(Education, Economy and Society) Jan Germen Janmaat, Marie Duru-Bellat, Philippe Méhaut, Andy Green - The Dynamics and Social Outcomes of Education Systems (2013, Palgrave Macmillan) PDF

The Dynamics and Social Outcomes of Education Systems
Education, Economy and Society
Series Editors: Andy Green, Institute of Education, University of London, UK;

Lorna Unwin, Institute of Education, University of London, UK; Karen Mundy,
Ontario Institute for Studies in Education, University of Toronto, Canada.
Titles include:
Andy Green
EDUCATION AND STATE FORMATION 2ND EDITION
Europe, East Asia and the USA
Andy Green and Jan Germen Janmaat (editors)
REGIMES OF SOCIAL COHESION
Societies and the Crisis of Globalization
Jan Germen Janmaat, Marie Duru-Bellat, Andy Green and Philippe Méhaut
THE DYNAMICS AND SOCIAL OUTCOMES OF EDUCATION SYSTEMS
Maha Shuayb (editor)
RETHINKING EDUCATION FOR SOCIAL COHESION
International Case Studies
Emer Smyth, Maureen Lyons and Merike Darmody (editors)
RELIGIOUS EDUCATION IN A MULTICULTURAL EUROPE
Children, Parents and Schools
Education, Economy and Society

Series Standing Order ISBN 978–0–230–29007–9 Hardback
ISBN 978–0–230–29008–2 Paperback
(outside North America only)
You can receive future titles in this series as they are published by placing a
standing order. Please contact your bookseller or, in case of difficulty, write to
us at the address below with your name and address, the title of the series and
the ISBN quoted above.
Customer Services Department, Macmillan Distribution Ltd, Houndmills,
Basingstoke, Hampshire RG21 6XS, England
The Dynamics and Social
Outcomes of Education
Systems
Edited by
Jan Germen Janmaat

Institute of Education, University of London, UK
Marie Duru-Bellat
Sciences Po Paris, France
Andy Green
Institute of Education, University of London, UK
and
Philippe Méhaut
LEST, France
Selection and editorial matter © Jan Germen Janmaat, Marie Duru-Bellat,
Andy Green and Philippe Méhaut 2013
Individual chapters © Respective authors 2013
All rights reserved. No reproduction, copy or transmission of this
publication may be made without written permission.
No portion of this publication may be reproduced, copied or transmitted
save with written permission or in accordance with the provisions of the
Copyright, Designs and Patents Act 1988, or under the terms of any licence
permitting limited copying issued by the Copyright Licensing Agency,
Saffron House, 6–10 Kirby Street, London EC1N 8TS.
Any person who does any unauthorized act in relation to this publication
may be liable to criminal prosecution and civil claims for damages.
The authors have asserted their rights to be identified as the authors of this
work in accordance with the Copyright, Designs and Patents Act 1988.
First published 2013 by
PALGRAVE MACMILLAN
Palgrave Macmillan in the UK is an imprint of Macmillan Publishers Limited,
registered in England, company number 785998, of Houndmills, Basingstoke,
Hampshire RG21 6XS.
Palgrave Macmillan in the US is a division of St Martin’s Press LLC,
175 Fifth Avenue, New York, NY 10010.
Palgrave Macmillan is the global academic imprint of the above companies
and has companies and representatives throughout the world.
Palgrave® and Macmillan® are registered trademarks in the United States,
the United Kingdom, Europe and other countries.
ISBN 978–1–137–02568–5
This book is printed on paper suitable for recycling and made from fully
managed and sustained forest sources. Logging, pulping and manufacturing
processes are expected to conform to the environmental regulations of the
country of origin.
A catalogue record for this book is available from the British Library.
A catalog record for this book is available from the Library of Congress.
Contents
List of Tables vii
List of Figures ix
Acknowledgements xi
Notes on Contributors xii
Introduction 1
Jan Germen Janmaat, Marie Duru-Bellat, Andy Green
and Philippe Méhaut
Part I The Variation and Dynamics

of Education Systems
1 The Dynamics of Education Systems: Convergent and

Divergent Trends, 1990–2010 15
Andy Green and Tarek Mostafa
2 From Upper Secondary to Further Education: European

Models of Post-Compulsory Learning 46
Audrey Dumas, Philippe Méhaut and Noémie Olympio
3 Lifelong Learning Regimes versus Vocational Education

and Training Systems in Europe: The Growing
Hybridisation of National Models 70
Eric Verdier
4 Formal or Actual Convergence? Three Cases of

Hybridisation 94
Philippe Méhaut
5 Curriculum Patterns in Citizenship Education:

A Cross-National Study 116
Christine Han, Jan Germen Janmaat, Tom May
and Paul Morris
v
vi Contents
Part II The Social Outcomes of Education Systems

6 Social Cohesion as Paradigm 141
Francois Dubet
7 Pre-School Education and Care: A ‘Win-Win’ Policy? 160

Tarek Mostafa and Andy Green
8 Educational Models and Their Impact on Student

Attitudes 181
Nathalie Mons, Marie Duru-Bellat and Yannick Savina
9 Educational Differentiation and Inequalities of Civic

Engagement 202
Jan Germen Janmaat
10 Lifelong Learning and Social Cohesion 223

Audrey Dumas, Philippe Méhaut, Noémie Olympio
and Isabelle Dimeglio
11 Education and Social Cohesion in a Comparative

Perspective 245
Marie Duru-Bellat, Antoine Vérétout and Francois Dubet
12 The Mismatch between Level of Education and Job

Qualifications: A Source of Mistrust and Intolerance
Depending on National Context? 267
Vanessa di Paola and Stéphanie Moullet
13 A Common Civic Culture in Europe: Has the Process of

European Integration Been Followed by the Convergence
of Citizenship Values? 281
Bryony Hoskins and Rebecca Ridley
Conclusion 299
Appendices 315
Author Index 324
Subject Index 328

Tables
1.1 Student–teacher ratios by country groups, 2000, 2009 22

2.1 Indicators 53
2.2 Example of membership 55
2.3 Level and mode of vocational provision in secondary
education 57
2.4 Access to and success in upper secondary education 58
2.5 Access to higher education 61
2.6 Access to continuing vocational training 64
3.1 ‘Decommodified’ LLL regimes 75
3.2 Market LLL regimes 77
5.1 Correlations of state control and collective values with
indicators of modernisation 131
7.1 The impact of PSEC participation rates on female and
total employment 166
7.2 Country-specific effects on the relation between
Logpresch and Logempf 167
7.3 PSEC participation (more than one year) before
universalisation 173
7.4 Cumulative total PSEC participation after progressive
universalisation for each group 173
7.5 Regression results 174
7.6 Average predicted performance scores for each ESCS
group before and after universalisation in the UK 175
8.1 Attitude variables 194
8.2 Explanatory variables by level 194
8.3 Variance component (in percentage) at each level for
attitudes and reading ability 195
8.4 Competitive Learning – nested multilevel models 196
8.5 Co-operative Learning – nested multilevel models 197
9.1 Between-classroom differences in six indicators of civic
engagement 210
9.2 Educational differentiation and inequalities of civic
engagement among grade eight students (correlations) 212
9.3 Between-classroom differences in civic engagement
(upper secondary students) 215
vii
viii List of Tables
9.4 Educational differentiation and inequalities of civic

engagement among upper secondary students
(correlations) 216
10.1 Typologies and scores of social cohesion according
to QCA 233
10.2 Access to upper secondary education and indicators of
social cohesion 235
10.3 Access to upper secondary and higher education and
indicators of cohesion 237
10.4 Continuing training and social cohesion 238
11.1 Inequalities, GPD, dynamics of the labour market and
social cohesion 250
11.2 Educational predictors of social cohesion 252
11.3 Educational variables, impact of educational
qualifications and social cohesion 259
11.4 Social and educational variables and social cohesion 260
11.5 Educational cohesion, inequalities and perception of
inequalities, and social cohesion 260
12.1 Mismatch and indicators of social and macroeconomic
cohesion by country (in %) 273
12.2 Correlation matrix for macro-level variables 275
12.3 The significance of the level 2 variables 276
13.1 Mean scores for the four scales in 1999 and 2009 by
country 291
13.2 Overall mean score and variation for the 16 European
countries for 1999 and 2009 with p-values given from
T-tests on mean differences 292
13.3 The 16 European countries’ contributions to the overall
variation for the four scales 292
Figures
1.1 Total public expenditure on education as a percentage

of GDP 22
1.2 Trends in private shares in total educational spending,
2000, 2008 24
1.3 Proportion of total tertiary education funding from
private sources, 2000 and 2008 24
1.4 Proportion of all schools which are public 25
1.5 Proportion of heads saying ‘always select by ability’ 25
1.6 Proportion of heads saying ‘never select by ability’ 26
1.7 Proportion of schools reporting ability grouping in all
subjects by country group, 2003 and 2009 26
1.8 Proportion of schools reporting ability grouping in ‘no
subjects’ by country, 2003, 2009 27
1.9 Proportion of decisions made at school level across
country groups in 2003 and 2011 29
1.10 Country average scores in literacy, 2000–2009 29
1.11 Trends in graduation rates at upper secondary level,
1995 to 2009 30
1.12 Graduation rates (in Type A Tertiary Education), 2002,
2006, 2009 31
1.13 Proportion of adults (25 to 64) with less than upper
secondary education, 1997–2009 31
1.14 Proportion of adults with tertiary education, 1998–2009 32
1.15 Changes in total variance by country group, 2000, 2009 33
1.16 Changes in within-school variance by country group,
2000, 2009 34
1.17 Changes in between school variance by country group,
2000, 2009 34
1.18 Trends in Between School Variance in ESCS 35
1.19 Clustergram 40
5.1 The composite state control indicator 126
5.2 The composite collective values indicator 127
5.3 The relation between state control and collective values 129
5.4 The relation between democratic consolidation and
state control 131
ix
x List of Figures
5.5 The relation between democratic consolidation and

collective values 132
7.1 The coefficient of variation 177
8.1 An original database about curricula in OECD countries 185
8.2 Graphic representation of Cramér’s V >0.40 matrix 187
8.3 Country clusters and distinctive categories 188
8.4 Correspondence analysis of curriculum variables 191
10.1 The link between education and social cohesion 226
11.1 Datasets and methods used 246
11.2 Composite index of social capital 248
11.3 Composite indexes for trust and tolerance 249
11.4 Impact of diploma and social cohesion 254
11.5 Perceptions of inequalities and social cohesion 257
11.6 A multilevel analysis of social cohesion 262
12.1 Indicators of social cohesion by country 274
Acknowledgements
This book marks the result of an inspiring cooperation between British

and French scholars interested in the links between education and social
cohesion. This issue is a core theme of both the ESRC-funded Centre
for Learning and Life Chances in Knowledge Economies and Societies
(LLAKES) (Grant number RES-594-28-0001), to which the British schol-
ars are affiliated, and the ANR-funded project ‘Education et Cohesion
Sociale’ (EDESCO), in which the French scholars have participated.
We would like to extend our gratitude to all the participants in the
joint meetings we have had over the past years for their valuable intel-
lectual input, as well as to those researchers who provided useful data
or comments on draft chapters. Special thanks go to the two funding
agencies, ESRC and ANR, whose financial support was crucial in making
this book possible.
xi
Contributors
Isabelle Dimeglio completed her doctoral research at the Institute of

Labour Economics and Industrial Sociology (LEST, Aix Marseille Uni-
versity, France). This research focused on the relationship between
trust, social cohesion and economic growth. Currently, she teaches
management and the economy of sports at Aix Marseille University,
France. Her research at the Institute of Sport Management (SMG, Aix-
Marseille University, France) examines the relationship between sports
and society.
Vanessa di Paola is Associate Professor at Aix-Marseille University,

France, and Labour Economist in LEST, France. She has published widely
on issues of over-education, unemployment duration among young
people and youth trajectories on the French labour market. She was
one of the French members of the ‘Atypical Employment and Welfare
Regime’ group of the European Economic change, Quality of Life and
Social Cohesion (Equalsoc) research programme.
François Dubet is Professor of Sociology at the University of Bordeaux,

France, and Director of Research at the École des hautes études en sci-
ences sociales (EHESS), France. He has published widely on social issues,
education and social justice. His most recent book is Pourquoi Moi?
L’Expérience des Discriminations.
Audrey Dumas is Assistant Professor (Maitre de Conference) in Eco-

nomics at the Centre for Analysis of Efficiency and Performance in
Economics and Management (CAEPEM) at the University of Perpignan
Via Domitia (FR), France. Her research area concerns the applied econo-
metrics of education and training. She has also published on measures
of inequality. Her research is comparative with a focus on European
countries and emerging countries such as Morocco.
Marie Duru-Bellat is Professor of Sociology at Sciences Po-Paris, France.

She has written numerous books and articles on gender and social
inequalities in education, on the ideological justification of these
xii
Notes on Contributors xiii
inequalities, and on the links between education and jobs; she has also
led many comparative studies on educational systems and reforms.
Andy Green is Professor of Comparative Social Science at the Institute of

Education, University of London, UK, and Director of the ERSC Research
Centre on Learning and Life Chances in Knowledge Economies and
Societies (LLAKES). His main field of research is the comparative (his-
torical and sociological) study of education and training systems, their
origins and social and economic consequences. His most recent book is
Education and State Formation in Europe, East Asia and the USA.
Christine Han is Lecturer in Education at the Department of Life-

long and Comparative Education, Institute of Education, University
of London, UK. She is the programme leader of the MA in Lifelong
Learning, Singapore, and also leads the Education and Development
in Asia module (MA in Comparative Education). Her areas of research
interests include values and citizenship education in East Asia (includ-
ing Singapore) and related issues such as social cohesion and civic
participation.
Bryony Hoskins is Senior Lecturer in Education at the University of

Southampton, UK. She is an internationally renowned expert on Active
Citizenship and has published widely in this field. Recent EU-funded
research projects include the Effects of Austerity on Active Citizenship
in Europe.
Jan Germen Janmaat is Reader of Comparative Social Science at the

Centre on Learning and Life Chances in Knowledge Economies and
Societies (LLAKES), Institute of Education, University of London (UK).
He is teaching on the MA in Comparative Education and has published
widely on the relations between education, civic values and social cohe-
sion. His most recent book is Regimes of Social Cohesion: Societies and the
Crisis of Globalization (co-author).
Tom May is a researcher at the UK Institute of Leadership and Man-

agement. He was previously a research manager at the former UK
Qualifications and Curriculum Authority where he specialised in inter-
national research into curricula and assessment. He is a research
associate at the centre for Learning and Life Chances in Knowledge
Economies and Societies (LLAKES), Institute of Education, University of
London, UK.
xiv Notes on Contributors
Philippe Méhaut is Senior Researcher at the Institute of Labour Eco-

nomics and Industrial Sociology (LEST), Aix Marseille University, France,
and CNRS, Aix en Provence, France. He has published widely on
comparative analyses of labour markets and educational policies.
Nathalie Mons is Professor of Sociology at the University of Cergy-

Pontoise, France. She has written many comparative studies of educa-
tional policies across OECD countries.
Paul Morris is Professor of Education at the Institute of Education,

University of London, UK.
Tarek Mostafa is a research officer at the centre for Longitudinal Studies

at the Institute of Education, University of London, UK. He is mainly
interested in economic and social policy with a particular emphasis on
the economics of education, educational inequalities, political economy
and quantitative methods. Tarek worked on various projects funded by
the European Commission and the Economic and Social Research Coun-
cil (ESRC). He is the principal investigator on the ESRC-funded project
on the challenges of consent in linking survey and administrative data.
Stéphanie Moullet is Labour Economist at the Labor Economics and

Sociology Laboratory (LEST), Aix en Provence and associate Professor
at Aix-Marseille University, France. She is teaching in a unit which is
mainly focusing on union’s representatives. Her main research themes
relate to the efficiency of education regarding labour market outcomes,
to the gender wage gap and to wage discrimination. She also published
several articles on the school-to-work transition, on the issue of over-
education and atypical employment at the beginning of working life.
Noémie Olympio is a PhD student in labour economics at the Institute

of Labour Economics and Industrial Sociology (LEST-CNRS), France. She
is working on a comparison between the Swiss and the French education
system in the light of the Amartya Sen’s Capability Approach. She has
recently published a comparative study on European education systems
in a Canadian book and an article on French educational trajectories in
the journal Economie et Sciences Sociales.
Rebecca Ridley is a PhD student at the University of Southampton,

School of Education, UK. Her research focuses on young people’s active
and political citizenship learning in both formal and informal settings
Notes on Contributors xv
and involves field work in secondary schools, the third sector and with
statutory youth service users. She has carried out field research in the
UK and in Hungary and worked on the Participatory Citizenship in the
European Union study coordinated by Bryony Hoskins.
Yannick Savina is Research Engineer at the Observatoire Sociologique

du Changement (OSC), Sciences Po & CNRS, Paris (FR). He specialises in
data analysis for the social sciences.
Eric Verdier is Research Professor at the National Center for Scientific

Research (CNRS), Institute of Labour Economics and Industrial Soci-
ology (LEST), Aix-Marseille University, France, and CNRS. His areas
of interest include vocational education and training policies and the
sociology of work and employment.
Antoine Vérétout is a sociologist in the Department of Sociology at the

University of Bordeaux, France. His most recent book is Les Sociétés et
Leur École : Emprise du Diplôme et Cohésion Sociale (co-author).
This page intentionally left blank
Introduction
Interest in education has continued to grow along with the development

of the modern global economy. Policymakers across Western states are
convinced that a country’s competitiveness in the global economy cru-
cially depends on its ability to develop a knowledge-based economy, and
they see education as a key strategic tool in this process. Education is
considered vital for the formation of a skilled and knowledgeable pool of
workers on whom the knowledge economy relies for innovation, growth
in labour productivity and the production of technologically advanced
and high value-added goods and services. Education is also interest-
ing for national policymakers in particular as it is one of the last fields
in which the national state, at least formally, has complete autonomy.
Thus, through educational policies, national politicians can potentially
still demonstrate their agency and power. A third reason why education
continues to attract attention is that it is believed to be an important
means to sustain social cohesion, as supported by research showing
that educated people are more tolerant, more civically and politically
active, more law-abiding, healthier and less inclined to develop addic-
tions. Referring to this research, politicians across the political spectrum
have been quick to point out that education cuts both ways – fostering
both economic competitiveness and social cohesion – and that rais-
ing the general level of education should consequently be a national
priority.
The first objective: Assessing the link between

education and social cohesion
However, the claim that education is the panacea for both economic
and social problems is not uncontested. Notwithstanding the numerous
1
2 Introduction
studies showing the economic and sociocultural benefits that education

brings for individuals,1 there are still controversies and omissions in
educational research, particularly regarding the link between education
and social cohesion. For instance, the debate on whether the benefits
of education for individuals can be generalised to whole societies is still
alive and kicking. In other words, if educated individuals tend to be
more tolerant, more trusting and more law-abiding, are more educated
societies also more open-minded, more cooperative and less criminal in
the aggregate? While, on the one hand, there are scholars who argue
that except for a few outcomes educational expansion is indeed advan-
tageous for society as a whole (e.g. Nie et al., 1996; Campbell, 2006;
McMahon, 2009), others are more sceptical of the ability of rising lev-
els of education to cure society’s social ills (Green et al., 2006; Stubager,
2008; Borgonovi, 2012).
Aside from this ongoing debate, research on the link between educa-
tion and social cohesion is anything but complete. Most of this research
has narrowly focused on the link between educational attainment, mea-
sured as years of education enjoyed or highest qualification attained,
and outcomes relevant for social cohesion such as trust, tolerance, par-
ticipation and crime, leaving a wide gap in our knowledge about the
impact of all the other aspects of education on these outcomes. This
is surprising as it is generally acknowledged that education shapes val-
ues and behaviour in many ways, for instance, through the curriculum,
peer socialisation, practices of selection and assessment and through the
labour market opportunities it provides. With regards to the latter, it has
been noted in French scholarly literature that societies with high rates
of return to educational qualifications are also societies with relatively
low degrees of intergenerational social mobility, suggesting that such
societies are not true meritocracies but stratified environments where
the middle and upper classes use the education system as a means
to pass privilege onto their offspring (Dubet et al., 2010). Yet, in the
Anglo-Saxon literature, the wider social consequences of the value of
qualifications in the labour market have hardly been explored. The first
objective of this book is thus to significantly advance our knowledge
of how all these other facets of education, including its link with the
labour market, contribute to or undermine social cohesion.
What makes our book stand out in pursuing this objective is its
cross-national comparative approach. There are two reasons why we
believe that only such an approach can increase our insight into the
link between education and social cohesion. Firstly, many aspects of
Jan Germen Janmaat, Marie Duru-Bellat, Andy Green and Philippe Méhaut 3
education other than attainment are properties of national systems (e.g.

curricula, age of first selection, age of enrolment, number and kind of
tracks in upper secondary, modes of assessment). In order to assess their
links with social cohesion outcomes, variation on these properties is
required, and this can only be achieved through cross-national anal-
ysis. Secondly, despite the varying understandings of social cohesion
in substantive terms by the contributors to this volume, we all share
the belief that social cohesion should be seen as a property of a soci-
ety, not of an individual or a small community. Only a conceptualising
of social cohesion as a societal phenomenon enables the identification
of a cohesion-threatening condition such as inter-community conflict
within a society. The conception of social cohesion as an individual or a
community trait risks overlooking inter-community strife and thus over-
estimating the degree of harmony in society at large. In other words,
an approach equating social cohesion with bonding social capital only
focuses on intra-community relations and is thus blind for bridging
social capital, that is, intergroup relations (cf. Chan et al., 2006; Green
et al., 2006; DiMeglio et al., 2013). Our understanding of social cohe-
sion as a societal phenomenon implies, as with system characteristics,
that cross-national analysis is needed to establish variation in social
cohesion and to link this variation with variation in system proper-
ties. Hence, all the contributions engage in cross-national comparison
with many of them using international datasets such as the Programme
of International Student Assessment (PISA), the OECD ‘Education at
a Glance’ series, the World Values Survey (WVS), the Civic Education
Study (Cived) and the International Civics and Citizenship Education
Study (ICCS). Many contributions will rely on aggregate attitudinal data
as measures of social cohesion and will relate these to particular system
characteristics.
The second objective: Exploring the variation

and dynamics of education systems
The cross-national comparative approach also necessitates a rigorous

examination of patterns and trends in education systems, which con-
stitutes the second main objective of this book. Such an examination
is required to determine whether countries show sufficient variation
in system properties to make research into the connections between
system properties and social cohesion worthwhile. This is an espe-
cially expedient task in view of the influential theoretical perspective
4 Introduction
claiming that the education systems of modern service-based economies

are becoming increasingly alike. Scholars sharing this view do not nec-
essarily agree on the reasons for this alleged system convergence. While
some have highlighted globalisation and the concomitant emergence
of transnational political actors influencing national policies as the
main driver of convergence (Held, 1989; Ohmae, 1990; Sklair, 1997),
others have pointed to the increasing dominance of an international
normative discourse emphasising human rights, democracy and liberal-
ism (Ramirez et al., 2006; Wiseman et al., 2011), and yet again others
have underlined the importance of the ever faster pace in which suc-
cessful models are emulated by other countries (Meyer, 1977; Baker
and LeTendre, 2005). However, whatever the reasons for the purported
convergence, if it is indeed the case that international variation in
education systems is diminishing then the legitimacy for undertak-
ing cross-national comparative research is compromised (Green, 2003).
After all, using cross-national variation to examine the social con-
sequences of system properties would then become increasingly less
effective. Indeed, some scholars have accepted convergence as a given
truth and have consequently argued that cross-national comparative
research is now obsolete and needs to be replaced by an analysis of
supranational and global processes (Beck, 2000; Jarvis, 2000). In this
regard, Beck (2000) has famously argued that comparativists should
abandon the ‘methodological nationalism’ of their research.
However, the convergence claim has also attracted considerable crit-
icism (e.g. Steiner-Khamsi, 2004; Carney et al., 2012). A well-known
critique has been that the advocates of convergence have tended either
to argue at a very high level of abstraction ignoring empirical evidence
altogether or to base their claim of convergence only on trends in formal
and structural properties (such as constitutions, curriculum objectives
and duration of primary and secondary education) while disregarding
developments in the content of the curriculum, the implementation of
policies and classroom practice.
Moreover, at least three theoretical perspectives on system change
can be identified predicting anything but convergence: (1) parallel
development, (2) regimes and (3) cultural traditions.
In the first perspective, the development of an education system is
seen as a linear process closely related to socio-economic development
more generally. Although national education systems are argued to be
changing in the same direction, they are not seen as necessarily drawing
closer to one another (e.g. Caillods,1994 on VET systems). In fact, if
the front runners on this trajectory develop faster than the laggards,
education systems will diverge. On the other hand, where there is a

ceiling effect and front runners reach a plateau, others may catch up.
The second perspective has been developed by scholars working in the
field of comparative political economy (CPE). They have argued that
notwithstanding the similarities between modern post-industrial soci-
eties in terms of the structure of their economies, levels of affluence
and demography, there are conspicuous differences among them in the
organisation of the worlds of work, education and welfare. These differ-
ences are seen as reflecting certain ‘regimes’, which can be understood as
unique constellations of social and institutional characteristics of mid-
range durability (roughly from several decades to more than a century).
A regime is seen as the product of a specific social formation charac-
terised by a distinctive class and political alliance. Regimes are not only
shaped by certain institutional path dependencies, but they are also sub-
ject to crises, external shocks and political transformations and are thus
not wholly immutable (Esping Andersen, 1990; Iversen and Stephens,
2008). Broadly, scholars working within this perspective have identified
four regimes characterising the labour market, social welfare and educa-
tion systems of Western countries: (1) a liberal regime observable in the
English-speaking countries, (2) a social-democratic group of countries
exemplified by the Scandinavian countries, (3) a conservative group of
countries represented by the German-speaking countries and the Low
Countries and (4) a southern European group comprising France, Spain,
Portugal, Italy and Greece. Unlike the parallel development perspective,
scholars associated within the regimes school of thought hold that edu-
cation systems develop in qualitatively different ways. In contrast to
convergence theory, they do not see these trajectories of development
as leading to wholesale convergence.
The third perspective differs from all the others in postulating pro-
found continuities in system characteristics. Differences among Western
states in these characteristics are lasting and are believed to be rooted
in distinct historical, political and cultural traditions. This perspec-
tive is exemplified by scholars arguing that conspicuous differences
between England, France and Germany in school socialisation and cit-
izenship policies reflect enduring liberal, republican-universalist and
ethnic-differentialist political legacies, respectively (e.g. Brubaker, 1992;
Baumann, 2004; Mannitz, 2004; Doyle, 2006). The contributions in
this book assessing patterns and trends in system properties will
explain the different perspectives on system change in greater detail
and will examine the empirical evidence for their claims (see the last
section).
6 Introduction
Other distinctive aspects
The book is distinctive in three other respects. Firstly, we proceed from a

lifelong learning understanding of education. It is now accepted knowl-
edge that learning is not confined to any particular life phase and it
therefore makes sense to examine all stages of education. The chapters
in our book will accordingly cover the whole spectrum of education,
from pre-primary through compulsory on to post-compulsory and adult
education, and will assess the contribution of each phase of learning to
a variety of social outcomes.
Secondly, we do not see education as an exogenous factor, exerting
its influence from some isolated external position. Rather we con-
sider education to be thoroughly embedded in the wider sociopolitical
environment. This interconnectedness of education with other societal
domains necessitates in our view a team of researchers from a variety
of backgrounds. Consequently, among the contributors to the current
volume we find sociologists, economists, historians, geographers and
political scientists, making the book a truly interdisciplinary enterprise.
The inclusion of scholars from many different disciplinary backgrounds
has generated a wealth of theoretical perspectives and methods, whose
application in the contributions to this book has greatly enhanced its
intellectual depth and width.
Thirdly, the book arouses curiosity by its adventurous use of quan-
titative and qualitative methods. Many chapters use state-of-the-art,
innovative and still unknown methods or are otherwise original by
applying methods that are rarely used in education. Thus, there are
contributions (1) using the coefficient of variance and an original appli-
cation of straightforward correlations to assess degrees of convergence
or divergence, (2) employing correspondence analysis to identify clus-
ters of countries sharing crucial similarities regarding the curriculum,
(3) applying Ragin’s (2000) ‘fuzzy-set’ qualitative comparative analy-
sis to develop a typology of post-compulsory education systems and
(4) using a simulation method to predict levels and inequalities of liter-
acy skills. In addition to these rarely used methods, many contributions
make use of multilevel analysis (MLA), a state-of-the-art technique to
assess the impact of variables pertaining to different levels of research
units on outcomes of interest at the individual level. It is worth not-
ing that a societal conception of social cohesion does not invalidate
using MLA with individual-level outcome measures of social cohesion,
as some contributions in this volume have done. To the contrary, MLA is
useful because it divides the variation in the outcome up into several
levels, usually those of the individual, the school and the country.
As long as the data on which MLA is performed includes a sufficient
number of countries and shows a sufficient amount of variance in the
outcome at the country level, the link between system characteristics
and social cohesion can be assessed. In fact, MLA has the additional
advantage of enabling the researcher to assess whether the links between
system properties and social cohesion represent real macro-level effects
or whether they are merely compositional in nature reflecting only the
sum of the effects of micro-level conditions. If they are compositional,
the effect of system properties is spurious. Thus MLA allows for a more
rigorous assessment of the effect of macro-level variables in relation to
that of other conditions.
The structure of the book
In view of the possibility of system convergence, an analysis of the

effects of education systems on social cohesion needs to be preceded
by an assessment of the variation in and dynamics of education systems
themselves. This logic has shaped the structure of the current volume.
Accordingly, the book is divided in two parts, with contributions in the
first part addressing variation in and dynamics of system properties and
chapters in the second part exploring the social consequences of a wide
range of educational conditions.
The contributions in the first part will broadly be informed by the
theoretical perspectives on system dynamics discussed above. They will
explain these perspectives in greater detail and will engage in a rigorous
scrutiny of a wide range of systemic properties to assess the explanatory
power of these perspectives empirically. This review of system proper-
ties will include levels of resources, mode of school organisation, school
regulation, take-up of apprenticeships and values in the curriculum, to
name but a few.
Chapter 1 of Part I is clearly the most comprehensive in terms of the
number of system properties and time period covered. Assessing trends
in no less than 26 indicators from the Education at a Glance and PISA
datasets over a period of two decades, it finds more evidence for conver-
gence than divergence. This convergence, however, is more pronounced
on output measures than on measures of resourcing and school sys-
tem characteristics. OECD states, for instance, appeared not to have
become more similar with respect to admissions, staffing, curriculum
and assessment decisions. They have become more alike on the issue
of decentralisation – not towards more school autonomy (as postulated
8 Introduction
by convergence theory) but in the opposite direction towards greater

centralisation of decision-making. In addition, the chapter identifies
five clusters of countries sharing crucial education system properties.
These clusters were found to correspond closely to the country groups
postulated by the CPE perspective.
Chapter 2 is highly original in that it represents a first attempt to clas-
sify European countries on the basis of their systems of post-compulsory
learning, including upper secondary, higher and adult education. It dis-
tinguishes countries with a weak from those with a strong system of
vocational education and training (VET) and concludes that the coun-
try groupings found are only partly in agreement with the shareholder
and stakeholder models predicted by the Varieties of Capitalism (VoC)
literature (which can be seen as a component of the CPE perspective).
In Chapter 3, Eric Verdier also observes that the regimes proposed by
the CPE perspective have only limited empirical value. Performing an
in-depth analysis of the trends in lifelong learning policies and struc-
tures in Denmark, Sweden, Germany, France and United Kingdom, it
concludes that all five countries appear to form hybrids of two or more
regimes. The chapter argues that the regimes of the CPE perspective
should therefore be seen as ideal-typical, whose value lies in illuminat-
ing the political choices that need to be made regarding a number of
sensitive questions. Chapter 4 takes this argument as a starting point to
examine the lifelong learning systems of three seldom explored coun-
tries – Denmark, Italy and the Netherlands. It concurs with Verdier’s
hybridity hypothesis as it shows that these countries are good examples
of compromises between different ideal-typical regimes as well.
Chapter 5 broadens the geographical scope by comparing a number of
prosperous East Asian states with a selection of Western states on their
citizenship education curricula. It finds pronounced differences between
these two groups of countries on two curriculum dimensions – state
control and political values. East Asian states appear to have a much
higher level of state control and to emphasise collectivism to a much
greater degree than any of the other states. These results are interpreted
as falling in line with the perspective postulating lasting cross-national
differences based on distinct political cultures and traditions.
In sum, the chapters outlined above show that there are still marked
differences between OECD states in terms of organisation, curricula and
inputs and outputs of their educations systems at each stage of learning,
which, moreover, correspond at least partly to the regimes of the CPE
perspective or to the cross-national differences proposed by the cultural
traditions perspective. Although they do find evidence for convergence
on some indicators, they also find examples of divergence on others.

What is more, the convergence found often appears not to go in the
direction postulated by convergence theory. Together these results point
to the limited empirical value of convergence theory, and in so doing
show that it still makes sense to carry out cross-national comparative
analysis to examine the links between education systems and social
cohesion.
Part II starts with an insightful chapter arguing that scholarly thinking
about how societies (should) cohere has changed and that the notion of
social cohesion has displaced terms such as ‘integration’ and ‘solidarity’
(cf. Chapter 6). It then proceeds with a string of chapters highlighting
the links between distinct system properties, relating to various stages
of lifelong learning, and various social cohesion outcomes. A num-
ber of marked links are found. Thus, with regard to the pre-primary
stage, Chapter 7 finds that universalising the provision of pre-school
education and care not only enhances overall levels of educational
achievement but also reduces inequality of performance and boosts
female employment rates. Regarding the compulsory stage, systems with
a less compartmentalised and hierarchical conception of knowledge
(i.e. those characterised by an integrated code of educational knowledge)
appear to foster more fluid relations among students and between stu-
dents and teachers (cf Chapter 8). Systems that minimise grouping by
ability in primary and lower secondary show the smallest disparities
across classrooms in civic engagement (cf. Chapter 9). Lastly, in rela-
tion to post-compulsory education, Audrey Dumas and co-authors et al.
find that high levels of access to adult learning are associated with
higher levels of trust, tolerance and participation among 15–30-year olds
(cf. Chapter 10).
Nevertheless, for many other system characteristics the link with
social cohesion outcomes appears more tenuous, leading a number of
contributors to this volume to conclude that either the impact of edu-
cation is overwhelmed by other factors or its impact is mediated by
other social domains. Regarding the latter, the interconnection with the
labour market appears crucial. In Chapter 11, Marie Duru-Bellat and co-
authors, for instance, find that countries with a high rate of return to
educational qualifications exhibit markedly lower social cohesion levels,
while Vanessa di Paola and Stéphanie Moullet show, in Chapter 12, that
the mismatch between an individual’s level of education and the skills
required for a certain job lowers interpersonal trust and ethnic tolerance
everywhere. This chapter further finds that the higher the aggregate
level of mismatch in a country, the lower the level of ethnic tolerance.
10 Introduction
The ambiguity of a direct link between education and social cohesion is

further illustrated by a contribution observing that despite an apparent
convergence across the European Union towards education policies fos-
tering cosmopolitan, tolerant and participatory notions of citizenship,
trends in public opinion among youngsters do not show any movement
in the direction of these civic-democratic ideals (cf. Chapter 13).
The concluding chapter tries to make sense of the findings of the
individual contributions by discussing them in greater detail and by
highlighting their implications for theory, research and policy.
Note
1. We are, of course, aware that education may also have negative effects for
individuals, such as frustrated expectations and alienation from friends and
family.
References
Baker, D.P. and LeTendre, G. (2005) National Differences, Global Similarities: World
Culture and the Future of Schooling. Stanford, CA: Stanford University Press.
Baumann, G. (2004) ‘Introduction: Nation-State, Schools and Civil
Enculturation’, in W. Schiffauer, G. Baumann, R. Kastoryano and S. Vertovec
(eds) Civil Enculturation: Nation-State, School and Ethnic Difference in The
Netherlands, Britain, Germany and France. Oxford: Berghahn Books, pp. 1–18.
Beck, U. (2000) What is Globalization. Cambridge: Polity Press.
Borgonovi, F. (2012) ‘The Relationship between Education and Levels of Trust and
Tolerance in Europe’, The British Journal of Sociology, 63, 1, pp. 146–167.
Brubaker, R. (1992) Citizenship and Nationhood in France and Germany. Boston:
Harvard University Press.
Caillods, F. (1994) ‘Converging Trends amidst Diversity in Vocational Training
Systems’, International Labour Review, 133, 2, pp. 241–257.
Campbell, D.E. (2006) ‘What Is Education’s Impact on Civic and Social Engage-
ment?’, in R. Desjardins and T. Schuller (eds) Measuring the Effects of Education
on Health and Civic/Social Engagement. Paris: OECD/CERI, pp. 25–126.
Carney, S., Rappleye, J., and Silova, I. (2012) ‘Between Faith and Science: World
Culture Theory and Comparative Education’, Comparative Education Review, 56,
3, pp. 366–393.
Chan, J., To, H., and Chan, E. (2006) ‘Reconsidering Social Cohesion: Developing
a Definition and Analytical Framework for Empirical Research’, Social Indicators
Research, 75, pp. 273–302.
Dimeglio, I., Janmaat, J.G., and Mehaut, P. (2013) ‘Social Cohesion and the
Labour Market: Societal Regimes of Civic Attitudes and Labour Market
Regimes’, Social Indicators Research, 111, 3, pp. 753–773.
Doyle, A. (2006) ‘Educational Equality, Religion and Social Integration: France
and England’, in J. Sprogøe and T. Winther-Jensen (eds) Identity, Education and
Citizenship – Multiple Interrelations. Frankfurt am Main: Peter Lang, pp. 55–72.
Dubet, F., Duru-Bellat, M., and Vérétout, A. (2010) Les sociétés et leur école. Emprise
du diplôme et cohésion sociale. Paris: Seuil.
Esping Andersen, G. (1990) The Three Worlds of Welfare Capitalism. Cambridge:
Polity Press.
Green, A. (2003) ‘Education, Globalisation and the Role of Comparative
Research’, London Review of Education, 1, 2, pp. 83–97.
Green, A., Preston, J., and Janmaat, J.G. (2006) Education, Equality and Social
Cohesion: A Comparative Analysis. Basingstoke: Palgrave.
Held, D. (1989) Political Theory and the Modern State: Essays on State, Power, and
Democracy. Stanford: Stanford University Press.
Iversen, T. and Stephens, J. (2008) ‘Partisan Politics, the Welfare State and
3 Worlds of Human Capital Formation’, Comparative Political Studies, 41,
pp. 600–637.
Jarvis, P. (2000) ‘Globalization, the Learning Society and Comparative Education’,
Comparative Education, 36, 3, pp. 343–356.
Mannitz, S. (2004) ‘The Place of Religion in Four Civil Cultures’, in W. Schiffauer,
G. Baumann, R. Kastoryano and S. Vertovec (eds) Civil Enculturation: Nation-
State, School and Ethnic Difference in The Netherlands, Britain, Germany and France.
Oxford: Berghahn Books, pp. 88–118.
McMahon, W.W. (2009) Higher Learning, Greater Good: The Private and Social
Benefits of Higher Education. Baltimore: John Hopkins University Press.
Meyer, J.W. (1977) ‘The Effects of Education as an Institution’, American Journal
of Sociology, 83, pp. 55–77.
Nie, N., Junn, J., and Stehlik-Barry, K. (1996) Education and Democratic Citizenship
in America. Chicago: Chicago University Press.
Ohmae, K. (1990) The Borderless World: Power and Strategy in the Interlinked
Economy. London: Collins.
Ragin, C. (2000) Fuzzy-Set Social Sciences. Chicago UP: Chicago.
Ramirez, F.O., Suárez, D., and Meyer, J.W. (2006) ‘The Worldwide Rise of Human
Rights Education’, in A. Benavot and C. Braslavsky (eds) School Knowledge
in Comparative and Historical Perspective: Changing Curricula in Primary and
Secondary Education. Hong Kong: Comparative Education Research Centre,
University of Hong Kong, pp. 35–52.
Sklair, L. (1997) ‘Social Movements for Global Capitalism; the Transnational
Capitalist Class in Action’, Review of International Political Economy, 4, 3,
pp. 514–538.
Steiner-Khamsi, G. (2004) The Global Politics of Educational Borrowing and Lending.
New York: Teachers College Press.
Stubager, R. (2008) ‘Education Effects on Authoritarian–Libertarian Values:
A Question of Socialization’, British Journal of Sociology, 59, 2, pp. 327–350.
Wiseman, A., Astiz, M.F., Fabrega, R., and Baker, D. (2011) ‘Making Citizens of the
World: The Political Socialization of Youth in Formal Mass Education Systems’,
Compare, 41, 5, pp. 561–577.
Part I
The Variation and Dynamics
of Education Systems
1
The Dynamics of Education
Systems: Convergent and Divergent
Trends, 1990–2010
Andy Green and Tarek Mostafa
Debates about societal convergence and divergence have persisted

since Enlightenment thinkers first developed theories about univer-
sal patterns of development (‘Progress’) which ‘counter-Enlightenment’
thinkers contested (Kumar, 1991). As sociology developed in the nine-
teenth century, universalistic conceptions, stressing convergence, were
typically associated with the ‘functionalist’ theories and positivist
methods of rationalist social science (e.g. Emile Durkheim), while
particularistic conceptions, stressing cultural and national differences,
were associated with anti-rationalist or heuristic traditions. The latter
derived from the eighteenth-century Sturm und Drang romantics, like
Johann Gottfried Herder (Greenfeld, 2003), and were also exemplified
in Germany by Friedrich List’s school of National Political Economy
(List, 1885) and later partially reflected in the works of comparative
sociologists like Max Weber and Ferdinand Tönnies. The debates have
continued to be a staple of modern social science, which has reprised
many of the nineteenth-century arguments between universalists and
particularists in new forms (Gray, 2007).
Modernisation theory argues that countries generally follow simi-
lar paths of socio-economic development, albeit at different speeds,
towards a Western model of urbanised, industrial societies organised
broadly along market lines and tending towards ‘democratic’ forms of
government. The latter is often seen as an ‘end state’ and therefore
implies that there is a long-term process of convergence. A modern
version of nineteenth-century notions of ‘universal progress’, this line
of thought received a major boost in the 1960s, particularly in Anglo-
American thought, from the near-simultaneous publication of seminal
15
16 The Variation and Dynamics of Education Systems
texts such as Walt Rostow’s Stages of Economic Growth (Rostow, 1960),

Lipset’s early work on capitalism and democracy (1959) and Clerk
Kerr’s work on the ‘logic of industrialisation’ (Kerr et al., 1960). Francis
Fukuyama’s notorious later claims (1992) about the global conquest
of liberal capitalism and ‘the end of history’ took the argument to
its hubristic apogee. However, since the late 1980s the mainstream
locus for the debate has been globalisation theory, with so-called
‘hyper-globalists’ (Ohmae, 1990, 1996) predicting wholesale global con-
vergence and ‘glocalists’ (Robertson, 1995) and ‘transformationlists’
(Held et al., 1999) providing more nuanced accounts of a simultaneous
dialectic of convergence and divergence.
While quantitative economists and sociologists tend to stress con-
vergence, many cultural and political theorists, both from the Right
and the Left, have continued to emphasise difference, most notably,
recently, in Samuel Huntington’s thesis about the Clash of Civilisations
(Huntington, 1997). World Systems theorists (Wallerstein, 2001) and
post-colonial theorists (Crossley and Tickly, 2004) stress the structural
asymmetry between core and peripheral economies. Comparative polit-
ical economists (e.g. Streeck, 1989, 2009; Hutton, 1995, 2002, 2007,
2011; Dore, 2000; Hall and Soskice, 2001; Thelen, 2004) and welfare
state theorists (Esping-Andersen, 1990, 1999), while acknowledging the
common forces of globalisation, still insist on the significance of histor-
ically evolved national institutional and cultural peculiarities and the
path-dependent processes which help to reproduce national differences.
These debates on convergence and divergence have also been played
out in educational theory and particularly in comparative education
where the distinction between ‘nomothetic’ (universalising) and ‘ideo-
graphic’ (particularising) modes of thought is constitutive of the very
field of study (Epstein and Caroll, 2005; Griffiths and Knezevic, 2012).
The most influential advocates of convergence have been the group
of sociologists around John Meyer and Francisco Ramirez at Stanford
University, whose research over decades consistently argues for a ‘World
Culture’ or ‘World Polity’ which they say is steadily eroding national
differences in education and causing education systems gradually to
converge around the world. The process, which they describe as largely
‘stateless’, is essentially one of cultural diffusion, not only between
the global policy elites and national policy actors, but also ‘on the
ground’, as the recognised global ‘standard’ affects the behaviour of
educational practitioners and consumers. ‘Many features of the con-
temporary nation state,’ they write, ‘derive from the worldwide models
constructed and propagated through cultural and associational process’
Andy Green and Tarek Mostafa 17
(Meyer et al., 1997, p. 145). This ‘institutionalisation of world mod-

els’, they claim, has led to a degree of ‘structural isomorphism’ in
national societies despite the latter’s ‘enormous differences in resources
and traditions’ (ibid, p. 145).
Meyer and his colleagues recognise that there is often a substan-
tial ‘de-coupling’ or ‘disconnect’ between the policy rhetoric and the
actual practises in many countries. But they argue, nonetheless, that
world models do impact over time on practice and cause convergence
in school structures and processes. Post-independence states, they say,
invariably create national education systems, replete with educational
bureaucracies, age-graded structures of schooling, national curricula and
trained teachers (Boli et al., 1985). Increasingly, at the national policy
level, ‘all national states define their fundamental purposes as having to
do with socio-economic development or welfare and individual justice,
rights and equality’ (Meyer, 2000, p. 237). National curricula become
increasingly similar, in both form and content, with an emphasis on
scientific and ecological principals and human rights (Meyer et al., 1997
Meyer, 2004), the latter exemplified, they say, by the increasingly nor-
mative reference to human rights in social studies textbooks (Meyer
et al., 2010). Convergence is also found in assessment methodologies
and enrolment rates in higher education.
Various criticisms have been made of World Culture theory, not least
in terms of its models of the processes underlying convergent global
trends and its claims about the degree of convergence actually mani-
fested. The most common criticism of the model of cultural diffusion
is that it ignores geo-political power relations, assuming – without
consideration of the power exercised by global institutions and aid
organisations – that policies and practices spread through the ‘volun-
tary’ or ‘spontaneous’ adoption of global ‘best practises’ by national
civil societies. Griffiths and Knezevic (2012), for instance, reject Meyer’s
‘liberal idealist’ argument in favour of a Wallersteinian Word Systems
approach which takes more account of the systematic material and
epistemological inequalities across countries which are built into the
core-periphery dualism of the modern world economy.
Such arguments do not necessarily demur on the degrees of conver-
gence to be found in the contemporary world. However, other critiques
suggest that the World Culture theorists exaggerate global convergence.
Steiner-Khamsi (2004), for instance, argues that their analysis ignores
the degree to which the global policies ‘transferred’ or ‘borrowed’ by
particular states and their actors are adapted to local contexts. This
mutation limits the degree of actual convergence that results. Thus
‘despite all the political and economic pressure on low income coun-
tries to comply with “international standards” in education, imported
policies do not have homogenising effects, that is, do not lead to conver-
gence of educational systems’ (Steiner-Khamsi, 2004, p. 203). A similar
argument was made by Green et al. (1999) in their survey of trends in
education and training systems the European Union during the 1980s
and 1990s. They found that while ‘policy rhetorics’ did tend to con-
verge, actual convergence in structures and processes on the ground was
much more limited.
Arguably, these and other similar critiques arise from a basic difference
in perspective. Meyer and colleagues find a great deal of convergence in
very general properties of systems. For instance, countries are said to be
converging when they mostly adopt written constitutions or when they
enact national curricula or compulsory schooling measures. This can, of
course, be said to represent convergence, but it is convergence at a very
high level of generality. Critics, on the other hand, will tend to focus
more on the details of what is in the constitutions or curricula, or how
they are put into practice, and at this level may find much more diver-
gence. Meyer and his colleagues often see things from a quite Olympian
height, as exemplified in Meyer’s comment that the cold war ‘can be
seen to have been about modest variations in models of social control in
the economy’ (Meyer, 2000). Others, taking a less elevated point of view,
would argue that despite common trends towards industrialisation and
urbanisation, modalities of economic organisation and political control
were significantly different in the opposing blocs during the cold war
era. Likewise, in the most recent era of globalisation, you can argue with
equal justification that the world is increasingly converging on capital-
ist modes of production and exchange or that the forms of capitalism
across the world are becoming increasingly diverse. Both are true. In a
sense, it all depends on your vantage point.
Universalists (including World Culture theorists) tend to measure con-
vergence by observing trends across a wide range of countries. If the
spread of values on any given measure becomes narrower over a period
of time you have convergence. An alternative approach has been to
identify the main distinguishing models of societal organisation, rep-
resenting clusters of countries, and to trace their trajectories over time
to see whether or not they remain distinctive. The models are essen-
tially Weberian ‘ideal types’ which are used to highlight the common
and distinctive characteristics of groups of countries and the functional
(or dysfunctional) relationships which pertain between the character-
istics in each model. Comparative methods (both quantitative and
qualitative) are used to show how these particular sets of (institutional

and cultural) relationships arise historically and how they may explain
the distinctive socio-economic outcomes typical of each model. Com-
parative political economists, for instance, identify different ‘varieties
of capitalism’ (Hall and Soskice, 2001) and different welfare regimes
(Esping-Andersen, 1990) in the developed world. The terminologies of
classification vary but a distinction is commonly made between a Social
Democratic model (of capitalism and welfare) (represented by Nordic
countries), a Social Market model (north-western continental European
countries) and a Liberal model (Anglo-American or English-speaking
countries), with southern European and East Asian states forming less
distinctive clusters and models (for a summary, see Green et al., 2010;
Green and Janmaat, 2011). Opinions vary, but a common view now (e.g.
Dore, 2000; Streeck and Thelen, 2005; Streeck, 2009) is that the differ-
ent models of Western capitalism are converging in a number of key
respects, under the pressure of globalisation, but still remain distinctive.
At the same time the rise of East Asian and other capitalisms provides
new sources of variation in the range models of worldwide (Hutton,
2007, 2011; Jacques, 2012).
Research on education and training has also identified distinctive
models in different groups of countries. Comparative educationalists,
going back to ‘founding fathers’ like Nicholas Hans, Isaac Kandel and
Michael Sadler, have often identified distinctive traditions of curric-
ula and school organisation in, inter alia, English-speaking, German-
speaking and Mediterranean countries, for instance (McLean, 1990).
More recently there has been a proliferation of research identifying dif-
ferent models of skills formation, which overlaps with, and draws on,
comparative political economy research on labour market institutions
and skills (e.g. Jallarde, 1989; Ashton and Green, 1996; Crouch et al.,
1999; Brown et al., 2001; Greinert, 2004; Busemeyer and Trampusch,
2011). Verdier’s chapter in this volume provides a new assessment of
these traditions.
Recent analyses of models of school systems (Mons, 2007) and life-
long learning systems (Green et al., 2006) focus on system characteristics
relating to institutional organisation, finance and regulation, and educa-
tional outcomes (levels and distributions of skills). Like previous studies,
they tend to find distinctive models associated with Nordic countries,
English-speaking countries, Mediterranean countries and Germany and
countries geographically proximate to Germany. Mons’ (2007) typol-
ogy of organisational models for the management of diversity in public
education systems includes a ‘model of separation’ (Austria, Germany,
Switzerland, Belgium); a ‘model of individualised integration’ (Finland,

Norway, Sweden plus Japan and Korea); a ‘model of a la carte inte-
gration’ (the UK, USA, Canada) and a ‘model of uniform integration’
(France, Italy, Spain and Greece). Green et al. (2006) relate models of
lifelong learning to varieties of capitalism and welfare regimes and iden-
tify lifelong learning models which cluster countries in broadly the
same groups as in the political economy literature. So English-speaking
countries form a Liberal Model, Nordic countries a Social Democratic
model and north-western continental European countries (including
the German-speaking countries) a Social Market model. Mediterranean
countries, which in the educational typology include France, form a
rather loose and diverse grouping, as do East Asian countries. The
models are differentiated primarily by different forms of institutional
organisation and regulation which correspond to different outcomes,
particularly in terms of skills distribution. For instance, the compre-
hensive systems of the Nordic countries, with lows levels of selection
and ability grouping, produce relatively egalitarian outcomes compared
with the countries in Liberal and Social Market models which have more
school diversity and ability-grouping (e.g. the UK) or selection by abil-
ity to secondary schools (particularly in German-speaking countries)
(Mostafa, 2011).
The methodology
This chapter seeks to provide some empirical tests of convergence and

divergence in the trends in OECD education systems over the past
two decades. We use 24 indicators relating to school organisation,
finance and regulation and skills outcomes from the OECD Education
at a Glance and Programme of International Student Assessment (PISA)
datasets. The data on school regulation and outcomes mostly come from
PISA and therefore cover the relatively short period from 2001 to 2009
which the survey covers. Other data from Education at a Glance tend to
cover a longer period, in most cases from 1990 to 2009. The objective
was to test the arguments arising from research in both the ‘universal-
istic’ and more ‘particularistic’ traditions, thus including the claims for
convergence from the World Culture theorists and the claims for contin-
uing diversity of models in the comparative political economy tradition.
In order to do this we tested for convergence and divergence over time
both across countries and across country groups. In the latter, the groups
are defined as ‘Liberal’, ‘Social Market’, Social Democratic, with coun-
tries classified to groups as above (Green et al., 2006). For the analysis of
trends across countries and across groups, we selected for each indica-
tor a sub-set of OECD countries for which there were values both at the
beginning and end of the reference period. While the classification of
countries to country groups remains constant for all indicators, missing
values may therefore mean that the number of countries in each group
may be different for different indicators, as in the cross-country analysis.
One should note that all our data are country-level macro-data.
The tests conducted are necessarily different in the analysis by coun-
try and the analysis by country groups. For the analysis across countries
we computed coefficients of variation (CV) at the beginning and end of
the reference period and used regression analysis to determine whether
any changes in the coefficients were significant.1 Convergence is there-
fore identified only where a decrease in the CV over time is confirmed
as significant by the regression analysis. For the analysis across country
groups, the number of cases was too small to conduct significance tests,
so we merely report on changes in the CVs. We report on both types
of evidence in the text below, but provide graphics only to illustrate
changes in the CV for the country groups, since it is not possible to pro-
vide visual representations of a large number of countries. For reasons
of space, we omit the data tables, which can be found on the website
(llakes.org).
The text below reports on the findings on convergence and divergence
across countries and country groups under the following headings:
1. Resources;
2. School organisation;
3. School regulation;
4. Skills outcomes.
Findings
Resources
Governments have generally given greater priority to educational
spending since 1990 (see the trends in Figure 1.1). Public spending on
education as a proportion of GDP has risen on average across countries
since 1990 (from 4.76 per cent in 1990 to 5.54 per cent in 2008 for the
country sub-set), with a non-significant convergence between countries
since 1995 (CV from 0.192 in 1995 to 0.149 in 2008; p = 0. 156). The
averages for country groups have diverged somewhat during the period
(CV from 0.179 to 0.226) due to the Scandinavian group pulling away
from the rest. Public spending per student as proportion of GDP per
8
7
6
5
4
3
90
92
94
96
98
00
02
04
06
08
19
19
19
19
19
20
20
20
20
20
Social democrats Liberals Japan
Social market Mediterranean
Figure 1.1 Total public expenditure on education as a percentage of GDP
capita has also risen on average across countries (from 21.75 per cent
in 1990 to 25.24 per cent in 2008 for the country sub-set) and shows a
substantial and significant convergence across countries (CV from 0.251
in 1990 to 0.075 in 2008; p = 0) and apparent convergence across coun-
try groups (CV from 0.213 to 0.136). However, student–teacher ratios
did not converge during the most recent period between 2000 and
2009, either within the country sub-set (CV remained unchanged at
0.1903; p = 0. 086) or across country groups (CV from 0.133 to 0.137)
(see Table 1.1).
Public educational spending as a proportion of GDP and education
spending per student as a proportion of GDP per capita only tell you
the policy priority a country’s government attributes to education, not
what it is actually spent, since GDP and GDP per capita vary substan-
tially across the countries. Only the student–teacher ratio measure gives
Table 1.1 Student–teacher ratios by country groups, 2000, 2009.
Regimes Student–teacher Student–teacher

ratio 2000 ratio 2009
Social democrats 11.52 11.28

Liberals 14.14 14.53
Social market 12.64 12.38
Mediterranean 11.18 9.96
Japan 15.10 12.23
Std 1.68 1.68
CV 0.133 0.137
a proxy for absolute levels of resources. So the general picture would

seem to be that countries have converged over 20-year period from 1990
on the policy priority given to education as a whole. But countries were
not converging in terms of the actual level of resources committed to
education during the 2000s.
There has been a small increase in the proportion of total educa-
tion spending deriving from private sources (sub-set country average
increased from 13.7 per cent in 2000 to 16.5 per cent in 2008), with
a non-significant convergence across countries during the period (CV
from 0.83 to 0.66; p = 0. 801). Liberal countries tend to have the high-
est proportion of spending coming from private sources (23.8 per cent
in 2008). The proportion was lower in the Social Market (11.2 per cent)
and Southern European countries (10.3 per cent) and lowest in the Social
Democratic countries (5.5 per cent in 2008). The country group averages
have not converged much during the period (CV from 0.7 to 0.6) and
there is a continuing polarisation of the Liberal and Social Democratic
states as, respectively, high and low private spenders.
The share of private funding in spending on tertiary education
is higher and has also risen during the period (from an average of
26.7 per cent in the OECD country sub-set in 2000 to 32.1 per cent in
2008) with no significant convergence across countries (CV from 0.883
to 0.774; p = 0. 67). Again the Liberal countries tend to have the high-
est share, followed by the Social Market countries, and with the Social
Democratic countries having the lowest, by a long way. There was no
significant convergence in the country groups during this period (CV
from 0.974 to 0.902).
School organisation
The data from PISA show little evidence of widespread privatisation of

schools between 2000 and 2009. The proportion of schools classified as
‘Private Independent’ went up only slightly across countries, from 4.25
per cent to 4.42 per cent, and there was a significant convergent trend
(p = 0. 026). The proportion classified as ‘Private Dependent’, conversely,
went down from 14.92 per cent to 14.01 per cent. As a consequence,
the proportion of all schools which are defined as ‘Public’ (i.e. neither
‘Private Independent’ nor ‘Private Dependent’) has actually increased
slightly between 2000 and 2009 (from 80.81 per cent to 81.57 per cent)
in a slightly convergent trend (CV from 0.29 in 2000 to 0.25 in 2009;
p = 0. 033). The slight increase we saw above in the relative shares of
private spending on education (Figures 1.2 and 1.3) may therefore be
25.0
20.0
15.0
10.0
5.0
0.0
Liberal Social market Social S. Europe
democratic
2000 2008
Figure 1.2 Trends in private shares in total educational spending, 2000, 2008
50.0
40.0
30.0
20.0
10.0
0.0
Liberal Social Social
democratic market
2000 2008
Figure 1.3 Proportion of total tertiary education funding from private sources,
2000 and 2008
due to increases in levels of fees or other educational costs to parents,

illustrating a trend towards ‘marketisation’ of public organisations, but
do not signal widespread privatisation of schools per se.
The differences between country groups in the proportion of dif-
ferent types of school remain quite large (Figure 1.4). As one might
expect, the countries in the Social Democratic group have the lowest
proportions of Private Independent schools and the highest proportion
of Public schools. The Liberal countries (and Japan) have the highest
proportion of Private Independent schools and the Social Market coun-
tries have the highest proportions of Private Dependent schools. The
proportion of schools which are public is highest in Social Democratic
and Mediterranean states and lowest in the Social Market States and
Japan.
100.00
80.00
60.00
40.00
20.00
0.00
ls
n
s
et
an
at
pa
ra
ne
r
ar
be
Ja
oc
rra
Li
m
al
ite
de
ci
ed
So
al
M
ci
So
2000 2009
Figure 1.4 Proportion of all schools which are public
The reports by head teachers on the practices relating to their own

schools suggest that selection to schools by ability (i.e. academic records)
has been decreasing. On average across OECD countries the proportion
of heads reporting academic records ‘always’ being used in selection
went down from 26.24 per cent in 2000 to 24.26 per cent in 2009.
Conversely, the proportion of heads reporting their schools ‘never’ used
academic records to select went up from 52.67 per cent in 2007 to 55.44
per cent in 2009. The coefficient of variation for the ‘sometimes’ ques-
tion remained the same at the end of the period as at the beginning
(CV: 0.65); it went up marginally but non-significantly on the ‘always’
question (CV: 1.09–1.10; p = 0. 205) and down on the ‘never’ question
(CV from 0.6 to 0.56).
The country groups are sharply distinguished on questions of selec-
tion, as Figures 1.5 to 1.8 show. The Social Democratic countries had
the lowest proportion of heads saying that their schools always selected
60.00
40.00
20.00
0.00
Year 2000 Year 2009
Liberal Social democratic
Social market Southern Europe
Figure 1.5 Proportion of heads saying ‘always select by ability’

100.00
80.00
60.00
40.00
20.00
0.00
Year 2000 Year 2009
Liberal Social democratic

Social market Southern Europe
Figure 1.6 Proportion of heads saying ‘never select by ability’
60.00
50.00
40.00
30.00
20.00
10.00
0.00
ts
n
an
ls
t
ra
ke
pa
ra
ne
oc
ar
be
Ja
rra
m
m
Li
de
ite
al
ci
ed
al
So
ci
M
So
Ability grouping 2003 Ability grouping 2009
Figure 1.7 Proportion of schools reporting ability grouping in all subjects by

country group, 2003 and 2009
by ability (1.3 per cent in 2000 rising to 1.77 per cent in 2009).
Next came the Mediterranean countries (1.63 per cent rising sharply to
16.46 per cent) and then the Liberal countries (17.11 per cent dropping
to 16.38 per cent). By far the most likely to select were the Social Mar-
ket countries where 50.10 per cent of head teachers reported ‘always’
selecting by ability in 2000 (dropping to 47.10 per cent in 2009). There
has been a slight convergence between the country groups during the
period (CV for ‘Always’ drops from 1.31 to 0.93; CV for ‘sometimes’
drops from 0.59 to 0.38 and CV or ‘never’ drops from 0.53 to 0.46).
70.00
60.00
50.00
40.00
30.00
20.00
10.00
0.00
ls
n
an
t
s
ke
pa
at
ra
ne
ar
r
be
Ja
oc
rra
m
Li
m
ite
al
de
ci
ed
So
al
M
ci
So
Ability grouping 2003 Ability grouping 2009
Figure 1.8 Proportion of schools reporting ability grouping in ‘no subjects’ by

country, 2003, 2009
However, the country groups still remain very distinctive at the end of
the period.
According to the reports by head teachers, ability grouping has
become less common across the OECD. The proportion of schools in
each country reporting ability grouping in all subjects has gone up in
two countries and down in 20 countries. The OECD country sub-set
average went down from 39.01 per cent to 12.7 per cent in a con-
vergent trend (CV from 1.82 to 0.97; P = 0). Conversely, the average
for the proportion of schools in each country not practising ability
grouping in any subject has gone up from 25.24 percent in 2003
to 32.04 percent in 2009 (with a slight apparent convergence as the
CV declines from 0.96 to 0.71). However, this is counteracted by the
higher proportion of schools adopting ability grouping ‘in some sub-
jects’ which has gone up in 15 countries and down in seven countries.
There appears to have been a slight divergence across countries on this
measure.
Country groups have remained quite differentiated in terms of the
use of ability grouping. Coefficients of Variation for the country groups
increased substantially over time on the measures for ‘all subjects’
(0.43–0.70) and on the measures for ‘some subjects’ (0.17–0.41) and
decreased only slightly on the measure for ‘no subjects’ (0.77–0.61). This
suggests an overall divergence across country groups during this period.
By 2009 the Social Democratic countries clearly had less ability grouping
(in all subjects and some) than the Liberal and Social Market groups and
are only surpassed by the Mediterranean countries in the proportion of
schools having ability grouping in no subjects.
Regulation and level of decision-making
Since the early 1980s there have been widespread debates about the
most appropriate levels for decision-making in education systems.
Global and regional policy bodies have tended to be strong advocates
of decentralisation and many governments have sought to introduce
measures which would reduce the number of decisions taken at the
central level and increase those taken at the level of the school.
The main argument for this has been to increase efficiency through
reducing bureaucracy, increasing responsiveness to local communities,
allowing more creative management of local resources and providing
more incentives for improving the quality of teaching (OECD, 2012,
p. 500).
During the past decade the OECD has been building up a dataset on
the level at which different types of decisions in lower secondary edu-
cation are made in different education systems, distinguishing between
‘central’, ‘regional, ‘local’ and ‘school-level’ decisions. The data are based
on surveys administered to panels of experts drawn from different
levels of decision-making in each country. The survey includes ques-
tions on 46 types of decision ranged across four domains relating to:
‘organisation of instruction’, ‘personnel management’, ‘planning and
structures’ and ‘resource management’. These have been developed into
a composite indicator (OECD, 2012, pp. 500–510).
Data on these indicators for the years 2003–2011 suggest that the
number of decisions made at the central level has risen on average
across countries whereas the number of decisions made at the school
level has decreased. During this period the number of decisions taken
at the school level, for instance, decreased in 10 out of 21 countries
and increased in only 4 (p. 500). There was no significant convergence
across countries on this measure (CV 0.477 in 2003 and 0.499 in 2012;
p = 0. 285) and across country groups there were signs of divergence
(CV from 0.195 to 0.273). This flies in the face claims of a common
and convergent trend towards decentralisation (Astiz et al., 2002). Most
of the country groups show similar levels of decentralisation on this
measure by 2011. However, the liberal group of countries remain con-
siderably more decentralised than the others and became more so (see
Figure 1.9).
70
60
50
40
30
20
10
0
Liberals SM SD Asian S. Eur
2003 2011
Figure 1.9 Proportion of decisions made at school level across country groups in
2003 and 2011
Source: Education at a Glance, Statistical Report, 2012, p. 509, Table D.6.6.
Skills outcomes
Country mean scores in PISA on literacy tests declined slightly on aver-

age across countries between 2000 and 2009, dropping in 18 countries
and rising in only 5 (CHE, DEU, GRC, LUX, PRT) and there has been
a slight but significant convergence on this measure (CV down from
0.048 to 0.032; p =0). The country group averages appear to have con-
verged slightly (CV from 0.034 to 0.026), with the scores for the formerly
leading Liberal states declining while those for the Social Market and
Mediterranean states remained fairly stable (Figure 1.10).
530
520
510
500
490
480
470
460
n
an
ts
t
ls
ke
pa
ra
ra
ne
ar
Ja
be
oc
rra
m
m
Li
al
ite
de
ci
ed
So
al
M
ci
So
Average scores 2000 Average scores 2009
Figure 1.10 Country average scores in literacy, 2000–2009

120
100
80
60
40
20
0
95
00
01
02
04
05
06
07
08
09
03
19
20
20
20
20
20
20
20
20
20
20
Liberal Social market Social democratic

S. European
Figure 1.11 Trends in graduation rates at upper secondary level, 1995 to 2009
While measured skills at 15 appear to have declined slightly on

average across countries, rates of graduation from upper secondary edu-
cation have increased (Figure 1.11). In 1995, 74 per cent of the age group
in the sub-set of OECD countries graduated from upper secondary edu-
cation. By 2009 this proportion had risen to 82 per cent. There was a
convergent trend towards higher rates of graduation within the sub-set
of OECD countries (CV down from 0.229 in 1995 to 0.17 in 2009; p =0).
The country groups appear to have converged strongly (CV from 0.205
to 0.061), but the number of cases in each group is rather small due to
missing data.
Graduation rates from long-cycle tertiary education converged across
countries in an upwards direction (CV from 0.322 to 0.263; p = 0. 021)
(Figure 1.12). In 2002 the proportion of an age group graduating
from this type of higher education was 31 per cent in a sub-set of
OECD countries. By 2009 it was 38 per cent. The country groups also
appeared to converge somewhat (CV from 0.272 to 0.179) although
the groups remained quite differentiated with the Social Democrat and
Liberal groups having much higher rates than the Social Market and
Mediterranean groups.
Changes in the flows of skills onto the labour market only alter the
stocks of adult skills very gradually, since the latter are the result of the
accumulation of skills over a 65-year period. The OECD cross-country
data drawn from the national Labour Force Surveys show a general
60
40
20
0
Liberal Social Social S.
market democratic European
2002 2006 2009
Figure 1.12 Graduation rates (in Type A Tertiary Education), 2002, 2006, 2009
80
60 62
56
47
43
40
36 34
29 27 29
25 25
20 29 30 21
24 21
0
1997 2000 2005 2009
Liberal Social market

Social democratic S. Europe
Figure 1.13 Proportion of adults (25 to 64) with less than upper secondary
education, 1997–2009
See Education at a Glance, 2011, Indicator A1.
decline in the proportion of adults aged 25–64 with low-level qualifi-

cations and an increase in the proportion with high-level qualifications
(Figure 1.13).
In 1997 the average across the sub-set of OECD countries for low-
level qualifications was 37 per cent. This declined to 28 per cent by
2009. There was a significant divergence between the countries during
the period (CV from 0.53 to 0.61; p = 0), as well as an apparent diver-
gence across country groups (CV from 0.46 to 0.49), even though each
country group showed an average decline in the proportion with low
levels of qualifications. The Social Market and Social Democratic groups
of countries had slightly lower proportions than the Liberal countries at

the beginning and end of the period. The Southern European countries
had substantially higher proportions than the rest at the beginning but
closed the gap somewhat by the end.
On the other hand, the proportion of adults with tertiary-level quali-
fications increases on average both across the sub-set of countries (from
21 per cent to 30 per cent) and across country groups. There was no
significant convergence during the period across countries (CV from
0.38 to 0.33; p = 0. 24) or country groups (CV from 0.21 to 0.23). The
Liberal group of countries have the highest average rate at the begin-
ning of the period, followed by the East Asian and Social Democratic
groups, the former of which overtakes the Liberal group towards the
end of the period (see Figure 1.14). The Social Market and Southern
European groups trail the leading group at the beginning by 7 and 11
percentage points, respectively, and at the end remain behind by 12 and
18 points.
The overall trend over the period in terms of adult skills is clearly
upward, with increasing proportions of more qualified adults and
declining proportions of less qualified adults in most countries. How-
ever, in contrast to the convergent trend on PISA scores and upper
secondary graduation rates, there is more evidence of divergent trends.
Overall, this can be seen as more a question of countries and country
groups travelling in the same direction than of converging. Given that
skills stocks, by their nature, can only change very slowly, a convergent
50
40
30
20
10
-
l
ia
pe
tic
ra
ke
As
be
ra
ro
ar
oc
Eu
Li
E.
m
al
S.
de
ci
So
al
ci
So
1998 2003 2009
Figure 1.14 Proportion of adults with tertiary education, 1998–2009

See Education at a Glance, 2011, Indicator A1.
12000
10000
8000
6000
4000
2000
0
an
ls
n
t
at
pa
ke
ra
ne
r
be
oc
Ja
ar
rra
m
m
Li
ite
de
al
ci
ed
al
So
M
ci
So
Total variance 2000 Total variance 2009
Figure 1.15 Changes in total variance by country group, 2000, 2009
trend in flows of upper secondary graduates will only translate into

convergence in adult skills stocks over a very long period of time.
The total OECD variance in PISA scores has declined on average (total
variance down from 9444 to 9090) and there has been significant con-
vergence on this measure both across countries (CV from 0.149 in 2000
to 0.124 in 2009; p = 0) and across country groups (from 0.123 to
0.070). The total variance declined in the Social Democratic, Liberal and
Social Market states by similar amounts, so it is the substantial rise in
variance in Japan, and the lesser rise in the Mediterranean countries,
which is driving country group convergence (Figure 1.15). The total
variance for each country can be divided into within-school variance
and between-school variance.
Within-school variance has gone down in the majority of coun-
tries (declines in 13 countries and increases in 10) with the country
average declining from 6160 to 5825 in a significantly convergent
trend (CV from 0.284 to 0.253; p = 0. 007) (Figure 1.16). Reduction in
within-school variation in many countries would be consistent with the
greater social homogeneity in student admissions to schools which one
might expect to result from the increasing number of countries adopt-
ing school choice policies (OECD, 2012). However, this reduction in
within-school variance is not accompanied by greater variation between
schools.
The average of the between-school variance for the set of countries
has stayed much the same (3240–3223), and there has not been any
significant convergence on this measure across countries (CV for the
country set goes from 0.559 to 0.544; p = 0. 095) (Figure 1.17). Across
10000
8000
6000
4000
2000
0
an
ls
n
at
et
pa
ra
r
ne
k
oc
ar
be
Ja
rra
m
m
Li
de
ite
al
ci
ed
al
So
ci
M
So
Within school variance 2000 Within school variance 2009
Figure 1.16 Changes in within-school variance by country group, 2000, 2009
6000 4817 5090

5000 3994
4000
3000 2310
2000 1260
1000
0
ts
an
ls
n
t
ke
pa
ra
ra
ne
ar
be
oc
Ja
rra
m
m
Li
ite
de
al
ci
ed
al
So
M
ci
So
Between school variance 2000 Between school variance 2009
Figure 1.17 Changes in between school variance by country group, 2000, 2009
country groups there appeared to be a slight divergence (CV from 0.454

to 0.513). This suggests that the reduction in within-school variance in
many countries may be due to other factors – such as greater enforce-
ment of standards and targets and performance monitoring of students –
rather than to greater homogenisation of school intakes as a result of
policies of school choice and selection. This explanation is supported
by the trends in distributions of Economic, Social and Cultural Sta-
tus (ESCS), the OECD’s measure of the economic and social status of
students’ parent(s).
The average of the total variance in students’ ESCS for countries
remained stable between 2003 and 2009 (average from 0.839 in 2003
to 0.801 in 2009) and there was no significant convergence or diver-

gence between countries on this measure (the CV of country variances
rose only marginally from 0.278 to 0.280; p = 0. 243).
Within-school variance in student ESCS declined in 12 countries and
increased in 11 countries, with a slight decline in the average across
countries (CV from 0.623 to 0.595). Liberal, Social Democratic and
Social Market country groups all saw declines in within-school vari-
ance in student ESCS, whereas there were increases in Japan and the
Mediterranean states. There was a significant convergence in within-
school variance in student ESCS across the set of countries (CV from
0.262 to 0.239; p = 0. 025) and apparent convergence across the different
country groups (CV from 0.248 to 0.204).
Between-school variance in ESCS declined slightly across countries
between 2003 and 2009 (CV from 0.457 in 2003 to 0.443 in 2009) and
there was no significant convergence across countries on this measure
(CV from 0.211 in 2003 to 0.230 in 2009; p = 0. 971). Liberal, Social
Democratic, Mediterranean and Social Market country groups all show,
on average, a slight decline in between-school variance in school ESCS
(Figure 1.18).
The general reduction in the variance within countries on PISA scores
seems to be related to a general attrition across countries in academic
selection to schools and ability-grouping within schools, as reported
previously. This puts into question the widespread assumption that
the prevailing trend is towards rising inequality in education. In fact,
0.700
0.600
0.500
0.400
0.300
0.200
0.100
0.000
an
ts
an
t
ls
ke
ra
ra
p
ne
ar
Ja
oc
be
rra
m
m
Li
al
ite
de
ci
ed
So
al
M
ci
So
Between school variance Between school variance

on ESCS 2003 on ESCS 2009
Figure 1.18 Trends in Between School Variance in ESCS

inequality in outcomes for 15-year-olds has reduced slightly. This

change is also reflected in the trend across countries for the ‘education
Gini’ measure, which also shows a slight reduction in inequalities in
adult qualification levels (see Green et al., 2006). The trend in relation to
the country groups seems to have been largely static with one group of
countries (the Nordics) continuing to be comparatively egalitarian while
the Liberal and Social Market countries continued to be comparatively
unequal. The biggest absolute change is for Japan which, between 2000
and 2009, became very much more unequal in terms of the spread of
PISA scores. It moved from being the most egalitarian country in terms
of variance, to being one of the most unequal.
Convergence or divergence?
What does our analysis tell us about whether the OECD countries are
generally converging or diverging on the indicators considered here?
How distinctive do the different country groups remain?
Across the countries in the OECD sub-set there is more evidence of
convergence than divergence. There was significant convergence on 10
of the 24 indicators where significance tests were done, while there was
significant divergence in only two cases. However, on 12 indicators there
was no significant convergence or divergence.
In terms of the country groups, there was also more convergence
than divergence with convergence on 14 of the 24 indicators and diver-
gence on only 8. Although there were more convergent and divergent
indicators across country groups than countries, this may not be very
significant. In the case of the country groups there were too few cases
to apply the significance tests so it may well be that a number of the
examples of convergence and divergence listed here are not significant,
even though they show a change on the coefficient of variation.
There appears to be more convergence in the trends on outcomes
measures than on measures of resourcing and school system character-
istics which suggests the countries are aiming for similar objectives but
sometimes using different policies to get there.
In terms of resourcing there was some convergence on indicators
(like education spending /GDP) which reflect the policy priority gov-
ernments placed on education, but no convergence on the measure
of actual resources deployed in absolute terms (student–teacher ratio).
The share of private spending in total educational spending converged
across countries but not across country groups, and the share of private
spending on tertiary education did not converge significantly at all.
School system characteristics also present a mixed picture. There was

an increase in the proportion of private independent schools across
countries and country groups but the increase was small and does not
indicate the substantial privatisation claimed in some of the literature,
although it is not inconsistent with a picture of increasing marketisa-
tion of public schools (Ball, 2012). The trend was convergent across
both countries and groups but the country groups remain strongly dif-
ferentiated on this measure. Selection to schools by ability generally
decreased across countries but the trend was not convergent. It appeared
to be convergent across country groups but they remained quite dif-
ferentiated. There was a convergent trend across countries towards
less ability grouping within schools but country groups diverged and
remained strongly differentiated. School autonomy was decreasing in
more countries than it was increasing but there was no significant con-
vergence on this measure across countries and divergence across country
groups.
The trends in outcomes, in terms of ‘flows’ of skills, were highly
convergent. Countries and country groups converged on: the aver-
age scores in PISA tests (downwards); the proportions graduating from
upper secondary education (upwards); and the proportions graduating
from long-cycle higher education (upwards). However, the pattern for
trends in ‘stocks’ of adult qualifications was less clear with divergence
across countries in the proportion of adults with qualifications below
lower secondary level (downwards on average) and no significant con-
vergence in proportions with tertiary-level qualifications. Results for
measures of skills distributions were generally convergent. The variance
in PISA scores converged across countries and country groups (down-
wards), as did the within-school variance, but there was divergence
across groups in between-school variance. Trends in ESCS (the measure
of parental class/status) did nor converge or diverge significantly across
countries.2
Overall, we can say there is more evidence of convergence than diver-
gence, both across countries and across country groups. This should
not be surprising since during the period under consideration there
was considerable pressure for the adoption of common policies from
supra-national agencies, and most particularly from the OECD and
the European Commission. This finding would seem to give support
to the view of the World Culture theorists about the general world-
wide tendency towards convergence in education systems. However,
the trends are by no means uniform. On the majority of indicators for
the country-by-country analysis there is no significant change in the
degree of variation between countries. For the country groups there are
more indicators showing convergence (on the CVs) but, as indicated,
some of these may not be significant, so we cannot place much reliance
on this.
While there appears to be more evidence of convergence than diver-
gence during this period, even across country groups, the latter seem
to remain quite distinctive. The most distinctive of the country groups
are the Social Democratic, Social Market and Liberal groups. The
Mediterranean countries form a less distinctive group (and Japan is
omitted from the characterisations below since it does not constitute
a group).
The Nordic countries tend to have high levels of educational funding,
both relative to GDP and in absolute terms, with only a small propor-
tion of this coming from private sources. Their school systems have
rather few private schools and are characterised generally by low levels
of academic selection to schools and ability grouping within schools.
In terms of outcomes they tend to achieve relatively high mean scores
in PISA tests and have high rates of graduation from both upper sec-
ondary education and tertiary education. Their adults are less likely to
have the lowest level of qualifications than in other groups and perform
relatively well in terms of tertiary level qualifications. Distributions of
skills are narrower than in most country groups and between-school
variation in performance is exceptionally low.
The Social Market countries have average levels of funding, com-
pared to our other country clusters, with a relatively low proportion
of this coming from private sources. Consistent with their historical
traditions (Green, 1990) their school systems have relatively few pri-
vate independent schools but a relatively high proportion of private
dependent schools, so that the proportion of full public schools is lower
than in the Nordic countries. Selection to schools by ability is much
more common than in other groups but within school ability-grouping
is at average levels. In terms of outcomes the countries in this group
perform in the middle range of our country groups. PISA scores were
on average lower than in all groups except the Mediterranean group
and the proportion gaining long-cycle tertiary qualifications was lower
than in other groups, but upper secondary graduation rates were high-
est. The proportion of adults with tertiary qualifications is relatively
low, as is the proportion with less than upper secondary qualifications,
confirming claims that Social Market countries tend to have relatively
un-polarised distributions of adult skills (Crouch et al., 1999; Brown
et al., 2001).
The Liberal countries have spending levels which are in the middle
range for our country groups, in spite of relatively high student–staff
ratios. Private spending as a proportion of total education spending is
higher than in most groups, and there is a higher proportion of pri-
vate independent schools than in the other groups. Selection to schools
by ability and ability grouping in schools is in the middle of the range
for the country groups. Countries in this group generally have more
school autonomy than countries in other groups. In terms of outcomes,
the Liberal countries perform above the average in PISA and have rela-
tively high rates of tertiary graduation and high proportions of adults
with tertiary qualifications. However, upper secondary graduation rates
have been relatively low, at least until recently, and the proportion of
adults with less than upper secondary qualifications is still higher than
in Social Democratic and Social Market countries. This latter character-
istic, combined with the relatively high rates of tertiary qualifications
among adults, confirm the characterisation of countries like the United
States and the United Kingdom as having relatively polarised skills distri-
butions among adults (Crouch et al., 1999; Brown et al., 2001). On the
PISA variance measure, skills outcomes at 15 are more unequal than
in Social Democratic and Mediterranean groups and only slightly less
unequal than in the Social Market Group.
Cluster analysis
The distinction between the main country-groups is also brought out by

our cluster analysis. The main objective of cluster analyses is to identify
groups of individuals or countries that bear some similarities according
to a number of dimensions. Hence, individuals in the same group must
be sufficiently different from those in other groups. In a nutshell, clus-
ter analyses consist of three steps. First, one has to choose the variables
according to which the observations are clustered. Secondly, the vari-
ables have to be standardised so they contribute equally to the distance
or similarity between cases. Finally, one has to choose which clustering
procedure to use.
In our analysis the aim is to cluster the countries according to a
number of factors: school choice, school autonomy and the level of
centralisation, student–teacher ratio, selection practices, the proportion
of private dependent and independent schools, levels and dispersion
of performance scores, ability grouping, school social composition,
and expenditure on education. All variables were standardised to have
an average of 0 and a standard deviation of 1. Finally, after some
experimentation we decided to use K-Means clustering to construct the

clusters.
The algorithm used in K-Means clustering starts with an initial set of
means (selected arbitrarily) and classifies cases based on their distances
to these means. Next, it computes the cluster means again, using the
cases that are assigned to the clusters. Then, it reclassifies all cases based
on the new set of means. The process is repeated until cluster means
do not change much between successive steps. Finally, the means of
the clusters are computed once again and cases are assigned to their
permanent clusters.
In Figure 1.19 we present the clustergram of our analyses. Each node
represents a particular number of clusters. For instance the first node
consists of one cluster, the second of two clusters and so forth until
reaching six clusters. Each node is a unique solution, and as the number
of clusters increases some countries move from one cluster to another.
The thickness of the line indicates how many countries we have in the
cluster, the thickest the line is the more countries it has.
In our analysis, the best solution that fits the theory is the five clusters
solution. For this solution we have a distinctive cluster for the Social
Democratic countries (cluster 1), for the Liberal countries (cluster 3),
for the Social Market countries (cluster 5) and for the Mediterranean
.5
AUS, CAN, DEU
GBR, IRL, JPN, 4
NZL, USA 2
NLD
3 LUX, PRT
2
BEL, NLD ESP, GRC,
ESP, GRC, 4 LUX PRT 4
LUX PRT, NLD
ESP, GRC, LUX PRT 2
BEL 3 AUS, DEU, IRL,
Cluster mean
DEU, IRL, AUT, BEL, 2 JPN, NZL, USA

ESP, GRC,
ITA 6
LUX PRT
2 3 FIN,
3
0 AUS, DEU, GBR, ESP,
1 AUS, DEU, GBR, 5 GRC,
IRL, JPN, NZL, IRL, JPN, NZL,
AUT, BEL CAN
USA, NLD, AUT, ITA USA, AUT, ITA
DNK, FIN, ISL, NOR, SWE CHE, ITA
AUS, CAN, GBR, JPN, NZL, 5
USA, NLD, CHE. 1 1 AUT, BEL,
DNK, FIN, CHE, ITA,
ISL, NOR, DNK, FIN, GBR, SWE,
SWE, CAN, ISL, NOR, DNK
CHE SWE, CAN,
CHE 1
DNK, FIN,
ISL, NOR, SWE
1
–0.5 ISL, NOR
1 2 3 4 5 6
Number of clusters
Figure 1.19 Clustergram

countries (cluster 2). The only exceptions are: Germany, which is part
of the liberal cluster; Italy, which is part of the Social Market clus-
ter; Luxembourg, which belongs to the Mediterranean cluster; and the
Netherlands, which forms a cluster of its own.
Conclusions
The analysis here – albeit limited by the restricted range of indicators

available, and the relatively short time period considered – seems to
suggest that there is a basis both for arguments about global conver-
gence and for the claims for persisting distinctiveness between different
country groups. The World Culture and Globalisation theorists and
the comparative political economists and comparative educationalists
would both seem to have some support for their claims. What is per-
haps more surprising in the findings is that the convergent trends often
do not seem to be going in the directions claimed by many theorists of
educational globalisation.
There is a quite widespread belief among globalisation theorists that
‘neo-liberal’ models of education – which emphasise school choice and
diversity, new public management techniques, and the marketisation
and privatisation of aspects of education – are becoming increasingly
dominant worldwide, not least through the influence of supra-national
bodies like the OECD and World Bank (e.g. Lingard et al., 2001; Maroy,
2001; Ball, 2012). The evidence that such policies are increasingly
adopted in the policy rhetoric of OECD countries is certainly strong,
and it confirms Meyer et al.’s notion of a World Culture, at least in
terms of the ostensible policies of the policy elites. However, the trends
observed here for OECD countries do not seem to confirm that struc-
tures and practises are uniformly moving in the directions suggested –
and advocated – by the policies.
Countries are indeed tending to adopt more school choice poli-
cies, albeit of various different kinds (OECD, 2012), and there have
been widespread attempts to implement forms of regulation which give
schools greater autonomy (Maroy, 2001). However, in the 2000s schools
appear to have experienced decreasing levels of autonomy in most
countries. Many countries, including Nordic countries (Wiborg, 2012),
have implemented new measures to encourage private schools. How-
ever, the proportion of fully private schools (i.e. ‘Private Independent’
schools in the OECD categories) and the proportion of funding from
private sources have only increased by a rather small amount on aver-
age across the OECD. This does not mean that public schools systems
have not become increasingly ‘marketised’ through the out-sourcing

of school services and use of private sector management techniques
within schools and education authorities (Ball, 2012). However, in most
OECD countries this does not equate to wholesale privatisation and
the dismantling of public provision, nor even, in the Nordic case, to
the abandonment of comprehensive schooling. It is often predicted
that increasing privatisation and school choice in public schools will
be accompanied by more academic selection and ability grouping in
schools, which will lead in turn to rising inequalities in educational
outcomes (Maroy, 2001). However, the evidence from PISA and other
cross-national sources is that inequalities of outcomes have been gen-
erally decreasing across the OECD, despite, paradoxically, increasing
income inequality. It is undoubtedly the case that a number of English-
speaking countries, including New Zealand, the United States and the
United Kingdom, have introduced more selection and ability grouping
(Whitty, 2002; Ball, 2008, 2012). This may have increased inequalities
in education over the longer term, although the evidence for this is still
disputed. But on evidence here, inequalities in outcomes have actually
decreased (although not by much) in these countries during the 2000s.
Notes
1. One should note that, on the one hand, we compute the CV by dividing the
standard deviation of a particular indicator by its average at the beginning
and end of the reference period. These standard deviations and averages are
computed for the OECD as a whole and for each country group. On the other
hand, the significance of the change between the initial and the final periods
is ascertained using simple regression analyses. First, we compute the differ-
ence between the final and the initial periods for a particular indicator then
we regress this difference on the initial value. A negative regression coeffi-
cient indicates convergence while a positive one indicates divergence, and the
p-value indicates whether the convergence or divergence is significant. In the
text we only report the CVs and the p-values without the regression coeffi-
cients. This is done because it is easier to compare the magnitude on CVs at
the start and the end periods.
2. For the raw trend data, see http://www.llakes.org/publications/the-dynamics-
of-education-systems-convergent-and-divergent-trends-1990-to-2010/
References
Ashton, D. and Green, F. (1996) Education, Training and the Global Economy.
Aldershot: Edward Elgar.
Astiz, M. F., Wiseman, A. W. and Baker, D. P. (2002) ‘Slouching towards Decen-
tralisation: Consequences of Globalisation for Curricular in National Education
Systems,’ Comparative Education Review, 46, 1, pp. 66–88.
Ball, S. (2008) The Education Debate. Bristol: Policy Press.

Ball, S. (2012) Global Education Inc. New Policy Networks and the Neo-Liberal
Imaginary. London: Routledge.
Boli, J., Ramirez, F. and Meyer, J. (1985) ‘Explaining the Origins of Mass
Education,’ Comparative Education Review, 29, 2, pp. 145–167.
Brown, P., Green, A. and Lauder, H. (2001) High Skills: Globalisation, Competitive-
ness and Skill Formation. Oxford: Oxford University Press.
Busemeyer, M. and Trampusch, C. (2011) The Comparative Political Economy of
Collective Skill Formation Systems. Oxford: Oxford University Press.
Crossley, M. W. and Tickly, L.P. (2004) ‘Postcolonial Perspectives and Com-
parative and International Research in Education: A Critical Introduction,’
Comparative Education, 40/41, pp. 147–156.
Crouch, C., Finegold, D. and Sato, M. (1999) Are Skills the Answer? Oxford: Oxford
University Press.
Dore, R. (2000) Stock Market Capitalism Versus Welfare Capitalism. Oxford: Oxford
University Press.
Esping-Andersen, G. (1990) The Three Worlds of Welfare Capitalism. New Jersey:
Princeton University Press.
Esping-Andersen, G. (1999) Social Foundations of Post-Industrial Economies. Oxford:
Oxford University Press.
Epstein, E. and Carroll, K. (2005) ‘Abusing Ancestors: Historical Functionalism
and the Postmodern Deviation in Comparative Education,’ Comparative Educa-
tion Review, 49, 1, pp. 62–88.
Fukuyama, F. (1992) The End of History and the Last Man. New York: Free Press.
Gray, J. (2007) Black Mass: Apocalyptic Religion and the Death of Utopia. London:
Allen Lane.
Green, A. (1990) Education and State Formation: The Rise of Education Systems in
England, France and the USA. London: Macmillan.
Green, A., Preston, J. and Janmaat, G. (2006) Education, Equality and Social
Cohesion. Basingstoke: Palgrave.
Green, A., Mustafa, T. and Preston, J. (2010) The Chimera of Competitiveness: Vari-
eties of Capitalism and the Economic Crisis, LLAKES Research Paper 8. London:
Institute of Education.
Green, A. and Janmaat, J.-G. (2011) Regimes of Social Cohesion: Societies and the
Crisis of Globalisation. Basingstoke: Palgrave.
Green, A., Wolf, A. and Leney, T. (1999) Convergence and Divergence in European
Education and Training System. London: Institute of Education.
Greenfeld, L. (2003) Nationalism: Five Roads to Modernity. Boston: Harvard
University Press.
Greinert, W.-D. (2004) ‘European Vocational Training “Systems” –Some Thoughts
on the Theoretical Context of the Historical Development,’ European Journal of
Vocational Education, 32, CEDEFOP, pp. 18–25.
Griffiths, T. and Knezevic, L. (2012) ‘World Systems Analysis in Comparative
Education: An Alternative to Cosmopolitanism,’ Current Issues in Comparative
Education, 12, 1, pp. 66–75.
Held, D., McGrew, M. A. Godblatt, D. and Parraton, J. (1999) Global Transforma-
tion: Politics, Economics and Culture. Pal Alto: Stanford University Press.
Hall, P. and Soskice, D. (eds) (2001) Varieties of Capitalism: The Institutional
Foundations of Comparative Advantage. Oxford: Oxford University Press.
Huntington, S. (1997) The Clash of Civilizations and the Re-Making of the World
Order. New York: Touchstone.
Hutton, W. (1995) The State We’re In. London: Jonathan Cape.
Hutton, W. (2002) The World We’re In. London: Little, Brown.
Hutton, W. (2007) The Writing on the Wall: China and the West in the 21st Century.
London: Little, Brown.
Hutton, W. (2011) Them and Us. Changing Britain – Why We Need a Fair Society.
London: Little, Brown.
Jacques, M. (2012) When China Rules the World. Harmonsworth: Penguin.
Kerr, C., Dunlop, J.T., Harbison, F. H. and Myers, C. A. (1960) Industrialism
and Industrial Man: The Problem of Labor and Management in Economic Growth.
Boston: Harvard University Press.
Kumar, K. (1991) Prophesy and Progress. London: Penguin.
Jallarde, J. (1989) ‘Recent Trends in Vocational Education and Training:
An Overview,’ European Journal of Education, 24, 2, pp. 103–125.
Lingard, B., Rizvi, F., Henry, M. and Taylor, S. (2001) The OECD, Globalisation and
Education Policy. Oxford: Pergamon Press.
Lipset, M. S. (1959) ‘Some Social Requisites of Democracy: Economic Develop-
ment and Political Legitimacy,’ The American Political Science Review, 53, 1,
pp. 69–105.
List, F. (1885) The National System of Political Economy, trans. Sampson Lloyd:
London Longmans, Green and Company.
Maroy, C. (2001) Regulation and Inequalities in European Education Systems. Final
Report of the research project: ‘Changes in regulation modes and social produc-
tion of inequalities in education systems: A European Comparison.’ Louvain:
Université Catholique de Louvain.
McLean, M. (1990) Britain and a Single Market Europe. London: Kogan Page.
Meyer, J., Boli, J., Thomas, G. and Ramirez, F. (1997) ‘World Society and the
Nation-State,’ American Journal of Sociology, 103, 1, pp. 144–181.
Meyer, J. (2000) ‘Globalisation: Sources and Effects on National States and
Societies,’ International Sociology, 15, pp. 233–248.
Meyer, J. (2004) ‘The Nation as Babbitt: How Countries Conform,’ Contexts, 3,
42–47.
Meyer, J., Bromley, P. and Ramirez, F. (2010) ‘Human Rights in Social Science
Textbooks. Cross-national Analyses, 1970–2008,’ Sociology of Education, 83, 2,
pp. 111–134.
Mons, N. (2007) Les Nouvelles Politiques Educatives. Paris: Presses Universitaires de
France.
Mostafa, T. (2011) ‘Decomposing Inequalities in Performance Scores: The Role
of Student Background, Peer Effects and School Characteristics,’ International
Review of Education, 56, 5, pp. 567–589.
Ohmae, K. (1990) The Borderless World. London: Collins.
Ohmae, K. (1996) The End of the Nation State: The Rise of Regional Economies.
London: Harper Collins.
OECD (2012) Education at a Glance: Highlights. Paris: OECD.
Robertson, R. (1995) ‘Globalisation: Time-Space Homogeneity-Heterogeneity,’ in
M. Featherstone and S. Lash and R. Robertson (eds) Global Modernities. London:
Sage, pp. 25–44.
Rostow, W. W. (1960) The Stages of Economic Growth: A Non-Communist Manifesto.

Cambridge: Cambridge University Press.
Steiner-Khamsi, G. (2004) The Global Politics Of Educational Borrowing and Lending.
New York: Teachers College Press.
Streeck, W. (1989) ‘Skills and the Limits of Neo-Liberalism: The Enterprise of the
Future as a Place of Learning,’ Work, Employment and Society, 3, 1, pp. 89–104.
Streeck, W. (2009) Re-Forming Capitalism: Institutional Change in the German
Political Economy. Oxford and New York: Oxford University Press.
Streeck, W. and Thelen, K. (eds) (2005) Beyond Continuity: Institutional Change in
Advanced Political Economies. Oxford: Oxford University Press.
Thelen, K. (2004) How Institution Evolve: The Political Economy of Skills in Germany,
Britain, the United States and Japan. Cambridge: Cambridge University Press.
Wallerstein, I. (2001) The Essential Wallerstein. New York: New Press.
Whitty, G. (2002) Making Sense of Education Policy. London: Sage Publications.
Wiborg, S. (2012) ‘Neo-Liberalism and Universal State Education: The Cases of
Denmark, Norway, and Sweden, 1980-2011.’ Comparative Education. Published
on-line 24th July 2012.
2
From Upper Secondary to Further
Education: European Models of
Post-Compulsory Learning
Audrey Dumas, Philippe Méhaut and Noémie Olympio
As part of the so-called Lisbon process, the European Union (EU) has
strongly emphasised the importance of education and training systems
for the knowledge society. Recommendations aimed at member states
have gradually been developed, based on indicators of objectives to
be achieved. The open method of coordination has been extended to
encompass vocational training. In the field of higher education, we are
familiar with the so-called ‘Bologna’ process, which seeks among other
things to harmonise the three main levels of education across Europe.
Over the past 30 years, the majority of European countries have seen
strong growth in the numbers of pupils pursuing their studies to upper
secondary. This trend has been continued by the advent of mass higher
education. As part of this exchange of good practice between coun-
tries, objectives have been set, including in an area entitled ‘making
lifelong learning a reality’, another entitled ‘improving equity in educa-
tion and training’ and a third called ‘key competences for young people’
(Commission Européenne, 2007, pp. 3–4).
Given these common objectives, it might be thought that hypotheses
concerning system convergence would increasingly be finding credence.
The aim of this chapter is to use a recent dataset in order to test whether
the idea of a European model can be validated or whether structural dif-
ferences persist and are sufficiently marked to warrant talk of a plurality
of models within the EU. Unlike the many studies focusing on one or
other component of education and training systems (compulsory edu-
cation, initial vocational training, continuing vocational training and
so on), the analysis we offer endeavours to cover both post-compulsory
secondary education, whether of a general or vocational nature, higher
46
Audrey Dumas, Philippe Méhaut and Noémie Olympio 47
education (whether university-based or otherwise) and continuing voca-

tional training. We also look at virtually all EU member states and
Switzerland.
In the first section, we survey the main studies of education and train-
ing models, paying particular attention to the varying potential offered
by these systems in terms of equality. The second section presents the
sources and methodology used. The latter is based primarily on the
‘qualitative comparative analysis’ method (QCA), adapted for small sam-
ples. The third section sets out the results and discusses them in the
light of the content of the literature in the second section. The results of
this chapter also serve as the basis for an analysis of the links between
education and social cohesion, presented in Chapter 10 of this book.
Education and training models: A survey
Many existing studies already offer an analysis of education and/or

training systems and produce typologies. They have a variety of differ-
ent aims (inequalities in access and in success, forms of governance,
relationship to the labour market (Shavit and Muller, 2000; Couppié
and Mansuy, 2004) and to social protection systems). Here, we offer a
brief review of these studies and look at possible convergences between
these analyses with a view to constructing an interpretative framework
covering the state of development and characteristics of education and
training systems towards the end of the 2000s.
Analysing compulsory education: The key question

of inequalities in access and success
There are many studies of compulsory education and its role in reducing
inequalities. They draw in particular on the Programme of International
Student Assessment (PISA) studies, which measure the scale of inequal-
ities in results at age 15, that is the age when compulsory education
normally comes to an end. They show significant differences between
countries, notably depending on whether they have opted for a com-
prehensive education system or practice early selection. However, there
is no clear link between the reforms that have significantly affected
these systems since the early 1960s, and any reduction in inequalities.
Nathalie Mons (2004) has examined the reforms introduced in the late
1990s in a whole series of European countries (introduction (or not)
of a comprehensive system, decentralisation, school autonomy, and
so on). She offers an interpretative framework based on four models.
She firstly distinguishes those systems that have placed the emphasis
on separation as a tool for managing heterogeneity (usually via early
selection and tracking), with the aim of creating uniform groups at a
fairly early stage, from those that regard diversity as an asset and as a
strength to be exploited (comprehensive schools) by enabling pupils to
follow a long common-core syllabus; she calls these ‘integrated systems’.
Although the comprehensive school model officially seems to predomi-
nate in the Organisation for Economic Co-operation and Development
(OECD) countries, it appears that, in reality, this model has gradually
been called into question by a growing diversification of the curricu-
lum. Thus within the integrated model as a whole, there appears to be
(i) an individualised integrated model, which places the main empha-
sis on one-to-one teaching for pupils in difficult circumstances as a tool
for managing heterogeneity; (ii) an à la carte integrated model in which
unofficial selection takes place right from the beginning of secondary
school by means of tracking; and (iii) a uniform integrated model that
does not specifically include any means of managing heterogeneity
and is sometimes characterised by the management of educational fail-
ure. The Scandinavian countries (Finland, Sweden and Norway) have
largely adopted individualised integrated systems. Southern Europe
(France, Italy, Spain and Greece) has placed greater reliance on uni-
form integrated systems, while Continental Europe (Germany, Austria
and Switzerland) has chosen a separation-type model. The Anglo-Saxon
world, for its part, has opted for an à la carte integrated model (United
Kingdom). Her analysis also largely confirms the findings of Green et al.
(2006, Ch. 5, p. 118).
Vocational training models: The key question

of the modes of training
Other studies have focused more particularly on vocational training

models. Greinert (2004), for example, adopts an historical approach in
order to investigate the way systems were created during the Industrial
Revolution. Thus he identifies a market model, a bureaucratic, state-
regulated model and a dual model. His hypothesis is that, even now,
all European countries still rely to a greater or lesser extent on one of
these basic models or on hybrid forms. Each of these models is based on
a dominant form of legitimation and a particular approach to producing
qualifications.
A contrast is usually drawn between models based on apprenticeship,
which may or may not be managed by the social partners, and those
based on schools, which may or may not be state-controlled. Bosch and

Charest (2009) emphasise the dynamics of these systems. In particu-
lar, they stress that the centre of gravity in most vocational training
systems has now shifted towards the ‘upper secondary’ level (and that
they should therefore be linked to compulsory education, of which they
constitute a natural extension). However, they also argue that, with the
advent of mass higher education, these systems have made it possible,
to a greater or lesser extent, for apprentices to gain access to higher
education.
Against this background, and to exaggerate almost to the point of cari-
cature, a contrast might be drawn between: (a) systems that continue the
early tracking practised in compulsory education into upper secondary
education; these systems do little to correct inequalities and provide
few openings for further study; (b) systems that are more open, both
socially and in terms of subsequent opportunities; and (c) highly inte-
grated systems which, in a way, maintain the comprehensive education
model.
Models of skill production and use
Rubery and Grimshaw (2003) focus on analyses based on the opposi-

tion between internal and occupational labour markets. They combine
this with the forms of regulation governing training systems (lightly
regulated, regulated by consensus or state-regulated). In those countries
with a ‘purely’ market-led regime (mainly the Anglo-Saxon countries),
there is little coordination of training. It is up to individuals them-
selves to create their own ‘stock of human capital’ by drawing on a
more or less competitive training offer. This is similar to Mons’ ‘à la
carte’ integrated model. The importance attached to formal qualifica-
tions, especially in recruitment processes, is said to be lower, but new
entrants gain rapid access to the labour market. In countries with strong
occupational labour markets and a high degree of consensus, the sys-
tems not only appear to work well in terms of providing access to the
labour market, but they also seem to play an important role in creating a
person’s occupational identity (‘Beruf ’) and in fostering integration into
social networks. Consequently, it is argued, they contribute to a differ-
ent form of cohesion from the previous case. In countries with strong
internal labour markets, the vocational training systems are said to fall
into one of two categories: poorly developed systems, with most training
taking place in firms (Japan) or systems that are more highly developed
but reliant on the state for coordination (France). In the latter case, it
seems to be harder to gain a foothold in the labour market but subse-

quent careers appear more stable, though on the other hand they rely
to a greater extent on seniority rules. Consequently, inequalities do not
appear to materialise in the same place, depending on how training is
regulated. Duru-Bellat and co-authors (Ch. 11, this volume) speak of ‘the
ascendancy of formal qualifications’ to characterise the links between
vocational training and the labour market but they address different
issues and use a different mode of analysis from those of Rubery and
Grimshaw.
Since the early 2000s, it is studies by adherents of the VOC (Varieties
Of Capitalism) School (Estevez-Abe et al., 2001; Iversen and Stephens,
2008, and so on) that have become the benchmark. For Estevez-Abe
et al. (2001), skill profiles (and thus the various education and training
systems) depend on the prevailing type of welfare state and more partic-
ularly on social protection systems (employment protection, unemploy-
ment insurance, etc.). A low degree of protection for the unemployed
and for employees will encourage individuals to invest to a greater
extent in general skills (in order to facilitate their own mobility within
the labour market). The composition of the labour force will then influ-
ence the strategies pursued by companies, which will adopt technologies
that require fewer specific skills and production strategies that require
general skills. On the other hand, a system that provides a high degree
of job protection for employees and/or a high degree of protection for
the unemployed will favour a production system geared to specific skills,
as investment (by students and companies) in these skills is then less
risky. However, E. Cognard (2010) has argued that VOC analyses tend
to underestimate the ‘political’ aspect of the way in which vocational
training is organised (because employers’ associations are also political
organisations that structure the balance of power).
Building on these studies, and following the logic of the theory of
power resources expounded by Esping-Andersen (1999) and the VOC
School, Iversen and Stephens (2008) explain the emergence of dif-
ferent worlds of human capital formation by historical differences in
terms of the organisation of capitalism and alliances of political parties.
According to them, there are three regimes of human capital formation
(associated with three welfare state regimes):
– Systems (like those of the Scandinavian countries) characterised by a

high level of redistribution, a high level of investment in the public
education system (from pre-school through to higher education) and
in company and/or industry-specific skills and active labour market
policies but a moderate level of employment protection (coordinated

market economy in which social democratic parties are prominent);
– Systems (such as in Germany, Austria or Italy, for example) charac-
terised by a high level of employment protection, vocational train-
ing courses in company and/or industry-specific skills but with a
lower level of expenditure on the public education system (a coor-
dinated market economy in which Christian Democratic parties are
prominent);
– Systems (as in the United Kingdom, for example) characterised by
substantial private investment in education with regard to general
skills but fairly low expenditure both on company and/or industry-
specific skills and on the public education system, redistribution and
employment protection (liberal market economy).
The contribution of the VOC School lies in the connection made

between the organisation of the welfare state (notably the social pro-
tection system) and human capital formation. However, such analyses
underestimate the differences within coordinated market economies,
notably those emphasised by Greinert (2004) or by Gehin and Méhaut
(1993).
Public action regimes and education models in Europe
Building on the societal effect approach and attempting to reconstruct

the direction the EU has taken in recent years in relation to education
and training, Verdier (2009, see also Chapter 3 of this work) describes
five ideal-typical action regimes in the area of lifelong learning (LL). His
aim here is to assess implementation of the principles of justice and
effectiveness against the background of the ‘configuration of political
principles, actors’ logics, rules and instruments’. The various education
systems are generally a combination of these various regimes, for exam-
ple the structure of the English education system contains elements of
both the ‘organised market’ regime and the ‘universalist’ regime. Thus
countries do not fall into any particular category here but constitute
hybridised models.
Olympio (2012) has sought to analyse the various existing education
models through the prism of Amartya Sen’s ‘capabilities’ theory (Sen,
1985, 1992) in order to assess the degree of freedom enjoyed by individ-
uals and the ‘opportunity space’ that exists in the various models. Her
analysis is similar to that of Mons (2004) but includes vocational train-
ing. She identifies four separate education models in Europe. The first is
a purely comprehensive model which, over and above the formal mech-
anisms in place to ensure equality of opportunity, seeks to ensure that
genuine possibilities exist, for example for individuals of modest origin,
by virtue of ‘conversion factors’ that make these mechanisms opera-
tional or real. The second is a market comprehensive model, which,
despite seeking to ensure a single rate of progression, may sometimes
reduce the genuine opportunities/options open to individuals through
‘negative conversion factors’, such as selection or tracking. The third is
a formal comprehensive model that generally limits itself to large-scale
formal mechanisms, without putting in place any genuine possibili-
ties for individuals (for example, by allowing a system of selection by
failure to operate in the area of vocational training). The fourth and
final one is a separation model, which, despite the use of early selec-
tion (and the consequent risk of influence by social circumstances),
puts in place conversion mechanisms and factors, organised around
high-quality vocational training. It should be noted that no European
country has a purely ‘formal comprehensive’ or ‘pure comprehensive’
system: these are ideal types and not a pure and simple characterisation
of education systems.
Generally speaking, education systems appear to be as varied in nature
as the historical and socio-institutional characteristics of the countries
concerned and dependency paths seem to persist (Green et al., 1999).
Have the objectives pursued by the EU changed this state of affairs?
Data, indicators and method
Although there is a considerable quantity of high-quality data available

on compulsory general education, whether it be indicators provided by
the various states via international organisations or data from major
international studies (e.g. PISA), the data on vocational training is
more recent and more patchy. The OECD and Eurostat did not start
to publish comparable indicators until the early 2000s. In 2006, the
UNESCO statistical institute embarked on a similar exercise at global
level. It emphasised the difficulty of defining the scope of vocational
training (Unevoc, 2006). The OECD launched a review of a group of
countries, which was intended to make good this omission (Field et al.,
2009). Eurostat, for its part, makes available a wide range of indica-
tors covering the countries of continental Europe. This is the main
source from which we draw our data, sometimes supplementing it
with data from national sources. The sample covers 25 EU countries
plus Switzerland for data covering the period 2003–2007. This data has
enabled us to produce an appraisal of the situation as it was in the

middle of the previous decade. However, because insufficient time has
elapsed, it does not yet enable us to analyse how the various systems are
evolving.
We should also highlight certain difficulties and weaknesses inherent
in our sources. Even though the classifications are governed by a num-
ber of international standards and recommendations, interpretation of
them is a matter for individual countries. In addition to ambiguities
relating to definitions (e.g. the distinction between post-secondary but
‘non-university’ vocational courses and ‘short vocational degrees’, that
is between ISCED 4 and ISCED 5), there are a number of uncertain-
ties relating to the differences between ‘theory’ and ‘practice’ (see also
Sauvageot, 2008). For example, even though a short vocational degree
course opens up the theoretical possibility of continuing to a high
level on successful completion, it will not be classified as ISCED 3c,
even though in reality very few young people continue their education
beyond this stage. At this stage, all we have at our disposal is indicators
of stocks and flows.
Indicators used
We have used the following indicators (see Table 2.1):
Table 2.1 Indicators.
Level of vocational provision in upper secondary education

Level of PROFSS Students at ISCED Eurostat 2007
vocational 3-PROF level as a % of
provision in all ISCED 3-level
upper secondary students
education
Importance of APPRSS Percentage of OECD 2005
apprenticeship individuals taking ‘Education at a
within secondary apprenticeships in upper Glance 2007’;
education secondary education Cedefop monograph
2009 (LIT, LAT,
ROM, ITA, POR, UK)
Success rates in upper secondary education
Rate of access to ESPSS Participation by Eurostat 2007
upper secondary students at the end of
education compulsory school age
(X + 2) – as a % of the
population aged (X + 2)
Table 2.1 (Continued)
Difference in DTSS Difference in success rate OECD 2005

success rate between vocational and ‘Education at a
between general education (that is Glance 2007’;
vocational and percentage of those with OECD 2006
general education upper secondary leaving ‘Education at a
certificate in the Glance 2008’
population of typical age (FRA); Eurostat
for obtaining this 2007 (LAT, LIT,
qualification) ROM)
Higher education
Rate of access to ESPSUP (ISCED 5–6) by sex and Eurostat 2007
higher education aged 20–24 – as a % of the
population in the
corresponding age bracket
Level of SUPC Importance of vocational Eurostat 2007;
vocational education within Eurostat 2003
provision in post-secondary + higher (LUX)
higher education (ISCED 4 + ISCED
5B/ISCED 4 + 5)
Continuing vocational training
Rate of access to a FC Percentage of the Eurostat 2007
continuing population aged 25–64 ‘Labour Force
vocational that has taken part in Survey’
training course training or teaching over
the past four weeks
Rate of access to a FCD Rate of access to a OECD 2005
continuing training course leading to ‘Education at a
vocational a qualification among Glance 2007’
training course 30–39 year-olds as a % of
leading to a the age group
qualification
For the country codes, see Appendix 2.1.

For some descriptive statistics relating to these rough indicators, see Appendix 2.2.
Methodology
To produce country-specific typologies as a function of the characteris-

tics of education and training systems, we use the QCA method (Ragin,
2008). The QCA method can be broken down into two stages. Stage one
involves creating groups of individuals (or of countries for the purpose
of our study), based on indicators. Stage two then seeks to determine
what links may exist between the defined groups and a result variable.
Only stage one is used in this chapter.
The QCA method requires indicators between 0 and 1. Thus our rough
indicators need to be ‘calibrated’ by determining which values of our
rough indicator will be assigned a value of 0.05, 0.5 and 0.95 for the
calibrated indicator. The way in which these value thresholds are deter-
mined draws on fuzzy set theory. In fact, the whole range of possible
replies should be taken into consideration in order better to assess the
extent to which a given country’s education system has specific charac-
teristics. For example, the vocational pathway’s share in all secondary
pathways may in theory vary from 0 to 100 per cent, but in practice
is seldom equal to 0 per cent or 100 per cent. The thresholds for the
fifth percentile (C05 (x)), the average (x̄) and the 95th percentile (C95 (x))
are chosen on the basis of the distribution of the indicator x. Taking
the average as transition thresholds, we consider that those countries
which, for a characteristic of the given education system, have a degree
of membership above (and, conversely, below) the average of European
countries have a HIGH (and, conversely, LOW) degree of membership
of this characteristic.
The calibration formula applied by the QCA method is as follows:
calibrated indicator = exp(log odds(x))/(1 + exp(log odds(x))).
where log odds(x) = (x − x̄) × In(0. 95/(1–0. 95))/(C95 (x) − x̄) when the
country has a value for indicator x above the average of the sample
considered.
where log odds(x) = (x − x̄) × In(0. 05/(1–0. 05))/(C05 (x) − x̄) when the
country has a value for indicator x below the average of the sample
considered.
The QCA method involves calculating each country’s membership
score for a combination of indicators, adopting the minimum of the
various degrees of membership for each indicator included in the com-
bination. The indicators considered should be between 0 and 1, with a
value close to 1, indicating that a country is strongly characterised by
the indicator. For example, let us consider three indicators for countries
X and Y (Table 2.2).
Table 2.2 Example of membership.
A B C (1-A)
X 0.6 0.7 0.8 0.4

Y 0.4 0.8 0.9 0.6
The membership score obtained by country X for the combination

characterised by high levels of A, B and C is 0.6, and 0.4 for country Y.
On the other hand, if we consider the combination characterised by
high levels of B and C and a low level of A, that is the inverse of A (1-A),
then the score is 0.4 for country X and 0.6 for country Y. A country is
then characterised by the combination of indicators whose membership
score is above 0.5. In point of fact, this means that for each characteristic
in the combination, the country has a greater chance of belonging to it
(because the degree of membership is above 0.5) than of not belonging
to it. In the above example, country X is therefore characterised by an
education system with high levels of A, B and C, whereas country Y is
characterised by an education system with high levels of B and C and a
low level of A.
Which configurations for national systems?
Our analysis involves looking beyond compulsory education in order

to investigate the different configurations of education and training
systems in Europe. It is also our aim to explore a number of more or
less hybrid configurations, following Verdier (Chapter 3 of this book).
Lastly, it seeks to establish (or refute) links between system architectures
at lower levels and their relationships at higher levels.
Upper secondary education configurations
We are concerned initially with the characteristics of post-compulsory

secondary education, the dichotomy between general and vocational
education (as measured by the level of vocational provision (‘vocation-
alisation’) in secondary education (PROFSS), and with the importance
of apprenticeships within secondary education (APPRSS)). This brings
us close to the debates opened up by Greinert (2004), Estevez-Abe et al.
(2001) and Bosch and Charest (2009). We then combine this initial anal-
ysis with the question of access to and success rates in upper secondary
education. Is there a relationship between the form taken by upper sec-
ondary education and the development of mass access and the success
rates in general and vocational education?
Initially we reach a fairly familiar finding, in which a distinction
is made between countries with a high level of vocational education
at secondary level, coupled with a predominant apprenticeship system
(Austria, Switzerland, Germany, Denmark, the Czech Republic, Slovakia
and the Netherlands) and the rest (see Table 2.3). This first group
Table 2.3 Level and mode of vocational provision in secondary education.
Name PROFSS APPRSS Number of Country’s name and degree of

of countries membership of group
group concerned
G0 LOW LOW 11 GRE (0.92); EST (0.92); IRL (0.9);

POR (0.84); LAT (0.84);
LIT (0.84); UK (0.78); SPA (0.73);
FRA (0.72);
POL (0.71); HUN (0.66);
G1 HIGH LOW 8 BEL (0.91); SLN (0.84); ROM (0.84);
FIN (0.76);
ITA (0.73); SWE (0.66); NOR (0.65);
LUX (0.64);
G2 HIGH HIGH 7 CZE (0.86); SWI (0.84); AUS (0.82);
DEN (0.825);
SLK (0.81); GER (0.67); NL (0.58);
Interpretation: Three groups of countries are obtained, to which are assigned a name (G0,
G1 and G2) and a font code (normal, italics, bold) in order that these initial groups can be
identified in each of the subsequent typologies.
The statistic in brackets after the country name denotes the degree of membership of the
grouping. The higher this statistic, the more the country is representative of this typology.
clearly belongs to the ‘Rhenish corporatist’ model. However, Germany

and, in particular, the Netherlands have a lower membership index,
reflecting changes in the importance of apprenticeships (see Bosch and
Charest (2009), for Germany; in the case of the Netherlands, Westerhuis
(2011) emphasises the co-existence of an apprenticeship pathway and
an academic pathway leading to the same qualifications). Interestingly,
although our data relates mainly to the late 2000s, two of the former
Eastern bloc countries have maintained (or rediscovered) systems with
a strong emphasis on apprenticeships, which contrasts with the other
former Eastern bloc countries (e.g. Poland and Hungary). It is known,
for example, that Poland, which in the early 1990s still had an appren-
ticeship system in the manual trades and in SMEs (OCDE, 1993), has
now lost this characteristic.
However, this familiar distinction is by no means the end of the
story. There is a second sub-group of countries whose secondary edu-
cation systems provide considerable amounts of vocational education
and training, though not predominantly in the form of apprentice-
ships. It includes the Nordic countries (Sweden, Finland, Norway),
Belgium and Luxembourg and the Latin fringe of Southern Europe (Italy,
Romania, Slovenia). Norway, Luxembourg and Sweden have the lowest

membership indices. The first two have a modest mix of academic and
apprenticeship pathways. In Sweden, which does not have an appren-
ticeship system, general and vocational education are closely integrated,
which must explain its membership index. We shall see in due course
that this in fact reflects two different organising principles governing
the development of vocational training pathways.
A third and final group is characterised by a fairly small vocational
element in their secondary education systems, combined with low use
of apprenticeships. It includes the countries characterised by the VOC
school as general systems (Ireland, the UK), most of the countries
with a ‘general’ meritocratic regime in the Latin tradition (Spain, Italy,
Portugal, Greece, France) and a number of former Eastern bloc coun-
tries (the Baltic states, Hungary and Poland). Within this group, Spain,
Poland, France and Hungary have the lowest membership indices. In the
case of the last two countries, and to a lesser extent Poland, this is linked
to the existence of an apprenticeship pathway within vocational educa-
tion; admittedly this is lower in the first group, but it is nonetheless
significant.
Can a link be established between this structuring of upper secondary
education and the characteristics of access and/or success? The answer
varies, depending on the groups just outlined above (see Table 2.4).
Table 2.4 Access to and success in upper secondary education.
ESPSS DTSS Typology No. of Countries

Countries
LOW LOW G0 4 SPA (0.73); FRA (0.66); HUN (0.66);

POR (0.56),
G1 1 ROM (0.57)
G2 1 DEN (0.56)
LOW HIGH G1 2 ITA (0.62); LUX (0.6)
G2 1 NL (0.58)
HIGH LOW G0 5 LIT (0.84); POL (0.71); LAT (0.68);
GRE (0.66);
EST (0.55)
G1 2 SWE (0.66); NOR (0.65);
HIGH HIGH G0 1 IRL (0.53);
G1 2 SLN (0.83); FIN (0.52),
G2 5 CZE (0.86); AUS (0.82); SWI (0.79);
SLK (0.58); GER (0.55)
Not classified: BEL, UK.

Among the countries with high levels of vocational education and

apprenticeships (corporatist regimes in Verdier’s definition), the major-
ity (five out of seven) are characterised by both very high rates of access
to upper secondary education (around 90 per cent) and a differences in
success rates that favours vocational education. Although most of these
countries have selective compulsory education (as defined by Mons,
2004, but with the notable exception of Denmark, which has adopted
the comprehensive education model), this is not necessarily accom-
panied by early exclusion from secondary education, even though it
is a predictor of ‘tracking’ into vocational education in the form of
apprenticeship. Denmark lies outside this group, with a rate of access to
post-compulsory secondary education that is below the European aver-
age. Here, however, age management, which enables students to re-enter
post-compulsory education at a later date, is likely to have an effect (our
indicator relates to access two years after compulsory education). Only
the Netherlands deviates significantly (confirming its already marginal
position with regard to the ‘vocationalisation’/apprenticeship combi-
nation). Although the ‘success premium’ in vocational education is
significant, the access rate to upper secondary education does not lie at
the top of European countries. In addition to early selection in compul-
sory education (pathways are determined at 12), this country appears to
have a particularly hybrid system in terms of the status of vocational
education: academic and apprenticeship pathways co-exist, together
with an organising principle based on division that borrows from the
meritocratic model.
Within the group of countries in which a high level of vocational
education is provided within the secondary system but which do
not necessarily have an apprenticeship system, a number of different
configurations can also be discerned. The Nordic countries (Norway,
Sweden) have rates of access to secondary education that are similar or
even higher than the previous group. Comprehensive lower secondary
education is followed by mass access to upper secondary education
which, though admittedly ‘vocationalised’, is generally fairly integrated.
It could be described as a ‘comprehensive’ secondary education system.
The differences in success rates between general and vocational educa-
tion are not significant. Slovenia and, to a lesser extent, Finland (with an
index of 0.52) form a second sub-group; this too has a high rate of access
to upper secondary education, but it goes hand in hand with a ‘success
premium’ for vocational education. The third sub-group encompasses
countries, which, despite their high levels of vocational provision,
have fairly low rates of access to upper secondary education (Italy,
Luxembourg, Romania), despite a success premium favouring vocational
education (in the case of Italy and Luxembourg). Unlike countries in the
apprenticeship and integrated vocational training groups, countries in
this group seem to regard the vocational pathways as destinations for
pupils deemed to have failed academically. Pupils relegated to the ‘pre-
vocational’ streams in lower secondary school tend to remain in the
vocational pathways for their post-compulsory education.
Let us turn now to the group in which both vocational education and
the apprenticeship system are underdeveloped. It divides broadly into
two sub-groups, although the relationship to the differences in success
rates between vocational and general education is not significant (except
in the case of Ireland). The first sub-group consists of countries that are
above or close to the European average for access to secondary educa-
tion. In this group, the scores are sometimes close to or even higher
than those of the previous groups: an upper secondary school system
that is open to a wide range of students and offers both general and (in a
minority of cases) vocational education allows mass education to be pro-
vided beyond the age of compulsory education. There is no ‘success
premium’ for vocational education. This situation obtains in Greece,
Poland and the Baltic States. In the second sub-group, opportunities for
post-compulsory study are below or close to the European average; in
principle, therefore, early exit rates from education will be higher. This
particularly applies in the United Kingdom, Spain, and Hungary and,
to a lesser extent, Portugal and France. Here, the principle of merito-
cratic selection seems to prevail and vocational education is not really
an option: it is largely underdeveloped and there is no significant differ-
ential in success rates. The only exception is Ireland, where the highly
developed upper secondary school system has a ‘success premium for
vocational education’.
Access to higher education
Can a link now be established with access to higher education? In this

section we are exploring, firstly, rates of access to higher education,
and secondly, in connection with those access rates, the question of
short vocational degrees. Most of the literature is, after all, concerned
with university-based higher education only. In the period under review,
short vocational degrees offered opportunities in some countries for
large numbers of individuals to pursue their studies. This is a factor
in the appeal of vocational education at secondary level (access can be
gained to higher education) and may indeed help to compensate for
inequalities of opportunity in secondary education. Conversely, it may
be that a low level of vocational education at secondary level can be

offset by an expansion of short vocational degrees in order to cope with
certain needs in the labour market.
Unsurprisingly, in the light of numerous other studies, our own anal-
ysis tends to show that, as far as higher education ‘for young people’
is concerned (our variable concerns the 20–24 age group), most of the
countries in which vocational training and apprenticeship predomi-
nate belong to the low access rate category (Austria, Slovakia, Denmark,
Germany and Switzerland) (see Table 2.5). Switzerland is an exception
here, though the existence of the vocational pathways does not com-
pensate for the overall low rate of access to higher education (access
to general higher education continues to be dependent on obtaining
the general or vocational upper secondary school leaving certificate
(maturité/Matura), which is still difficult for apprentices). Even though
‘mass vocational training through apprenticeship’ is a factor in devel-
oping mass access to and the homogenisation of secondary education,
it does not act as a gateway to higher education, at least not directly.
The construction of ‘intermediate’ categories positioned between skilled
workers and technicians which, as we shall see, tends to take place
through continuing training and the divide between vocational train-
ing and a university education system that recruits primarily on the
basis of academic criteria appear to prevail. However, two countries – the
Table 2.5 Access to higher education.
ESPSUP SUPC Typology No. of countries Countries
Low Low G0 2 POR (0.84); SPA (0.65);

G1 2 ROM (0.69); SWE (0.6);
G2 4 AUS (0.77); SLK (0.74); DEN
(0.63); GER (0.58);
Low High G0 3 IRL (0.9); FRA (0.62), UK (0.54)
G2 1 SWI (0.57)
High Low G0 3 POL (0.71); LAT (0.64); HUN
(0.52);
G1 3 FIN (0.76); NOR (0.61);
ITA (0.55);
G2 2 NL (0.58); CZE (0.51);
High High G0 3 GRE (0.85); LIT (0.73); EST
(0.56)
G1 2 SLN (0.84); BEL (0.57)
Not classified: LUX.

Netherlands and the Czech Republic – go their separate ways in this

regard. In the first case, the ambiguous nature of the Netherlands’ posi-
tioning, which seems strongly hybrid, once again becomes evident.
It may be that the mix of general and vocational education, com-
bined with the system’s division into different levels, opens doors to
higher education (level five in the Netherlands’ schema), even though
short degrees (which do exist) are not classified as higher education
qualifications.
In the group of countries with significant vocational provision at sec-
ondary level but no apprenticeship system, the heterogeneity previously
noted is confirmed at higher education level. In some of the countries,
there is a high rate of access to higher education (Finland, Norway, Italy),
though this is not attributable to the (statistically recorded) existence
of short vocational pathways in higher education. In some cases, it is
due to the integrated and comprehensive nature of general and voca-
tional secondary education, which is highly likely to make it accessible
to all categories of pupils (Finland, Norway). In the case of Italy, on the
other hand, the explanation lies more in the large numbers of individ-
uals continuing their studies at the end of general secondary education
(even though this country is not noted for a particularly strong system
of upper secondary education), but with poor opportunities for those
from a vocational education background. Slovenia and Belgium, which
also have high rates of access to higher education, owe this more to the
existence of short vocational degrees. There are two countries in this
group to which these findings do not apply. Firstly, Sweden’s rate of
access to higher education is classified as low (at 29.2, it is below the
European average), whereas there is mass access to its upper secondary
education system. However, this may be due to our indicator, which
takes account only of the 20–24 age group, whereas it is known that
opportunities exist here for later entry. Secondly, Romania, a country
characterised by a poorly developed upper secondary education system,
has, fairly logically, a low rate of access to higher education.
Turning now to the last group with its low levels of vocational pro-
vision at secondary level, we find the same two sub-groups that were
initially identified with regard to access to upper secondary education.
The countries that were characterised by mass access to upper secondary
education, without any significant difference in success rates between
the dominant general and underdeveloped vocational streams, are also
those with high rates of access to higher education (except for Hungary),
five of which are among the new EU accession states (Poland, Latvia,
Lithuania, Estonia and Hungary) plus Greece. Nevertheless, two higher
education strategies can be discerned. The first is characterised by mass

access, due in part to the existence of short vocational degrees (voca-
tional provision deferred until after secondary education in Greece,
Lithuania and Estonia), to traditional full-length degree courses (Poland,
Latvia). Conversely, the countries that were characterised by low rates
of access to upper secondary education (Portugal, Spain, France) are in
the low category for access to higher education (with the exception of
Ireland) and are joined by the United Kingdom. In all of these coun-
tries, after all, access to upper secondary education is below average,
the logical consequence of which is a ‘delay’ in access to higher educa-
tion, though this can also sometimes be explained by higher exit rates
at secondary level. In three cases (Ireland, France and the UK), short
vocational pathways exist on a significant scale.
Continuing vocational training from the lifelong

learning perspective
We now come to the question of access to continuing vocational train-

ing. This may, after all, constitute an alternative pathway, compensating
for the effects of a lower access rate to upper secondary education (or
early exits), or even offer opportunities to pursue studies at a later stage
at higher education level. Here too, there are correlations with the char-
acteristics of initial training. The Nordic countries, regardless of their
configuration in terms of initial training, are characterised by a high
access rate to continuing vocational training, including courses leading
to qualifications with Denmark confirming its atypical position relative
to the countries with an apprenticeship system (see Table 2.6). In all of
these countries, to which must be added the United Kingdom (though it
should be stated that the data for this country is often called into ques-
tion), the intergenerational gap is small – which might be explained by
the high level of continuing vocational training.
The group of countries with strong apprenticeship systems, which
had been relatively homogeneous up to this point, now splits into two
when we take continuing vocational training into account. Switzerland,
the Netherlands, Denmark and Austria are at a somewhat higher level
in terms of access to continuing vocational training, though not nec-
essarily leading to a qualification, except in the case of Denmark, but
this is probably more as a result of large access and later access to
apprenticeship (the average age of apprentices is 25).
In the case of countries with vocational pathways but no apprentice-
ship system, two groups can be identified. The Scandinavian countries
Table 2.6 Access to continuing vocational training.
FC FCD Typology No. of Countries

Countries
Low Low G0 5 POL (0.71), POR (0.84); IRL (0.77)

SPA (0.57); FRA (0.72);
G1 2 ITA (0.73); LUX (0.64)
G2 3 CZE (0.85); SLK (0.81); GER
(0.67);
Low High G0 2 HUN (0.51); GRE (0.55)
G1 1 BEL (0.75)
High Low G2 3 SWI (0.84); NL (0.58); AUS (0.56);
High High G0 1 UK (0.78)
G1 3 FIN (0.76); SWE (0.66); NOR (0.62)
G2 1 DEN (0.7)
Not classified: EST; LAT; LIT; ROM; SLN.
(Finland, Sweden and Norway) do not have high rates of access to con-
tinuing vocational training leading to a qualification. On the other
hand, Italy, Luxembourg and Belgium have low rates of access to contin-
uing vocational training, though this still leads to a qualification in the
case of Belgium. Lastly, all of the countries within the group with a low
level of vocational education, except for the United Kingdom, have low
rates of access to continuing vocational training. These training courses
do not usually lead to qualifications, except in the United Kingdom.
All of the other countries, regardless of their configuration in terms
of initial training (there may be a strong emphasis on apprenticeship,
as in Germany, a strong vocational secondary education system, as in
Belgium, or a strong general secondary education system, as in Poland)
have a low rate of access to continuing vocational training. Among
them, only Hungary, Greece and Belgium have high rates of access to
training courses leading to qualifications (it should be borne in mind,
however, that the values for this indicator are often very low, i.e. around
2–3 per cent).
Thus it is difficult in these countries to attribute the intergenerational
differences in qualifications to continuing vocational training. They are
more likely to be a reflection of the time at which mass secondary edu-
cation came into being. This appears to have been relatively early in
recent decades in the remainder of the countries with highly developed
apprenticeship systems (Germany, the Czech Republic, Slovakia), as well
as in Poland and Luxembourg, where the intergenerational differences
are small, in contrast to the situation in Poland, Ireland, Spain, France,

Greece and Belgium.
Conclusion
In the countries with strong vocational training systems, three mod-

els can be identified. The first of these models can be described as
‘comprehensive’. Vocational training is integrated into the system, and
failure rates are low. We can perhaps speak of an ‘integrative comprehen-
sive education system’, which transmits a common set of values and is
intended to forge common identities, regardless of the educational path
chosen. This is the situation in Finland, Sweden and Norway. This model
is similar to Verdier’s ‘universalist’ regime, where the aim is to promote
education for all in accordance with the equal opportunities principle
(this seems to be borne out by access to training beyond the level of
compulsory education and equality of opportunity between vocational
and general education). To some extent, this group might also be com-
pared to Mons’ ‘individualised integration’, in the sense that the main
goal of this model is to adhere to the principles of comprehensive edu-
cation. The Netherlands and Belgium have a number of similarities with
this model.
The second can be labelled a first separation-type model, but with
an integrative vocational training system. This applies to the countries
with a strong emphasis on apprenticeship: Austria, the Czech Repub-
lic, Slovakia, Switzerland and to a lesser extent, Denmark and Germany.
Vocational training is separated from general academic education, on
the basis of ‘early tracking’. Prospects for access to higher education are
poor. However, due to the homogeneity and strength of the appren-
ticeship system, this separation does not prevent the construction of
strong occupational identities, which relate to large numbers in any
given generation. In this sense, this particular group appears compa-
rable to Verdier’s ‘vocational regime’, insofar as it involves an education
model providing ‘access to a professional community’.
The third model can be described as the second separation-type
model, but one which is less integrative. Vocational training (some
of which is school-based) is strong but often provided through short
courses. ‘Drop-out’ rates are average or high, and the opportunities to
carry on into higher education are fairly poor. In some ways, there is a
kind of tracking into upper secondary education, which may or may not
follow on from ‘comprehensive’ lower secondary education. This group
appears to be positioned mid-way between the ‘separation’ model and
Mons (2007)’ ‘uniform integration’ model, in which the inclusive nature

of comprehensive education (based on a common-core syllabus through
to the end of lower secondary school) is more a formal concept than a
reality. Within this group, the separation of pupils into different upper
secondary pathways is not offset as previously by the integrative nature
of the vocational training. Italy, Romania and Slovenia belong to this
group, although it does not clearly equate to the ‘republican’ cohesion
model identified by Green and co-authors.
In the countries that have a fairly poorly developed vocational train-
ing system, there is a greater degree of heterogeneity, and above all,
there is nothing allowing a contrast to be drawn, as the VOC litera-
ture typically does, between market economies and the rest. In point of
fact, this group encompasses a number of Continental Europe countries
(mostly in the South), together with Ireland and the United King-
dom, the Baltic States, and some Central European countries (Poland,
Hungary). In most cases a ‘meritocratic’ principle seems to be at work
throughout these systems, which therefore seem comparable to Verdier’s
‘academic’ regime (Verdier, 2009), which is characterised by the ascen-
dancy of qualifications and educational meritocracy. Nevertheless, these
mechanisms may come into play as early as lower secondary education
or only after the end of a more or less comprehensive lower secondary
education (as in the UK). They may be strong and exclusive in upper
secondary education (Spain, the United Kingdom) or weaker (Poland).
Here, we may speak of a separation between general and vocational
education (in which the latter usually takes in those excluded from
general education). This does not, however, completely close the door
to higher education, especially where short vocational degree courses
exist. Yet it may be thought that these ‘deferred’ vocational pathways
act initially as a destination for some of those who quit general edu-
cation, with vocational education then tending to be a second-best
option.
Ultimately, our results show that the common trends in the devel-
opment of mass upper secondary education and, subsequently, higher
education have not eliminated the structural differences that exist
between Europe’s various education and training systems. The impor-
tance of vocational training and the forms it takes (apprenticeship
or otherwise) remain distinguishing variables, including within higher
education. The overall perspective adopted here (which includes con-
tinuing vocational training) reveals different combinations and forms
of continuity or separation between the various stages of these systems.
This analysis also highlights the fact that ‘pure’ models derived from the
literature (be it VOC or older studies) are sometimes shown to be more

complex as soon as all forms of skill production are taken into account.
Consequently, they are sometimes also hybrid in nature, although in
view of the longitudinal nature of our data it cannot be determined
whether this hybridisation process is becoming more accentuated.
Nevertheless, our results illustrate the robustness of education mod-
els, which maintain complex relationships with the societies that create
them and which they fashion. The French-style system of meritocracy
and the German notion of Beruf both have deep historical, social and
economic roots. Whether in the VOC literature or in studies adopt-
ing the societal effect approach (Maurice et al., 1982, 1986), it is the
interactions between the various parts of the education and training
system, labour markets and company policies that help to explain
dependency paths, even where there are strong internal developments
within systems. The way in which the various economies are based on a
distribution of skills (e.g. a high level of intermediate skills in Germany,
a bipolarisation between low skills and very high skills in the UK and a
more subtle gradation in France, but with very low skilled individuals
being excluded due to the abundant supply of graduates) and compa-
nies do or do not make use of continuing vocational training reinforces
the strategies pursued by pupils and employees. European injunctions
and examples of good practice are feeding into national debates, and
thereby contributing to forms of hybridisation. However, we shall see in
Chapter 10 that the effects of these systems in terms of social cohesion
are quite different from the usual presuppositions regarding equality and
equity in education and training systems.
References
Bosch, G. and Charest, J. (eds) (2009) Vocational Training, International Perspectives.
New York: Routledge.
Cedefop (2009) Initial Vocational Education and Training (IVET) in Europe. Review.
http://www.cedefop.europa.eu/FR/Files/IVET_Review_08.pdf
Cognard, E. (2010) Économie politique des employeurs et néo-corporatisme: Financer
la formation professionnelle continue en Europe, thèse de doctorat en sciences
économiques, Université Lumière Lyon 2.
Commission Européenne (2007) Un cadre cohérent d’indicateurs et de critères de
référence pour le suivi des progrès accomplis vers les objectifs de Lisbonne dans le
domaine de l’éducation et de la formation. http://eur-lex.europa. eu/LexUriServ/
LexUriServ.do?uri=COM:2007:0061:FIN-:FR:PDF
Couppié, T. and Mansuy, M. (2004) L’insertion professionnelle des débutants
en Europe: des situations contrastées, Economie et Statistique, CCCLXXVIII–
CCCLXXIX, 47–167.
Esping-Andersen, G. (1999) Les trois mondes de l’Etat-providence: essai sur le

capitalisme moderne. Paris, PUF, coll. ‘Le lien social’.
Estevez-Abe, M., Iversen, T. and Soskice, D. (2001) ‘Social Protection and the
Formation of Skills: A Reinterpretation of the Welfare State’ in Hall, P.A. and
Soskice, D. Varieties of Capitalism: The Institutional Foundations of Comparative
Advantages. Oxford: Oxford University Press, pp. 145–183.
Field, S., Hoeckel, K., Kis, V. and Kuczera, A. (2009) ‘Learning for jobs’, OECD
Policy Review of Vocational Education and Training (initial report, OECD).
Géhin, J.-P. and Méhaut, Ph. (1993) Apprentissage ou Formation Continue? Stratégies
éducatives en Allemagne et en France. Paris: L’Harmattan.
Green, A., Preston, J. and Janmaat, J.G. (2006) Education, Equality and Social
Cohesion: A Comparative Analysis. New York: Palgrave MacMillan.
Green, A., Wolf, A. and Leney, T. (1999) ‘Convergence and Divergence in
European Education and Training Systems’, Bedford Ways Papers, University of
London, Institute of Education.
Greinert, W.D. (2004) Les systèmes européens de Formation professionnelle,
réflexions sur le contexte de leur évolution historique. Formation Professionnelle,
XX, 19–27.
Iversen, T. and Stephens, J. (2008) ‘Partisan Politics, the Welfare State and 3
Worlds of Human capital formation’. Comparative Political Studies, XXXXI,
600–637.
Maurice, M., Sellier, M. and Silvestre, J.J. (1982) Politiques de l’éducation et Organ-
isation industrielle en France et en Allemagne: essai d’analyse sociétale. Paris, PUF,
coll ‘Sociologies’.
Maurice, M., Sellier, F. and Silvestre, J.J. (1986) The Social Foundations of Industrial
Power. Cambridge: MIT Press.
Mons, N. (2007) Les nouvelles politiques éducatives: la France fait-elle les bons choix?
Paris: PUF, coll. ‘Education et Société’.
Mons, N. (2004) De l’école unifiée aux écoles plurielles: évaluation internationale
des politiques de différenciation et de diversification de l’offre éducative. Thèse de
Doctorat inédite en Sciences de l’Education, Université de Bourgogne, 537p.
OCDE. (1993) Le marché du travail en Pologne. Paris, éditions de l’OCDE.
Olympio, N. (2012) Les contextes éducatifs européens à l’épreuve de la théorie
des « capabilités » d’Amartya Sen. In Picard et Masdonati (dir.) Les parcours
d’orientation des jeunes: Dynamique institutionnelles et identitaires, PUL.
Ragin, C. (2008) ‘Qualitative Comparative Analysis Using Fuzzy Sets’ in Rihoux,
B. and Ragin, C. (eds) Configurational Comparative Analysis, Sage Publications,
Chapter 5.
Rubery, J. and Grimshaw, D. (2003) The Organization of Employment, an Interna-
tional Perspective. London, New York: Palgrave Macmillan.
Sauvageot, C. (2008) Un outil au service des comparaisons internationales:
la classification internationale type de l’éducation. Education et formations,
CCCLVIII.
Sen, A. (1985) Commodities and Capabilities. Londres: Elsevier Science Publishing
Company.
Sen, A. (1992) Inequality Re-examined. Harvard: Harvard University Press.
Shavit, Y. and Müller, W. (2000) Vocational Secondary Education: Where Diver-
sion and Where Safety Net? European Societies, 2(1), 29–50.
Unevoc, UIS. (2006) Participation in Formal Technical and Vocational Education and
Training Programmes Worldwide, an Initial Statistical Study. Bonn: Unevoc.
Verdier, E. (2009) L’éducation et la formation tout au long de la vie: une orien-
tation européenne, des régimes d’action publique et des modèles nationaux en
évolution. Sociologie et Sociétés, XL.1.
Westerhuis, A. (2011) ‘The Meaning of Competence’ in Brockmann, M., Clarke,
L. and C. Winch (eds) Knowledge, Skills, Competence in the European Labour
Market: What’s in a Qualification? London, New York: Routledge, pp. 68–84.
3
Lifelong Learning Regimes versus
Vocational Education and Training
Systems in Europe: The Growing
Hybridisation of National Models
Eric Verdier
Any attempt to compare national vocational education and training

(VET) systems comes up against the constraint of complexity, for such
schemes are embedded in other social sub-systems (e.g. labour relations,
basic education, higher education, labour market, company manage-
ment). In addition, focusing on the national frame of reference alone
may be problematic because of the multiplicity of levels stemming,
on the one hand, from increased decentralisation and, on the other,
from a supra-nationalisation which introduces a new level of regula-
tion. In rhetorical terms, the expression ‘lifelong learning’ both reflects
this complexity and accentuates it still further, since it endows this
resource with a new purpose, namely that of helping to safeguard career
paths and job transitions, to the point of making it a ‘pillar’ of social
protection systems.
How then can we compare national systems without falling into the
trap of trying to compare the incomparable? Several approaches, rooted
in different theoretical and disciplinary traditions, may be drawn upon
(Giraud, 2004):
The comparative political economy approach known as ‘varieties of
capitalism’ (Hall and Soskice, 2001) focuses on the institutional arrange-
ments which coordinate individual choices and company strategies.
Individuals seek to protect themselves from the risks of unemployment
and downgrading of their investments in human capital (Estevez-
Abe et al., 2001). Liberal market economies and coordinated market
economies can be identified by the specific features of their modes of
70
Eric Verdier 71
coordination. In the latter, young people know that it is not irrational

to invest in specific or semi-specialised training. In the liberal mar-
ket economies, on the other hand, acquiring general skills allows
individuals to confront the uncertainties of the labour market.
The dualistic approaches (actors vs. institutions) emphasise intermedi-
ate regulations between individual choice and the macro-social order.
Thus Culpepper (2003) shows why and how the successes and failures
of new youth training schemes in France and the former East Germany
may be explained by the quality of the local networks appropriating
the national rules. Earlier research in the tradition of societal analysis
drew on a comparison between France and Germany in order to high-
light the coherence of the wage-labour nexus specific to each country by
analysing the interactions between three spheres: labour relations, work
organisation and education and training. Maurice et al. (1986) empha-
sised what, in their view, were the fundamental social mediations of
the institutions specific to each country, such as the dual system of
vocational training in Germany, which supports occupational labour
markets and favours a form of work organisation that is considerably
less hierarchical than in France, where the education system functions
primarily as a process for the reproduction of elites.
Finally, in socio-historical approaches, comparisons are effected by trac-
ing the genesis and evolution of the national institutions, public issues
and social categorisations at work. Applied to VET in three national con-
texts, the historical institutionalism of Thelen (2003) reveals forms of
change which are then conceptualised (Streeck and Thelen, 2005) and
denoted by the terms displacement, layering, drift and conversion.
The approach in terms of lifelong learning (LLL) regimes developed in
this chapter combines societal analysis with socio-historical approaches.
It takes as its starting point the hypothesis that several factors are foster-
ing the hybridisation of national VET systems. In a context of economic
uncertainty and demands for effective public spending, these factors
include the successive attempts of public policy to respond to recurring
labour market difficulties, especially with regard to the (re)integration of
young people and the unemployed. Quite often, training reforms have
drawn their inspiration from (supposed) ‘successes’ in reference coun-
tries. Similarly, changes in general education have led to a repositioning
of initial VET within young people’s study programmes because of their
increasing access to higher education. Last of all, European education
and training policies, which promote LLL through the open method
of coordination (OMC), have created a context favourable to hybridisa-
tion (see Verdier, 2008). The OMC structures the flow of ideas and ‘good
practices’ and encourages member states to undertake initiatives aimed

at attaining the objectives set in accordance with the so-called Lisbon
strategy (2000) and then the Europe 2020 strategy (see European Com-
mission, 2011). Over the past decade, LLL has gradually established itself
across Europe as an essential referent in political discourse on education,
skills and competences.
In order to explain this growing hybridisation in the context of the
reforms of national systems, it is necessary to construct ideal types
(Max Weber, 1965). In order to introduce a semblance of order into a
societal diversity that is considerably more complex, the ideal types pro-
posed here attach prime importance to the action targeted by the social
and political actors. Each LLL training regime is identified on the basis
mainly of the principles of justice informing the institutions respon-
sible for its regulation (see section ‘ “De-commodified” LLL regimes’).
Each national situation results from a compromise between several pos-
sible regimes, even if one of these ideal types may predominate (see the
section ‘LLL market regimes – but which market?’).
Five public policy regimes of lifelong learning
These regimes are built around the responses to a series of political

issues, with a minimum level of coherence established between the
various choices by virtue of the institutional complementarities and sol-
idarities between actors which these regimes generate. They have been
constituted on the basis of socio-historical analyses of the national sys-
tems in order to bring out the political principles which, at a given phase
of their history, have spurred a reorganisation of education and training
for young people. Drawing once again on Weber, these ideal types are
reconstructions of social and political trajectories in the area of educa-
tion and training; they provide a schematic description of reality that
makes it possible to move beyond the tension between the diversity of
particular historical occurrences and the general nature of these national
evolutions. This list of policy regimes draws on two kinds of typologies
of national systems which are ‘stylised’ in order to capture their more
general sense.
– One relates to national welfare states (see in particular Esping-

Andersen, 1990), which is all the more justified here because labour
market entry policies are becoming a new pillar of social protec-
tion (Esping-Andersen and Palier, 2008) and LLL is in some ways an
extension of this.
Eric Verdier 73
– The other relates to national systems of initial education and train-

ing or higher education and research (see Aventur and Möbus, 1998;
Buechtemann and Verdier, 1998; Maroy, 2000; Green et al., 2006).
Public policy regimes as a combination of political

principles, actors’ logics, rules and instruments
Every LLL public policy regime entails more or less explicit responses to
highly political questions:
1. Which principles of justice and efficiency are to be applied in the

area of education and training?
2. Is the individual as conceived within the regime integrated into
a professional community, line organisation, network or social
citizenship?
3. Who assumes responsibility for qualification and employment-
related risks (unemployment, precariousness, obsolescence of skills,
etc.): the individual and/or social insurance /the state?
4. What kind of governance is needed: which configuration of pri-
vate and public players (degree of decentralisation, role of private
training establishments and companies)?
5. How should education and training be organised: continuity or
separation between initial and continuing training, what place for
vocational training in the initial curriculum?
6. Which conception of knowledge is to be applied: primacy of aca-
demic knowledge, work-related knowledge, absence of distinctions
between the different kinds?
7. Which regulatory institutions are put in place: provision of infor-
mation, schemes compensating for initial inequalities, rules for
selection of individuals, negotiated collective agreements?
8. What are the means of access to both initial education and training
(e.g. accessibility of tertiary education) and continuing training (cf.
the question of guidance)?
9. What is the nature of the training to be provided for young people:
vocational, general, organised by levels?
10. Who funds the different kinds of education and training (public
authorities, companies, families or individuals)?
The principles of justice and efficiency underlie the legitimacy of the

rules in this area, all of which depend on various forms of justification
(Boltanski and Thévenot, 2006): vocation (Beruf ), academic meritocracy,
solidarity, utility of the service, transparency of the quality–price ratio.

These conventions, which are used by private and public players,
are associated with different cognitive, financial and organisational
resources that take concrete form as instruments, rules, social tech-
nologies and so on, all of which help to stabilise the regulations at
work. Notwithstanding their individual specificities, these five regimes
can be grouped into two categories in terms of linkages with market
rules. Three of them are more or less based on ‘de-commodification’
(Esping-Andersen, 1990).1 They are described as ‘academic’, ‘corporatist’
and ‘universal’ respectively. The other two are market-oriented but have
different conventions or principles and are described as ‘pure market
competition’ and ‘organised market’ regimes respectively.
‘De-commodified’ LLL regimes
Above and beyond compulsory schooling, these three regimes make ini-
tial education and training a central focus of the collective action which
is decisive for both individual and collective destinies, but they draw
on quite dissimilar rules: selection, vocation and cohesion. They also
decommodify education and training to varying degrees (see Table 3.1).
The academic regime is built around two processes: first of all, school-
based competition between individuals, the fairness of which must
be guaranteed by a public actor invested with incontestable political
legitimacy. It also relies on an objectified criterion, academic perfor-
mance, which, in principle, is not sensitive to local market influences
(Duru-Bellat, 1992). Qualifications therefore identify different levels of
general studies and constitute rules which are above all internal to the
educational system (Méhaut, 1997). The independence of such rules is
to be preserved from influences that might compromise the integrity
of the merit principle. It is up to individuals to enhance the value of
these ability indicators in the labour market, where legitimate posi-
tions in hierarchical organisations are established through the degree
of academic prestige they have acquired for themselves. Within this
framework, continuing training is above all a means of adapting to tech-
nical and organisational changes. The ‘training enterprise’ assumes most
of the funding during working hours. The main risk is an aggravation
of academic and, later, social inequalities that are all the more serious
because academic competition has lasting effects on the labour market.
The corporatist regime relies on occupational identities sustained by
individual commitment to a vocation, as well as on highly involved
social actors (Vinokur, 1995). The latter are in fact called upon to make
75
Table 3.1 ‘Decommodified’ LLL regimes.
Corporatist Academic Universal
Justice principle Access to an School-based Compensation

occupational or merit system for initial
craft community (‘rank’ and inequalities
(vocation) selection) (‘solidarity’ and
social inclusion)
Conception of Overall mastery Education levels Reconciliation of
skills in initial of a trade or basic knowledge
education and occupation and practical
training skills
Certification Recognised Certification by National diploma
qualification an academic
authority
Nature of Contents Subject-based Interaction
programme determined by standards between
negotiation different kinds of
knowledge
Area of Occupational Internal and ‘Multi-
recognition labour market hierarchical transitional’
market labour market
Key actor in Company Academic Community of
initial education education partners
and training institutions
VOTEC objective Occupational Indicators of Social citizenship
rules abilities
Main risk Stigmatisation of Sharp inequalities Increased
those without in schooling collective costs
qualifications
Key actor in Social partners at Educational Public authorities
institutional industry level institution
regulation
Continuing Higher levels of Short term Social autonomy
training occupational adaptation of
objectives mastery skills
Political Collective Companies and National
responsibility for agreements at public bodies tripartism
employability occupational
branch level
Funding of Vocational Companies and Public agencies
continuing training schools employers’ and mutual funds
training and individuals groups
a political commitment to the creation of vocational training quali-

fications so that they may enjoy high social esteem and become the
rules governing occupational labour markets (Eyraud et al., 1990). This
configuration generates social and economic legitimacy for skills, based
on mastery of a ‘trade’ or occupation and made up of a set of indis-
sociable capabilities. Contrary to what prevails with meritocracy, there
is no hierarchy of education/training levels that can be used to stan-
dardise the training programmes associated with a single occupational
field. Within this framework, work organisation is based on the pro-
fessional legitimacy associated with growing mastery of the ‘trade’, as
might be attested, in the course of working life, by occupational branch
certifications. The provision of an esteemed vocational route does not
exempt the individuals concerned from making efforts to demonstrate
their desire for promotion and peer recognition.
The universal regime is based on a principle of solidarity that attempts
to compensate at an early stage for inequalities stemming from a dis-
advantaged social environment (Wiborg, 2009). Pre-school education is
an established element of LLL. Thus early, irreversible tracking towards a
vocational pathway is excluded; conversely, individualisation based on
the construction of pathways adapted to the diversity of expectations
and career plans not only avoids repeating school years but also, and
above all, limits the importance attached to academic knowledge. Such
a regime is aimed at ensuring a linkage with other kinds of knowledge
and notably the capacity to translate what is learnt into competences:
the ability to work in groups or set up economic and social projects
of collective local interest. In this spirit, the opening up of training
establishments to their environment also requires the involvement of
different partners (non-profit general interest organisations, trade and
industry, users’ representatives, etc.) in the regulation of the estab-
lishment itself. In the area of continuing training, it is important for
collective actors to organise a second chance for returnees in the hope
of reducing the original inequalities that were inherited or engendered
by the earlier sequences of initial or continuing training. In addition,
adults in continuing training should be ‘actors’ in their own training; as
Merle (2006) puts it, this means restoring favourable conditions to the
‘desire to learn’ and giving substance to the exercise of a subjective right
which, ‘as a legal category, is inseparable from individual empowerment’
(Maggi-Germain, 2006). One of the risks of this form of collective action,
however, lies in the excessive costs of generous compensatory schemes.
Evaluation is thus an essential regulatory instrument that can be used
to justify the allocation of resources to a given person or group.
Eric Verdier 77
LLL market regimes – but which market?
Both of these regimes, the pure market competition regime and that
‘organised’ around networks linking public and private actors, approach
training in utilitarian terms (see Table 3.2). In the former, a price marks
the successful matching of service supply and demand in the labour
market; in the latter, work can be an alternative to training, provided
the individual concerned can resolve a double trade-off. What skills
does on-the-job training, in a work situation, offer? And how is time
to be divided between leisure, training and work, depending on the util-
ity/disutility of the various activities? The answers to these questions
Table 3.2 Market LLL regimes.
Market competition Organised market
Principle of justice Utility of services Fair price for quality

provided
VOTEC objective Human capital Social capital
Conception of skills in Meeting a demand Portfolio of operational
initial education and (possibly on the job) skills
training
Certification Level of remuneration Attestation of skills
(matching)
Nature of programme N/A (not available or Quality procedure
inadequate)
Area of recognition Immediate transaction External ‘organised’
(spot market) markets
Key actor in initial Individuals as consumers ‘Guided’ individuals
training
Main risk of failure Under-investment in Inefficient incentives
training
Key actor in Visible hand Public regulatory and
institutional regulation accreditation agencies
Conception of Utility of service Diversified skills
continuing training provision portfolio
Political responsibility Individuals ‘Active’ individuals and
and employability agencies
Funding Direct payment or Training vouchers,
loans for individuals individual training
(companies for specific accounts
training)
vary considerably from one regime to the other depending on the

distribution of responsibilities between private and public choices.
The market competition regime has to deal with the definition of edu-
cation and, to some extent at least, training as public goods. Given the
positive externalities attached to this resource, the state’s intervention
to make schooling compulsory is considered legitimate, but its dura-
tion should vary according to the market configurations involved: work
can be an alternative to school,2 even at an early age, depending on
the job opportunities proposed and the quality of the services provided
by training establishments (see Bougroum and Ibourk, 2006, on the
Moroccan case). What is involved in the case of a market agreement,
then, is the regulation not of a pure initial education and training mar-
ket but rather of a quasi-market, whose operating rules, especially in
the financial domain, are intended to ensure the pre-eminence of two
principles: competition, notably between training establishments, and
the individual’s free choice between different training methods, ranging
from structured versions to the informal varieties of on-the-job train-
ing, as well as the compromise situation of apprenticeship, all of which
eliminates the need for the prior determination of initial vocational edu-
cation and training programmes. In continuing training, what comes
into play is the standard rule of the market with which service providers
and seekers have to comply. In the case of general training, the seek-
ers will be individuals; when specific skills are to be acquired, the main
responsibility will fall on companies (Becker, 1964).
The organised market regime gives the market or quasi-market (Bartlett
and Legrand, 1993) a predominant role in regulating the access, use and
further development of training, whether initial or continuing. How-
ever, its workings are subject to a body of rules intended to guarantee
the transparency of quality–price relationships for all the protagonists
in order to control possible information distortions and arrive at the fair
price. In this case, we can speak of a market ‘organised’ by public inter-
vention, which then has to guarantee the reliability of the information
and the quality standards supporting the transactions taking place in the
market or quasi-market. The two cardinal principles of competition and
individual choice come into play but they are backed up by rules aimed
at eliminating, or at least reducing, selection bias and moral hazard,
which might compromise the transactions: the quality requirement for
the information system (notably based on certifications and outcomes
of training establishments) is the responsibility of the public authori-
ties (Chubb and Moe, 1990). Individuals are guided in these choices so
that they can make the best possible assessment of the consequences
Eric Verdier 79
(e.g. exiting initial education and training in favour of on-the-job learn-

ing, even if this means returning to structured programmes afterwards)
with the idea of making them aware of the potential risks to their
employability. Thus incentives are offered in the form of individual
training accounts, to which individuals themselves, public agencies or
employers can contribute (on this issue, see Giddens, 2001).
This approach, which has been extremely influential in the English-
speaking countries since the 1980s (Whitty, 1998), involves establishing
a market for individual skills, the first advantage of which is the trans-
parency of the quality–price ratio. The external markets have to be
endowed with the appropriate institutions, for even if they may be free
of long-term commitments they nonetheless require standards. In addi-
tion, the organisation of this skills market must be flexible so that
individuals can validate experiences acquired in a wide variety of ways,
from on-the-job learning to programmes in training establishments.
These signals, along with the individual training accounts, enable indi-
viduals to move through networks of private and public organisations
which are linked together by the standards developed under the aegis
of the public authority. The individual user is thus called upon to be a
vehicle for the management and evaluation of initial and continuing
training policies.
From policy regimes to national models
A national LLL model is the result of a specific compromise between

different ideal-typical regimes supported by specific coalitions of pub-
lic and private actors. This societal arrangement may be more or less
sustainable depending on the endogenous social dynamics and capac-
ity to cope with external changes. This comparative presentation is
based (1) on a set of quantitative indicators reflecting the models’ vari-
ous features and (2) on a comprehensive approach which aims to take
into account the institutional settings and the commitments of a social
coalition of actors.3
Sweden: A good record in LLL but with increasing

difficulties
In the early 2000s, Sweden accumulated a set of features corresponding
to a LLL regime with a dominant universal orientation: high rates of
access to the various levels of education; modes of funding and organ-
isation reflecting a political determination to channel the effects of
the market, academic selection and corporatist closure; student perfor-

mances generally less uneven than in comparable countries. All of these
findings and practices are inseparable from the ‘individualised model
of integration’ which regulates the long common core of the Swedish
education system by combining ‘integration through common objec-
tives and differentiation of pathways’ (Mons, 2007, p. 119). At upper
secondary level, one of the basic objectives was to reduce as much
as possible the gulf separating vocational training from general educa-
tion in order to offer all pupils access to higher education. In terms of
access rates, the Swedish results are among the best in Europe: (1) they
are high for pre-primary school, which is effective in reducing edu-
cational inequalities due to social background and gender disparities;
(2) like Denmark, Sweden is characterised by a low rate of early school
leavers; (3) it also has a fairly good level of access to higher educa-
tion, above the European average. Sweden is also characterised by the
lowest share of private expenditures on education and only a slight dif-
ference between individual expenditures on higher education and those
on primary schools, with the overall average significantly higher than
in Germany and France. Adult education is part of the public education
system: a very dense network of public centres devoted to adult educa-
tion (Abrahamsson, 1999) limits disparities by qualification, gender and
age. Public policies in training have been focused on less well-educated
populations: from 1997 to 2002, a vast public programme – called
‘Knowledge Lift’ – aimed to bring participants up to upper secondary
school level, thereby enabling them to go on to higher education as well.
In all, some 230 000 people, equivalent to 75 per cent of the young peo-
ple in upper secondary education, took part in the programme, although
its effectiveness was contested by certain evaluations (Stenberg, 2003).
If the unemployment rate for those without qualifications is fairly high,
the differences in earned income by qualificational level were (and are
still) considerably lower than in Europe as a whole, even though the
employment rate of university graduates is quite high.
However, the general pattern appears to be undergoing considerable
change. The average PISA performance (reading/literacy) has declined
significantly (by some 20 points, compared to 3 for the Organisation for
Economic Co-operation and Development (OECD) as a whole), in par-
ticular because of a sharp drop in the outcomes of the weakest students.
The impact of social inequalities on academic performances is now
greater than the European average, reflecting a decline in the effective-
ness of the individualised integration model, which was already having
Eric Verdier 81
difficulties in integrating young people from immigrant backgrounds

(OECD, 2007). Indeed, it might be asked whether the introduction of
individual vouchers,4 the liberalisation of the regulations governing
school catchment areas and reduced support for students with problems
in large urban centres such as Stockholm since the arrival of a conserva-
tive government coalition in 2007 have accentuated these trends, which
are reflected in a pronounced rise in the number of school leavers with-
out qualifications in the labour market. In any case, the gap between the
performance of the Swedish and Finnish systems is now quite marked
(OECD, 2011). The reforms undertaken by the Swedish government in
2011 reflect a new selective, corporatist agenda, in which access to sec-
ondary schooling becomes dependent on prior achievements, general
and vocational streams are once again differentiated (with the latter
not necessarily giving access to higher education) and an apprenticeship
training programme has been set up.
Germany: Continuity of the apprenticeship model

but a corporatist mode of regulation subject to
increasing limitations
Until the end of the 1990s, the predominant corporatist regime of
industry-level ‘private governments’ (Hilbert et al., 1990) linked appren-
ticeship to occupational certifications during the working life through
the intermediary of the Berufsakademie (‘vocational academy’). From
the outset, the system has been based on a market for apprenticeship
places that is itself highly regulated by quality standards laid down
at federal level in tripartite negotiations between trade unions, busi-
ness organisations and government (Koch, 1997). The predominance of
this corporatist mode of regulation helps to explain (1) the low rate of
youth unemployment maintained by the ‘regulated integration’ in the
labour market (Garonna and Ryan, 1991) and (2) the stagnating share
of university graduates among the younger generation at a time when
shares were increasing very rapidly in many other European countries,
such as Denmark, where apprenticeship is still predominant: access to
higher education remains very limited for young people following their
apprenticeships.
Since the mid-1980s, the German-style ‘primacy of occupational
know-how’ has had to take account of the increasing need for skills to be
adaptable. To this end, the number of qualifications has been consider-
ably reduced, curricula are revised more regularly and new qualifications
created more quickly (Bosch, 2009). However, while it remains attractive

to young people, the dual system as a whole has faced increasing
pressures from:
– a structural shortage of apprenticeship places, which makes entry

into vocational training ever more selective, at the expense of the
36 per cent of young people exiting at best with a certificate from the
shortest lower secondary programme (Hauptschule, ISCED Level 2), or
without any qualification at all; the issue is all the more important in
social terms given the fact that in Germany, more than elsewhere, a
lack of qualifications increases the risk of unemployment;
– the erosion of promotion prospects for holders of qualifications
acquired in the course of their careers following apprenticeship train-
ing, which could weaken the strong occupational identity based on
both initial and continuing vocational training (see Maurice et al.,
1986). The declining access to recognise further qualifications (e.g.
Meister and Techniker federal certifications) explains the current low
rate of access to education during working life, now at the same level
as in France, in contrast to the equivalent situations in the 1980s (see
Géhin and Méhaut, 1993). These trends have been accentuated by
the low outcomes of young Germans in the OECD’s first PISA sur-
vey in 2000 which tended to undermine the validity of the early
tracking of most young people into vocational education, especially
given the low share of students from immigrant families obtaining
apprenticeships. This societal situation results from a model based on
separation, which contrasts sharply with more integrative models of
secondary education (Mons, 2007). In addition, the relative lack of
access to pre-elementary school (in 2000, it was still 4 points below
the EU average and 25 points less than the score in France) runs
counter to the European objectives of ‘facilitating access for all to
education and vocational training’ and reconciling working life and
family life (Salzbrunn, 2007).
The ‘PISA Shock’, which generated considerable debate among politi-

cians, led to the introduction of both federal and regional retraining
programmes for young people excluded from the dual system (over a
third of young people entering vocational training in 2009). As for the
continuing training of the unemployed, recent reforms (Hartz IV) are
clearly based on an ‘organised market’ mode of regulation that is
intended to encourage a quick return to employment, regardless of its
quality. The upgrading training provided to the unemployed by federal
Eric Verdier 83
agencies has been sharply reduced (Bosch and Weinkopf, 2008). Admit-
tedly, this combination of schemes has led to a fall in the share of young
people exiting without a qualification, while the number of young chil-
dren enrolled in nursery school has increased considerably. For all that,
however, the school system as a whole basically remains unchanged.
Germany still spends fairly little on its primary schools; pupil–teacher
ratios are still high and the number of teaching hours lower than else-
where, while considerable resources have been made available for higher
education. In addition, even if the influence of social background on
the learning achievements measured by the PISA test has diminished
slightly, it remains higher than the average for OECD member states.
In this context, income disparities between higher education graduates
and those without qualifications showed a significant increase during
the first decade of this century.
Denmark: A virtuous compromise committed

to permanent education?
The Danish model constitutes an original compromise between a his-
torically corporatist convention (especially in initial education and
training) and a universal one (notably in continuing education and
training). Apprenticeship predominates within upper secondary educa-
tion, but unlike the German case, it is not associated with high rates
of early school leaving, very unequal outcomes in the PISA evalua-
tions or obstacles to entering higher education (in terms of the share
of higher-education graduates, the Danish situation is even better than
that of Sweden). As in Sweden, moreover, the share of private funding,
in secondary and higher education alike, is quite limited, which reflects
the universal principles strongly defended by the state. This position is
demonstrated by the ongoing policy of significant public investments
in education, especially at primary and secondary levels (Denmark is
characterised by an increase in public spending which has brought the
share of GDP allocated to education to more than 7%); in particular, this
commitment has led to an even greater reduction in the share of early
school leavers without qualifications. The same applies to continuing
training, where funding is essentially provided by the public authori-
ties. This approach produces high rates of access to continuing training,
whether leading to certification or directly job-related. It also turns out
to be considerably more egalitarian than the approaches adopted in
other OECD countries (cf. Denmark’s income inequalities by qualifica-
tional level, which are particularly low and which, contrary to those
found in many other countries, have not increased since the begin-
ning of this century), even if it remains relatively impossible to undo
the long-term effects of the corporatist regulation of initial training:
‘The number of unemployed who are the farthest from employment
(those without vocational training, ethnic minorities) and are depen-
dent on public assistance has scarcely changed since 1994’ (Lefresne,
2005). Nonetheless, Denmark remains an unavoidable reference point
for societal benchmarking on questions of training and unemployment.
The specific concept of folkeoplysning (literally ‘popular enlightenment’
or ‘popular education’) is not unrelated to this situation, insofar as its
combination of ‘personal development, sense of community, education,
vocational training and individual responsibility within the democratic
process’ (Meilland, 2006) generates a complex alchemy close to the
demands of the 1970s notion of ‘permanent education’.
It has given rise to a group of institutions intended to encourage indi-
viduals particularly young people, to exercise free choice so as to allow
them to experiment in the course of their studies and working lives alike
(van de Velde, 2008). Thus as soon as they are of legal age, young Danish
people benefit from a state-organised system that guarantees them broad
financial independence, whether they are students or in employment,
through a combination of allowances and earned income. In 2005, more
than 55 per cent of the 15–24 age group combined work and studies,
with many university students punctuating their studies with periods
of work experience. Nonetheless, while this scheme offers useful guar-
antees in times of crisis, the fact remains that initial training pathways
remain relatively inaccessible to young people from immigrant fami-
lies seeking apprenticeships. The discrimination encountered by young
people of non-Community origin in terms of access to and participation
in apprenticeship (Wiborg and Cort, 2010)5 constitutes, moreover, one
of the recurring problems of the corporatist convention prevailing in
Denmark: ‘It is almost entirely up to the students to find a firm willing
to take them on as apprentices’ (OECD, 2009, pp. 6–7). In addition,
although the period of transition between youth and adult life does
serve to stabilise career paths, it turns out to be quite expensive because
of the length of time it takes young people to complete their higher
education.
France: ‘Integrated vocationalism’ or a combination of academic

and corporatist regimes
Under the combined pressure of youth unemployment and a demand
for more democracy in education, France’s educational policy has, in
Eric Verdier 85
the course of 30 years of successive reforms, developed vocational cer-

tifications integrated into the hierarchy of general education levels
while promoting alternating training in the form of apprenticeships
or school-based programmes. In this respect, the creation of the voca-
tional baccalauréat marked a major turning point: for the first time, a
vocational high-school curriculum could not dispense with work place-
ments. ‘Integrated vocationalism’ was thus developed as a compromise
between, on the one hand, a neo-corporatist (or vocational) conven-
tion and, on the other, the historically dominant academic convention.
This compromise manifests itself in the spread of apprenticeships to all
vocationally oriented degree programmes, including engineering and
university master’s programmes, as evidenced in the Paris region by
the fact that more than half of apprentices are preparing for a higher
education qualification.
Indeed, the vocational stream is still subject to the standards of gen-
eral education, and notably the hierarchy of training levels. The fact
remains that if the education system is to be seen as just in the eyes of its
protagonists, it now has to guarantee each young person that he or she
will leave initial training with a ‘recognised qualification’ (Educational
System Orientation Law of July, 1989). The definition of this ‘qualifica-
tion’ is vague, however, given that it can cover certification by means
of a formal qualification, recognition within the classification grid of a
collective agreement or even the mere existence of opportunities within
the occupation targeted at the end of training. These uncertainties are
all the more difficult for young people because, to a greater extent than
in other European countries (OECD, 2009c), their school-to-work tran-
sition remains marked by the chronic inability of the schemes to curb
youth unemployment, especially for those with low educational levels
(17% of exits without qualifications).
The academic regime (‘everything is played out before age 25’, after
which companies adapt individual skills to their needs, with the active
support of the public authorities) is still prominent. First, the ‘French
meritocracy’ is based on rigorous selection throughout the school career.
The upper secondary cycle is structured around a three-way segmen-
tation (vocational/technological/general) and although there has been
undeniable progress, the democratisation of the education system is
so ambiguous that some authors speak of ‘segregative democratisation’
(Duru-Bellat and Kieffer, 2001). Access to higher education has signifi-
cantly expanded and the proportion of graduates in the 25–34 age group
is as high as in Sweden and Denmark, but this is due to the development
of short vocational degree programmes (France remains behind many
European partners in terms of the share of graduates at master’s or PhD
level) and there has been no change in the predominance of the highly
selective grandes écoles, which still shape the ‘scholar gentry’ (analysed
by Bourdieu). Indeed, the influence of social origin on the highest qual-
ification obtained is greater in France than in the United States (Meuret,
2007; see also the PISA results). School enrolment rates have shown no
increase over the past 15 years (they have even declined slightly), which
reflects the breakdown of the compromise between the academic and
corporatist regimes, while France is one of the OECD countries that has
spent least on compulsory schooling over the past decade.
Since 1971, training has been run on a tripartite basis (government,
trade unions and employers), which has made short courses more acces-
sible to the most skilled employees. Thus the role of business in the
organisation of training is predominant but the rate of access to train-
ing during the working life is less than half the European average. The
industry-wide agreement on ‘lifelong learning’ (2003), which was writ-
ten into law in May 2004, has not really changed the situation: it creates
an individual right to 20 hours of training per year but the transferability
of this right is limited and the number of hours in practice limits indi-
viduals to short courses designed to upgrade existing skills, although
it is possible to carry over the annual entitlement for up to six years,
enabling employees to accrue a total of 120 hours of training.
Great Britain: Development of an ‘organised market’

and universal principles
In the context of a regime that was primarily academic but also included
a corporatist apprenticeship system, the Thatcher government’s reforms
introduced a quasi-market logic (the training costs borne by households
were very high, double those in the United States), but one that was
organised from the outset around standards for certification. National
Vocational Qualifications (NVQs) were the cornerstone, in terms of
both initial and continuing training, while the content of training pro-
grammes was determined by training providers themselves. From the
year 2000 onwards, moreover, the Labour government invested in the
development of initial education and training as part of a redistributive
effort focused on individuals without qualifications that gave an increas-
ing role to the ‘organised market’. The government’s primary objective
was to reduce the share of young people who, at the end of compul-
sory schooling, found themselves unemployed, inactive and outside of
any training programme to the point of plunging into social exclusion
(Ryan, 2001). While the main activation programme – the New Deal
Eric Verdier 87
for Young People – privileged the target of a ‘first employment’, vari-

ous schemes such as the 14–19 Strategy (that aimed to find a suitable
study programme for every young person after compulsory schooling)
to some extent reflected a universal convention.6 Thus by combin-
ing loan facilities and public funding, training policy greatly improved
access to qualifications in the course of the working life (to an extent
clearly above the European average). More generally, participation in
continuing training reached relatively high levels (27% of employees
in 2003 compared to an OECD average of 18%), particularly because of
a new system recognising formal and informal learning (Tessaring and
Wannan, 2004).
Thus evaluations of the British trajectory must be regarded as some-
what mixed, particularly with regard to youth training, ‘ . . . while the
main activation programme for young people in the United Kingdom –
the New Deal for Young People – has helped many young people to
return to work, sustainable employment outcomes have proved difficult
to achieve’ (OECD, 2009b, p. 2). Indeed, one in five young people find-
ing work through the programme held a job lasting less than 13 weeks.
Under the coalition government, the role of charitable foundations and
private providers in offering services to youngsters not in education or
training (NEET) can be expected to increase. In the area of initial edu-
cation, the academic regime remains predominant, albeit with a strong
polarisation between significant numbers of early school-leavers and the
famous ‘Oxbridge’ elite (cf. the income inequalities between higher edu-
cation graduates and individuals with no qualifications). Nonetheless,
an active public policy has significantly increased funding to ensure a
basic level of education for all pupils, which seems to be attested by
the latest PISA outcomes (in terms of low achievers in reading) and the
decrease in early school leavers. Despite the development of appren-
ticeship, the only clear route is the one leading to A-levels and higher
education (taken by around 40% of a given cohort), while the majority
of young people must confront the profusion of qualifications proposed,
and for too many of them, this results in poor choices leading to dead
ends (Steedman, 2010).
Conclusion
Over the last 25 years, national VET systems in Europe have been sub-
ject to increasing hybridisation. However, among the countries reviewed
here, the extent and political significance of this phenomenon vary
considerably. Denmark has seen the introduction of an unprecedented
compromise between the universal and corporatist regimes. This com-

promise has been further consolidated over the past decade by means
of considerable public investment. Income disparities by qualificational
level remain relatively slight; the youth unemployment rate is lower
than the European average; there are few NEETs and employee access
to training courses leading to qualifications during the initial phase
of the working life is more open than in most other European coun-
tries. In France, the academic regime still predominates, but a series of
reforms, inspired by Germany in particular, have established a unique
regime which might be described as ‘integrated vocationalist’, with fea-
tures that have hardly evolved over the past 15 years, amid growing
tensions. In spite of this investment in VET for young people, this popu-
lation still faces a high risk of unemployment, while the ‘second chance’
offered to workers in the form of training programmes leading to quali-
fications remains very limited. In the United Kingdom, the Thatcher era
saw the establishment of an organised market regime targeted mainly at
individuals in transition (young people and the unemployed), while the
past decade has seen significant public–private investments in compul-
sory training along the lines of the universal regime, as shown by the
particularly sharp increase in expenditures for primary and secondary
education, but even so it is proving difficult to combat the pronounced
dualism of the education system. Enrolment rates for upper secondary
remain far below the European average, while the percentage of NEETS,
as well as youth unemployment, is still quite high, despite efforts to pro-
fessionalise their training. Sweden stands apart by virtue of the increasing
personalisation of educational pathways, which has led to growing con-
vergence of VET and general programmes. While this trend reflects the
growing universalism of the education system, its effectiveness seems
to have diminished, as demonstrated by the considerable influence of
social background on PISA performance and the considerable increase
in the proportion of poor readers according to the same tests. Some
observers would attribute these trends to the impact of neo-liberal
reforms introduced by the neo-conservative governments of the 1990s
and 2000s. At present, even though the universal regime remains pre-
dominant, the Swedish societal system is facing reforms on a scale that
is difficult to measure but which appear to reassert initial VET. Germany
has undergone fewer transformations: ultimately, the corporatist regime
linked to a market for apprenticeship places still stands out for its youth
labour-market entry, which is more successful than elsewhere in Europe.
It has shown little evolution over the past 25 years, as indicated by the
stability of the proportion of higher education graduates. Nonetheless,
Eric Verdier 89
the erosion of this regime is exposing it to growing pressures, notably

because of a lack of places in vocational programmes leading to qual-
ifications, which penalises the young people who have been the least
successful during compulsory schooling; schemes for entry into employ-
ment and continuing training do not seem able to provide adequate
solutions.
Ultimately, if there is convergence, it is probably to be found in the
extent of the governance challenges facing these societal regimes of LLL.
On the other hand, the actual form taken by the various ideal-type
regimes remains quite country-specific. Two factors may contribute to
this situation: the prevailing regimes differ sharply from one national
system to another, which means that the same reform, of neoliberal
inspiration for example, will take on a different meaning depending on
the societal context involved. In Sweden, the reforms aimed at intro-
ducing quasi-market rules have had relatively little effect so far on what
is still the strongly universal nature of the system in place, whereas in
the United Kingdom, within a national context historically marked by a
highly selective educational system, the neo-liberal reforms introduced
during the Thatcher era had increased inequalities so radically that the
Labour government was led, to a certain degree at least, to develop
instruments of a universal nature.
Notes
1. The concept of decommodification comes from the idea that, in
a market economy, individuals (and their labour) are commodified.
Decommodification refers to activities and efforts (generally by the gov-
ernment) that reduce individuals’ reliance on the market and their labour
for their well-being. In general, unemployment and sickness insurances and
pensions are used to measure a welfare state’s degree of decommodification
(Esping-Andersen, 1990).
2. Steedman (2010) explains that in the British case, ‘the historical legacy of
the early leaving/early employment tradition in White British culture can
still be seen in young people’s and their parents’ attitudes to staying-on in
full-time education after 16’ (p. 95).
3. The characterisation of each national LLL regime and its evolution relative
to the five ideal types is based, firstly, on a body of information and analy-
ses concerning the organisation and impact of education and training and,
secondly, on a series of indicators. The latter are intended to describe several
dimensions:
– the development of education and training: the rates of school enrolment

at different levels and the ability to maintain enrolment beyond the
compulsory level; the importance of continuing training, whether job
related or leading to a qualification;
– its forms of organisation (general training and apprenticeship within

upper secondary education; the share of pupils repeating a year and extent
of selection (early school-leaving without qualifications) – and funding
(notably for primary schooling, the share of private expenditure);
– its performances in terms of learning achievements (PISA), the proportion
of university graduates and labour-market entry (youth unemployment
and NEETs);
inequalities in terms of earned incomes and unemployment by qualifica-

tion levels, given that they reflect one part of the training-employment link.
For reasons of space, the corresponding tables can not be presented in this
chapter. For more details, see E. Verdier ‘European Lifelong Learning Strategy
and diversity of national devices: an interpretation in terms of public policy
Regimes’. The European Consortium for Political Research, General Conference
in Potsdam, 9–12th September 2009, Germany.
http://halshs.archives-ouvertes.fr/docs/00/43/65/45/PDF/Potsdam-Verdier.
pdf, date accessed 20th December 2009. For a discussion of OECD indicators
on education, see Chapter 1 in this book.
4. ‘ . . . It is not only housing segregation that causes the increase of vocational
education, but parental school choice has exacerbated this trend’ (Wiborg,
2010, p. 19).
5. ‘Another problem is the high dropout rate from VET programmes, especially
among ethnic minorities’ (p. 104).
6. See also the Education Maintenance Allowance (financial assistance) intro-
duced in 2004 and the September guarantee (guaranteed placement in an
educational establishment after age 16) launched in 2007.
References
Abrahamsson, K. (1999) Vocational Education and Training in Sweden. Thessaloniki:
Cedefop.
Aventur, F. and M. Möbus (1998) Formation professionnelle initiale et continue en
Europe. Paris: Magnard et Vuibert.
Bartlett, W. and J. Legrand (1993) ‘The Theory of Quasi-Markets’ in Bartlett,
W. and J. Legrand (eds) Quasi-Markets and Social Policy. Houndmills: MacMillan
Press, pp. 13–34.
Becker, G. (1964) Human Capital. A Theoretical and Empirical Analysis, with Special
Reference to Education. New York: Columbia University Press.
Boltanski, L. and L. Thévenot (2006) On justification: Economies of Worth.
Princeton: Princeton University Press.
Bosch, G. and C. Weinkopf (2008) Low-Wage Work in Germany. New York: Russell
Sage.
Bosch, G. and J. Charest (2009) Vocational Training. International Perspectives.
Bougroum, M. and O. Ibourk (2006) ‘Les déterminants de la demande d’éducation
chez les enfants travaillant dans le secteur de l’artisanat à Marrakech: anal-
yse micro-économétrique’ in Mourji, F., B. Decaluwé and P. Plane (eds) Le
développement face à la pauvreté. Paris: Economica, pp. 223–246.
Eric Verdier 91
Buechtemann, C. and E. Verdier (1998) ‘Education and Training Regimes: macro-

institutional Evidence’. Revue d’économie politique, CVIII, pp. 292–319.
Chubb, J.E. and T.M. Moe (1990) Politics, Markets and America’Schools.
Washington: Brookings Institutions.
Culpepper, P. (2003) Creating cooperation – How States Develop Human Capital in
Europe. Ithaca: Cornell University Press.
Duru-Bellat, M. (1992) ‘Evaluer les trajectoires sociales à l’aune de la méritocratie’.
Savoir, III, pp. 453–468.
Duru-Bellat, M. and A. Kieffer (2001) ‘La démocratisation de l’enseignement
en France: polémiques autour d’une question d’actualité’. Population, LV,
pp. 51–80.
Esping-Andersen, G. (1990) The Three Worlds of Welfare Capitalism. Princeton:
Esping-Andersen, G. and B. Palier (2008) Trois leçons sur l’Etat Providence: Coll. La
République des Idées. Paris: Seuil.
Estevez-Abe, M., T. Iversen T. and D. Soskice (2001) ‘Social Protection and
the Formation of Skills: A Reinterpretation of the Welfare State’ in Hall,
P. and D. Soskice (eds) Varieties of Capitalism – The Institutional Foundations of
Comparative Advantage. Oxford: Oxford University Press, pp. 145–183.
European Commission (2011) Analysis of the implementation of the strategic
framework for European cooperation in education and training – Country analy-
sis. Commission Staff Working Document, Accompanying document to the
Communication from the Commission to the European Parliament, the Coun-
cil, the European economic and social Committee and the Committee of
Regions. Brussels. http://ec.europa.eu/education/lifelong-learning-policy/doc/
joint11/wp2_en.pdf, date accessed 10 September 2012.
Eyraud, F., D. Marsden and J-J. Silvestre (1990) ‘Occupational and Internal Labour
Markets in Great-Britain and France’. International Labour Review, CXXIX,
pp. 501–517.
Garonna, P. and P. Ryan (1991) ‘The Regulation and Deregulation of Youth Eco-
nomic Activity’ in Garonna, P., P. Ryan and R. Edwards (eds) The Problem of
Youth: The Regulation of Youth Employment and Training in Advanced Economies.
London: MacMillan, pp. 35–81.
Géhin, J-P. and P. Méhaut (1993) Apprentissage ou Formation Continue? Stratégies
éducatives en Allemagne et en France. Paris: L’Harmattan.
Giddens, A. (2001) The Global Third Way Debate. Cambridge: Polity Press.
Giraud, O. (2004) ‘La Comparaison Comme Opération de Réduction de la
Complexité. Le cas de la formation professionnelle’. Revue internationale de
politique comparée XI, pp. 349–368.
Green, A., J. Preston and J.G. Janmaat (2006) Education, Equality and Social
Cohesion: A Comparative Analysis. Houndmills: Palgrave MacMillan.
Hall, P. and D. Soskice (2001) Varieties of Capitalism – The Institutional Foundations
of Comparative Advantage. Oxford: Oxford University Press.
Hilbert, J., H. Südmersen and H. Weber (1990) Berufsbildungspolitik – Geschichte,
Organisation, Neuordnung. Opladen: Verlag Leske und Bundrich.
Koch, R. (1997) ‘La rénovation des Formations professionnelles réglementées:
Instrument majeur de modernisation du système dual depuis les années 1970’
in Möbus M. and E. Verdier (eds) Les diplômes professionnels en Allemagne et en
France, conceptions et jeux d’acteurs. Paris: L’Harmattan, pp. 29–43.
Lefresne, F., (2005) ‘Les politiques d’emploi et la Transformation des Normes: une
Comparaison européenne’. Sociologie du travail, XLVII, pp. 405–420.
Maggi-Germain, N. (2006) ‘Herméneutique de deux idées’ in Morvan, Y. (ed.) La
formation tout au long de la vie. Nouvelles Questions, nouvelles Perspectives. Rennes:
PUR, pp. 23–38.
Maroy, C. (2000) ‘Une typologie des référentiels d’action publique en matière de
formation en Europe’. Recherches Sociologiques, XXXI, pp. 45–59.
Maurice, M., F. Sellier and J-J. Silvestre (1986) The Social Foundations of Industrial
Power: a Comparison of France and Germany. Cambridge: The MIT Press.
Méhaut, P. (1997) ‘Le diplôme, une norme multivalente?’ in Möbus, M. and
E. Verdier (eds) Les diplômes Professionnels en Allemagne et en France, Conceptions
et jeux d’acteurs. Paris: L’Harmattan, pp. 263–273.
Meilland, C. (2006) ‘Danemark: La Formation Continue des Adultes, Instrument
de politique de l’emploi et enjeu des négociations Collectives’. Chroniques
internationales de l’IRES, CI, pp. 23–29.
Merle, V. (2006) ‘La Formation tout au long de la vie, une Proposition pour les
sociétés démocratiques’ in Morvan, Y. (ed.) La formation tout au long de la vie.
Nouvelles Questions, nouvelles Perspectives. Rennes: PUR, pp. 23–38.
Meuret, D. (2007) Gouverner l’école, une comparaison France/Etats-Unis. Coll.
éducation et sociétés. Paris: PUF.
Mons, N. (2007) Les nouvelles politiques éducatives: La France fait-elle les bons
choix?.Coll. éducation et sociétés. Paris: PUF.
OECD (2007) ‘Briefing Note for Sweden’. Education at a Glance. Paris: OECD,
http://www.oecd.org/edu/39317576.pdf, date accessed 25 May 2013.
OECD (2009) Jobs for Youth/Des emplois pour les jeunes, Denmark: Summary in
English. Paris: OECD.
OECD (2009b) Jobs for Youth/Des emplois pour les jeunes, United Kingdom: Summary
in English. Paris: OECD.
OECD (2009c) Jobs for Youth/Des emplois pour les jeunes, France: Summary in English.
Paris: OECD.
OECD (2011) Education at a Glance 2011. Paris: OECD.
Ryan, P. (2001) ‘The School-to-Work Transition: A Cross-national Perspective’.
Journal of Economic Literature, XXXIX, pp. 34–92.
Salzbrunn, M. (2007) ‘Entre autonomie et insertion. Les grands dispositifs de la
politique de la jeunesse et de la famille en Allemagne’. Horizons Stratégiques,
April 2007, pp. 43–69.
Steedman, H. (2010) ‘England: Low Qualified People and the “NEET” ’ in
Steedman, H. and E. Verdier Les élèves sans Qualification: La France et les pays
de l’OCDE. Rapport pour le Haut Conseil de l’Education, http://www.hce.
education.fr, date accessed 2 February 2012.
Stenberg, A. (2003) ‘An Evaluation of the Adult education Initiative Rela-
tive Labor Market Training’. Doktorsavhandling, Ekonomiska studier, DCIX,
Nationalekonomiska institutionen. Umeå universitet, http://www.usbe.umu.
se/digitalAssets/11/11120_ues609.pdf, date accessed 25 June 2013.
Streeck, W. and K. Thelen (2005) ‘Introduction: Institutional Change in Advanced
Political Economies’ in Streeck, W. and K. Thelen (eds) Beyond Continuity:
Explorations in the Dynamics of Advanced Political Economies. New York: Oxford
University Press, pp. 229–254.
Eric Verdier 93
Tessaring, M. and J. Wannan (2004) Vocational Education and Training – Key to the
Future: Cedefop Synthesis of the Maastricht Study. Luxembourg: Office for Official
Publications of the European Communities.
Thelen, K. (2003) ‘How Institutions Evolve: Insights from Comparative Histori-
cal Analysis’ in Mahoney J. and D. Rueschemeyer (eds) Comparative-Historical
Analysis: Innovations in Theory and Method. Oxford: Oxford University Press,
pp. 208–240.
Van de Velde, C. (2008) Devenir adulte. Sociologie comparée de la jeunesse en Europe.
Paris: PUF.
Verdier, E. (2008) ‘L’usage des idées floues. L’Education et la formation tout au
long de la vie’ in Giraud O., P. Muller and P. Warin (eds) Politiques et démocratie:
Mélanges en l’honneur de Bruno Jobert. Coll. Recherches. Paris: La Découverte,
pp. 109–135.
Vinokur, A. (1995) ‘Réflexion sur l’économie politique du diplôme’. Formation
Emploi. LII, pp. 151–183.
Weber, M. (1965) Essais sur la théorie de la Science. Paris: Plon.
Whitty, G., S. Power and D. Halpin (1998) Devolution and Choice in Education:
The School, the State and the Market. Buckingam-Philadelphia: Open University
Press.
Wiborg, S. (2009) Education and Social Integration: Comprehensive Schooling in
Europe. New York: Palgrave MacMillan.
Wiborg, S. (2010) ‘Swedish Free Schools: Do they Work?’. LLAKES Research Paper
XVIII, http://www.llakes.org/wp-content/uploads/2010/09/Wiborg-online.pdf,
date accessed 11 September 2012.
Wiborg, S. and P. Cort (2010) ‘The Vocational Education and Training System
in Denmark’ in Bosch, G. and J. Charest (eds) Vocational Training: International
Perspectives. New York: Routledge, pp. 84–109.
4
Formal or Actual Convergence?
Three Cases of Hybridisation
Philippe Méhaut1
Three facts emerging from the debates on the convergence of education

and training systems that have unfolded in the preceding chapters are
sufficiently striking to be picked out and emphasised.
Firstly, the scope of the systems under investigation varies consid-
erably, since the notion of education and training system is defined
differently in each country and, in some cases, is dependent largely on
the available sources. Thus in the sociology of education (in France, but
elsewhere as well), the education system is defined first and foremost
as the compulsory and/or general education system (primary, lower sec-
ondary, upper secondary and possibly universities). Little attention is
paid to vocational training. The wide availability of international sta-
tistical sources, such as PISA, helps to reinforce this magnifying glass
effect: the object of comparison is pupils at age 15 and the correspond-
ing segment of the education systems in question (compulsory lower
secondary education).
This problem is exacerbated as soon as the focus of attention starts to
shift away from compulsory schooling (OECD, 2009). In the English
system, a large part of what is regarded as ‘lifelong learning’ (the
vocational pathways beyond compulsory education/training) would be
regarded as initial training in the French definition of the term. And
the ambiguities surrounding apprenticeship systems are well-known: in
Italy, they tend to be informal and outside the education and train-
ing system; in Denmark they are an integral part of the education
and training system while in countries such as France there is a con-
stant state of tension between apprenticeships as part of active labour
market policy and apprenticeships as part of education policy, a ten-
sion that manifests itself in a division of responsibilities between the
94
Philippe Méhaut 95
Ministry of Education and the Ministry of Labour. Consequently, certain

apparently homogeneous macro-level statistical indicators, such as the
enrolment rate at such and such an age, may conceal major structural
heterogeneities.
Finally, the notion of education system convergence itself warrants
debate. Above and beyond various, cultural, political and economic
hypotheses advanced to explain convergence (see Chapter 1 for a pre-
sentation), it is often reduced to a small number of major quantitative
indicators (enrolment rate, share of GDP) or to a few more qualitative
indicators (decentralisation, early tracking) whose formal convergence
is said to be sufficient to deduce actual convergence. However, some of
the studies in the present volume have revealed the limitations of these
indicators in describing actual convergence. They also demonstrated
the difficulty of capturing marginal situations in cluster analysis (e.g.
Chapters 1 and 2). Consequently, the hypotheses on system hybridisa-
tion advanced by Verdier in Chapter 3 have to be taken seriously. Such
hybridisation can be observed on two levels. The first is within a given
system, that is between the various levels of education within the same
country. Thus a country’s basic education system may fall within the
scope of one particular ‘regime’, while its upper secondary system is
governed by another regime and lifelong learning by yet another. The
second is between systems, with countries being subject to influence and
pressures exerted by economic factors, exchanges of practices, cultural
mimesis and international organisations such as the Organisation for
Economic Co-operation and Development (OECD) and the European
Union (EU), which together have given rise to some complex dynam-
ics. Regimes that were previously fairly ‘pure’ are becoming bastardised,
although the new mix cannot necessarily be described as ‘convergence’
(Bosch and Charest, 2009). This is all the more true since, as already
noted, the very notion of convergence warrants investigation. To take
just one example, drawn from biology, species convergence denotes the
propensity of different species to develop similar characteristics in simi-
lar environments. One example is that of the mole and the mole cricket,
both of which evolved under ground and developed shovel-like fore-
limbs enabling them to dig their tunnels. However, one is a mammal,
the other an insect. Their formal convergence does not mean actual
convergence.
It is not our aim in this chapter to examine all these questions. Our
objective is to use a detailed analysis of three countries as a starting point
for examining Verdier’s notion of hybridisation within and between
systems. We aim to expand and demonstrate Verdier’s hypothesis of
hybridisation and to show where countries are located in the conceptual

space of the ideal-typical regimes identified by Verdier (see Chapter 3)
for each of the four stages of their educational and training systems.
The changes that appear similar according to the major indicators
may in fact conceal differences in organisation, structure and societal
significance.
The three countries investigated (Denmark, Italy and the Netherlands)
can all to a greater or lesser extent be described as ‘unclassifiable’
according to the standard typologies and categories. Denmark was typ-
ically regarded as belonging to the corporatist apprenticeship model,
along with Germany. Italy is often classified alongside France in the
Southern European model (see, e.g. Iversens and Stephens, 2008). The
Netherlands, for their part, are sometimes assigned to the Nordic model,
sometimes to the German model. However, the results presented in
Chapters 1 and 2 pose problems from the point of view of the standard
categorisations. In the qualitative comparative analysis (QCA) analysis
in Chapter 2, Denmark is on the margins of the apprenticeship model
and the Netherlands, similarly, is on the margins of the ‘vocational’
model. There are grounds for investigating these borderline cases in
greater detail.
The four major segments of the education and training systems in
all three countries (compulsory schooling, upper secondary, tertiary
and further education) are analysed as a starting point for examining
hybridisation within and between systems. We draw on official sources,
both national and international, as well as on the existing literature.
Compulsory education: Comprehensive, but not

everywhere and with different principles
Compulsory education is said to be comprehensive in both Italy and

Denmark, but it nevertheless takes different forms. In Italy, compulsory
education begins at 6 and ends at 16, although the vast majority of chil-
dren (95 per cent) begin nursery school (‘scuola dell infanzia’) as early
at age 3. The initial stage of education, which lasts eight years, takes
place in primary schools (five years) and then in lower secondary schools
(three years) (Vaniscotte, 1997). Pupils undergo continuous assessment.
Pupils may be asked to repeat a year. Since the end of the 1990s (school
autonomy reform in 1997; revision of the constitution in 2001), various
reforms have strengthened the role of the regions and local authorities,
for instance with regards to the management of inputs. They have also
increased schools’ autonomy. In principle, schools are free to determine
Philippe Méhaut 97
how and what they teach, on the basis of a framework of competences

(Grimaldi and Serpieri, 2010). In reality, central government retains a
considerable role, since it determines the subjects to be taught and the
timetables (Salatin, 2011). However, with no inspections, schools are
playing their own game, with informal adjustments. While these same
reforms have also strengthened the role of head teachers, which is pre-
sented as a genuinely managerial one, many observers agree that this
change has been more formal than actual. Although now imbued to
some extent with the ethos (or rather the rhetoric) of the regulated mar-
ket, Italian compulsory education is labelled as a comprehensive system,
with no early selection, at least at the primary school. However, for the
final two years of compulsory schooling (from 14 to 16), pupils enter
the two first years of the upper secondary stage. This early transition to
tracked upper secondary (unlike in Denmark) is highly mitigating the
principle of comprehensive schooling. Italy’s lower secondary education
is also characterised by considerable within-school social segregation at
the class-level and by a relatively high rate of non-completion. Already
at age 13 pupils have counselling which marks the beginning of the
allocation to different tracks in upper secondary (Florenzano, 2009).
Interestingly, although at age 14 they have a national examination, the
results do not conclusively determine the students’ destination at the
upper secondary stage (general academic high school or vocational edu-
cation and training). Social backgrounds, specific requirements of some
schools and the transition from primary to lower secondary play at least
as important a role in the sorting process, and do so to such an extent
that one could speak of a kind of “early tracking” (Checci and Flabby,
2007).
Denmark also has compulsory education from 6 to 16 years of age.
Education at this stage is comprehensive but structured in line with the
‘Nordic’ model. The education system lasts nine or ten years. Grade 10 is
the additional orientation year. Thus unlike in the other countries there
is no distinction in Denmark between primary and lower secondary
school, which constitute a single stage. Education is provided by the
state and is free (although state-funded private schools are developing).
Pupils usually keep the same class teacher as they move from year to
year. The progression is individualised, even if there is a strong tradition
of group work, and there is no possibility of repeating a year. It was in
the 1960s and 1970s that the model of a comprehensive school, char-
acteristic of the Scandinavian countries, was consolidated (after intense
debates between supporters of comprehensive education and those of
tracking) (see Wiborg, 2009 for an historical perspective). Subsequent
reforms in the 1980s further strengthened the model. Today, this part
of the education system is regarded as having proved its worth. How-
ever, some changes have been introduced (decentralisation, free choice
for parents, stronger focus on academic achievement). As in other coun-
tries, national tests and evaluation of the schools are expanding, which
sit uneasily with the universalist character of Denmark’s system of com-
pulsory schooling (Wiborg, 2012). As in Italy, debates are also focusing
on the enrolment rate and success (or otherwise) of immigrant children
(Mrok and Rasmunssen, 1997). At the end of this stage of education,
national examinations play a part in deciding pupils’ options. Never-
theless, pupils can choose the level of the tests they take and may retake
them if necessary (particularly during the tenth year). There is no early
tracking and youngsters are prevented from dropping out of the system
altogether.
At this level, the Danish system is fully consistent with Verdier’s
‘universalist principles’, even if recent trends seem to somewhat
encroach on this regime (Wiborg, 2012) while Italy hovers between
universalism and the academic regime, largely because of the division
between primary and lower secondary education.
The Netherlands is clearly different. Firstly, compulsory education
lasts longer (from 5 to 18), and all pupils are expected to obtain an
upper secondary certificate, which means that compulsory schooling
lasts until the age of 18 at a minimum.2 Secondly, in 1985 the two years
of nursery school were integrated into primary school, which means
that the period of primary education is now exceptionally long, stretch-
ing from age 4 to age 12.3 There are both state and private schools.
Parents have freedom of choice when it comes to selecting a school.
The Netherlands is usually considered to be a country that practices
early selection (Vaniscotte, 1997). This takes place at age 12. We will
return to this question later. Before this age, the Dutch system hov-
ers between comprehensive education and early tracking. Schools enjoy
a fairly high degree of autonomy with regard to curriculum and their
weekly timetable (Eurydice, 2010). Some group pupils by ability, oth-
ers do not; some allow children to repeat a year, others do not. Since
2007, basic competence standards have been in place; they are based
on systems of tests chosen by the schools. A national test (CITO) was
introduced in the late 1960s. It is now widely used by 8 per cent of the
schools. The results of this test form the basis of pupils’ applications for
compulsory post-primary education. It is also used to assess the perfor-
mance of schools (which must be public and may guide parental choice).
There is a growing tendency to subject schools to quality evaluations
Philippe Méhaut 99
based on outcomes. This evaluation may lead to withdrawal of a school’s

accreditation and public funding. Thus what is emerging is a form of
organised market in education (in Verdier’s sense of the term), even
though schooling remains free to users.
At this stage, the question of system coherence and the linkage
between the various stages (i.e. hybridisation within the system) does
not arise. If we focus for the time being on hybridisation between sys-
tems, there are several major factors that seem to support the argument
that convergence is indeed taking place.
– There is a growing concern in all three countries with pre-school edu-

cation, one of the aims being to reduce inequalities due to social and
family background.
– There is a tendency to increase the length of compulsory education,
even though the extension takes different forms.
– There is an increasing use of standardised tests at the end of pri-
mary and lower secondary. This is certainly the case for NL with
the CITO test at the end of primary and the ‘Centraal Schriftelijk’
for VMBO and HAVO at ages 16 and 17 respectively (assuming no
repeating of year).
Nevertheless, as we have just seen, only Denmark has full comprehen-

sive education throughout the whole of compulsory schooling. Some
of the characteristics of Italian primary education are similar to the
Danish model. However, the division between primary and lower sec-
ondary education indicates a more academic regime at this level. While
the Dutch system has an old historical tradition of free choice, associ-
ated to the freedom for catholic, orthodox protestant and Jewish parents
of founding schools presenting a curriculum in line with their religious
beliefs/ worldview opinions, school league tables based on results and,
in particular, selection at age 12 put it more in the ‘regulated market’
category and, in subsequent stages, the corporatist regime.
Upper secondary (and access via lower secondary):

Enrolment rates of around 80 per cent in all three
countries but by different routes
Two of our three countries have seen a rise in access to upper secondary
education. Between 2000 and 2010, the share of 20–24-year-olds having
completed upper secondary school increased from 70 to 78 per cent in
the Netherlands and from 69 to 76 per cent in Italy (Eurostat, 2010).
Denmark, which was at 72 per cent in the year 2000, fell to 70 per cent
in 2010. For this country, we shall see that account has to be taken of late
entries/re-entries (the average age of apprentices is high), which should
help to improve Denmark’s score since an older age group would be
included (OECD, 1998). It should also be recognised that during the first
decade of the twenty-first century and before the crisis, unemployment
in Denmark was very low and that the resultant tensions in the labour
market encouraged some people to enter the labour market at a younger
age than usual, in the knowledge that they could always return to edu-
cation or training at a later date (Danish Ministry of Education, 2010).
In sum, all three countries can be said to have developed mass upper
secondary education. Furthermore, prompted by the EU, they have all
made policy pledges to raise the rate of access to upper secondary educa-
tion. The Netherlands had been the most ambitious in this regard, with
a target of 95 per cent to be reached by 2020.
How has this mass upper secondary education developed? Are upper
secondary ‘regimes’ tending to become standardised? In Italy, selection
occurs on the basis of a public examination taken at around 14 years of
age. The best pupils spend five years on academic studies in high school,
which is the fast track to higher education. Thus there are two years of
compulsory education after lower secondary school and selection takes
place on academic and social criteria at a fairly early stage, with few
opportunities to change course, as we shall see. Before the 2003 Moratti
Reform, there was a parallel vocational training system that existed vir-
tually outside the standard education system. Its sole purpose was the
development of very specific skills that would enable trainees to enter
the labour market very quickly. This training, provided by technical and
vocational institutes, was regarded as a ‘pre-vocational’, at level ISCED
2. The 2003 Act strengthened the introduction of more broadly based
courses into vocational training and tended to align it with the liceo
model. This process was continued with the Gelmini reform, which
brought the technical and vocational institutes into line with the gen-
eral academic licei by establishing a complete five-year cycle, albeit with
an intermediate exit point after three years in the case of the vocational
institutes (Mengucci and Romano, 2006; Cedefop, 2011). Alongside this
‘state’ vocational pathway, there is a regional vocational pathway, with
public and private training centres that provide three- and four-year
training programmes. This training is the responsibility of the regions,
which adapt it to match their particular characteristics and require-
ments. Before the reform, the qualifications obtained in the course of
one of these training programmes were valid solely within the region.
Philippe Méhaut 101
With the integration of vocational training into the education system,

vocational qualifications became equivalent to the other academic qual-
ifications awarded and recognised throughout the whole of the country
(third EQF level for the Qualifica and fourth for il Diploma).
Thus Italy has been through a dual process, gradually incorporat-
ing vocational training into the education system while at the same
time developing the vocational pathway. However, social divisions and
meritocracy remain very strong. Although there are, in principle, bridges
between the academic liceo and vocational training, they are in practice
highly theoretical. At the end of upper secondary school, 91 per cent
of young people attending licei go on to university, compared with
35 per cent of those from the technical and vocational institutes, even
though in theory they enjoy the same level of access. Young people in
the regional vocational training centres do not have access to higher
education. Those who have completed their upper secondary education
in the licei are much more likely than their counterparts from the tech-
nical and vocational institutes to pass their examinations on schedule
and not drop out of university. Both the technical and the vocational
institutes also have a high share of foreign pupils. Finally, one other
important piece of information is the examination failure rate in these
institutes, which is much higher than it is in the licei: 18.1 per cent com-
pared with 5.4 per cent in the classical licei in 2005–2006 (Florenzano,
2009).
It should be added that, in contrast to the other two countries, appren-
ticeships are virtually unknown in Italy. In quantitative terms, they are
marginal and even though apprenticeship contracts do exist they can
basically be regarded as informal. Nevertheless, a reform was introduced
in mid-2011, the impact of which remains to be seen.
Upper secondary education as a whole underwent a series of reforms
between 2003 and 2011, some of them introduced simply in order to
cancel out the previous one (Ballatore, 2010). Ultimately, it is fairly dif-
ficult to reconstitute the data, since the classifications by ISCED level
have changed. Thus in 2000, for example, Italy declared that about
20 per cent of its pupils were at ISCED vocational level 3; by 2010,
this figure had risen to almost 70 per cent! If the more precise Refernet
data are to be believed, in 2009/10 approximately 43 per cent of pupils
were in the ‘general’ streams, 31 per cent in the technical institutes,
19 per cent in the vocational institutes and 7 per cent in the regional
institutes (Cedefop, 2011).
Thus Italy, like our other two countries, has put in place a mass
upper secondary education system. It has done so by developing the
vocational track and integrating it into the rest of the system. How-
ever, this upper secondary system can be said to be dominated by an
academic regime, thus marking a break with the comprehensive system
of primary schools: general academic education is separate from voca-
tional education/training, there is a high level of selection and there is
a significant failure rate, particularly in the vocational track. Although
the drop-out rate has declined from 24 per cent to 18.2 per cent, it
is considerably higher than that in the other two countries (slightly
higher than 9 per cent), which further supports the academic regime
hypothesis. It should be added that the social partners play little role
in the vocational track, making it difficult to speak of corporatist
regulation.
As already noted, Denmark was traditionally regarded as belonging to
the ‘Germanic’ apprenticeship model, with a rigid separation between
the academic gymnasium and the vocational pathways. This is now only
partially true, largely because of the many reforms that have been intro-
duced into the apprenticeship system (Danish Ministry of Education,
2008; Cedefop, 2011). The selective gymnasium is certainly the fast track
to higher education and is the destination for the best pupils at the end
of compulsory education (45 per cent). There is another pathway (HF),
parallel to the gymnasium, in which students can prepare for university
entry in two years. However, this is intended for other groups, mainly
pupils whose education has been interrupted (because they left educa-
tion immediately after compulsory schooling or because they started
an apprenticeship, which they may or may not have completed) and
adults. The vocational pathways cater for a bare majority of pupils. One
of these is the ‘technological school’ (HHX and HTX), which leads on to
higher education (mainly non-university) (18 per cent), while the other
is apprenticeship (30 per cent) (Danish Ministry of Education, 2010).
A strong gender effect can be observed in this segment of the education
and training system (more than 50 per cent of girls continue in the gen-
eral academic stream, compared with 30 per cent of boys). However, a
significant proportion of girls (about 45 per cent) subsequently opt for
apprenticeship, thereby weakening the gender effect.
The apprenticeship system has undergone radical reform, to the point
where it is no longer possible to speak of ‘apprenticeship’ but rather
of basic vocational training (Ministry of Education, 2008; Wiborg and
Cort, 2009). Four striking developments in particular should be noted.
– Firstly, there has been a drastic reduction in the number of specialisms

as basic ‘common trunk’ courses have gradually been put in place and
the principle of free choice for pupils (with the possibility of changing
specialism during a course) has become established.
– Secondly, individuals now have the option of starting with either a
period of full-time education or a placement in a company.
– The length of an apprenticeship is now highly variable, depending on
the specialism and pupils’ own programme. The length ranges from
18 months to 3½ years depending on the specialism, with variations
depending on the evaluation of pupils’ attainment on entry to the
programme.
– Finally, the average age of apprentices, which is close to 23, clearly
reflects a system with ‘multiple entry and exit points’: apprentices
enter after experience of the labour market, after general or technical
secondary education or after an initial apprenticeship and adults even
embark on apprenticeships after the age of 25.
Finally, there are several characteristics that suggest that what can be
observed in Denmark is the hybridisation of a ‘corporatist’ regime as
it evolves towards a ‘universalist’ regime, in line with what is the rule
in compulsory schooling. The bridges between the various pathways
are numerous and real. There is much toing and froing between educa-
tion, employment and other personal choices, underpinned by a right
to change direction up to the age of 25, a right based on a social security
regime specific to young people (see also Van de Velde, 2008). Failure is
not irreversible. While the social partners play an important role at all
levels, they do so on a tripartite basis with the public authorities (central
government and local authorities).
The system in the Netherlands, where almost 80 per cent of any
one age group is now in upper secondary education, has also under-
gone considerable change. Selection by ability takes place at age 12,
at the end of eight years of primary education. Following national
tests, a choice is made between general education that will prepare
pupils for university entry (VWO), general education preparing pupils
for non-university higher education (HAVO) or pre-vocational and then
vocational education (VMBO). Since 1996, virtually all vocational edu-
cation and training has been managed within the education system by
the same rules (Westerhuis, 2009; Visel, 2011). Some 70 public institu-
tions offer a range of training programmes regulated by the state and
the social partners. As in Italy and Denmark, the number of training
specialisms has tended to decline in favour of more broadly based, less
specific programmes, particularly in the initial years. As in Denmark,
there are ‘bridges’ between general and vocational education. Two of
the most pronounced characteristics of the Dutch system bring it close

to the French model, making the situation in the Netherlands a very
hybrid one.
Firstly, at the end of lower secondary school, the upper secondary
vocational stream is divided into groups of occupations/ professions on
four levels (MBO 1 to 4), with outflows into the labour market at all
four levels. Rather like in Italy, tracking takes place as early as lower sec-
ondary school (VMBO in two years). These two years are generalist years
common to all those who enter the vocational stream. However, the
curriculum (and hence pupils’ subsequent prospects) varies according
to pupils’ ability. Thus, for example, VMBO-BL (21 per cent of pupils)
preparing for short vocational training courses at the upper secondary
stage (MBO2) coexists with VMBO TL (41 per cent of pupils), where the
programme is more general and theoretical and leads on to longer train-
ing courses (MBO4) (Facts and Figures, 2007). Secondly, a school-based
track and an apprenticeship track coexist, both leading to the same qual-
ifications. With almost 70 per cent of boys and 65 per cent of girls in
vocational education, the Netherlands is the country in our sample with
the highest share of vocational education (Eurostat, 2010). This share
remained relatively stable between 2000 and 2010, whereas it declined
by about ten percentage points in Denmark. The share of apprentice-
ships varies from speciality to speciality. However, with approximately
20 per cent of apprentices in upper secondary school, the Netherlands
has a significantly more ‘school-based’ upper secondary system than
Denmark. Vocational education and training is public and free. The
social partners play an essential role in all matters related to regulation
of training programmes and vocational qualifications.
Despite the division into different levels, the Netherlands manages
to educate the majority of each age group up to ISCED 3A level. The
percentage of those failing to attain this level has fallen to 9 per cent
(close to the figure for Denmark). As A. Westerhuis (2009) put it, the
Dutch vocational system provides a form of comprehensive vocational
education and training for all, even though it is based, ex ante, on early
tracking.
In short, the common trend towards the establishment of mass upper
secondary education is a powerful factor in internal hybridisation,
between corporatist and universalist regimes in Denmark and between a
corporatist regime and a combined universal/academic regime in the
Netherlands, with only Italy’s system of upper secondary education
clearly retaining its exclusively academic character. Formally, the influ-
ence of the EU, and particularly of its main benchmarks (drop-out rates,
rate of access to upper secondary education, and so on) as well as its tools
such as the EQF, makes itself felt, even though the subsidiarity principle
leaves room for a fairly wide range of different policy choices. Thus in all
three countries, and in particular in the Netherlands, the construction of
qualifications is now based on competences and outcomes (Westerhuis,
2009). Significantly, Denmark has reformulated its basic policy state-
ments (Denmark Strategy, 2007) by adopting European rhetoric with
the aim of presenting itself as the star pupil in the European class.
While Italy is experiencing a form of ‘legislative instability’ (Ballatore,
2010), many of its plans are also based on European guidelines. These
guidelines are used by national actors less in a spirit of mimesis but
rather as levers that can be pulled in order to advance (or, conversely,
thwart) national policy. Over the past few years, the authorities in the
Netherlands have been implementing a plan to reduce the drop-out rate,
with systematic individualised monitoring and financial incentives for
schools. Social partners and local authorities have pushed this issue for
different reasons (labour shortages, social risks linked to the drop out).
Influence of the EU benchmarks may be also playing a role.
As a result, national systems are developing in parallel with each other
as measured by the major indicators, and particularly in terms of the
rate of access to upper secondary schooling. However, this should not
conceal fairly different failure rates in secondary education (Italy vs.
Denmark and Holland) nor significant differences between the countries
with regard to the extent of vocational training, its mode of develop-
ment and the extent to which it is both integrative and integrated into
the education system as a whole.
Higher education: Purely academic or bipartite?
Rates of access to higher education have increased significantly. Here

too, a parallel form of development can be observed. Between 2000
and 2011, the percentage of the 30–34 age group with a CITE five or
six level qualification rose from 32 to 41 per cent in Denmark, from
26.5 to 41 per cent in the Netherlands and from 11 to 20 per cent
in Italy (Eurostat, 2010), which is still considerably lower than the
other two countries, thereby confirming the analysis of a ‘meritocratic’
regime. As far as convergence is concerned, it should be noted that the
Bologna process is being pursued in the universities in all three coun-
tries and that consequently the standard bachelor’s/master’s/PhD degree
hierarchy is in place. It should also be noted that, out of a concern to
increase access to higher education, some bridges have been established
between vocational training and higher education. Moreover, specific

vocational pathways have also been established within non-university
higher education.
In Denmark, alongside the 12 universities (65,000 undergraduates and
66,000 post-graduates), there are some hundred or so non-university
institutions offering either vocational bachelor’s degrees or short (two-
year) vocational training programmes (approximately 90,000 students)
(Jørgensen, 2012). Access to all forms of higher education has increased
considerably (to more than 45 per cent of the relevant age group,
with a target of 50 per cent). It varies depending on the type of
secondary schooling received. Of those completing the general aca-
demic upper secondary school (gymnasium), 45 per cent will go on to
university and 35 per cent to vocational courses and vocational bache-
lor’s programmes in higher education, compared with 27 per cent and
33 per cent respectively for those leaving the school-based vocational
stream and 9 per cent of those completing apprenticeships (The ministry
of Children and Education, 2012). The percentages are higher among
the children of university graduates, but the success rates are not signifi-
cantly different. In the light of what was said earlier about the toing and
froing between education, employment and other personal choices and
the opportunities for late entry via apprenticeships, it is not surprising
that the median age of Danish students (25) is the highest in our three
countries. In this respect, it is closer to the German than to the academic
model, in which entry into higher education generally follows on imme-
diately from general academic upper secondary school. As we saw above,
one of the routes into higher education (HF) can be described as based
on the principle of offering a second chance to pupils with discontinu-
ous school careers and to adults. Moreover, as in secondary education,
vocational qualifications are strictly regulated by the social partners.
Thus in line with its secondary school system, Denmark’s higher edu-
cation system is characterised by fairly individualised entry routes, based
on ‘bridges’ and opportunities for those in need of a second chance (see
below). The system could be described as bipartite, with the academic
sector based on the research-led universities accounting for just half of
the educational opportunities beyond secondary school. In the voca-
tional segment of higher education, corporatist modes of regulation are
very much the norm in the management of flows and specialisms.
Higher education in Italy is almost wholly synonymous with uni-
versity education and until the mid-1990s no real attention was paid
to non-university post-secondary education (Ballatore, 2010). Since
the regulation of 3 November 1999, which followed the Universities’
Autonomy Act of 1997, each university manages and organises its own
education provision and decides on the organisation and structure of
the teaching and courses leading to state-recognised degrees, which are
approved by the minister.
The unified Italian university system in fact conceals significant differ-
ences between universities and disciplines. What Roberto Moscati calls
the ‘social specialisation’ (specializzazione sociale) of the various facul-
ties (facoltà) is a marked feature of the system. The choice of faculty
depends on students’ socio-cultural background, the type of baccalau-
reate they possess and their previous academic results (Moscati and
Rostan, 2000). The prestige of the various courses, and the consequent
university hierarchy, is determined by the unequal social composition
of the registered students. As mass higher education has become estab-
lished, the dualist divide between top-ranking faculties (engineering,
materials science, medicine) and those in the second rank (arts, human-
ities, social sciences, law) has been reinforced. Delays in completion
and numbers of drop-outs have also increased. Thus there are many
students, the so-called ‘fuoricorso’, who have been registered for a sub-
ject for more years than are theoretically required to obtain a degree
(more than 40 per cent are in this situation). Until relatively recently,
there was only one university degree, Laurea, which could be obtained
after four or five years’ study, depending on the subject. The Bologna
process (called 3 + 2 in Italy) has led to the introduction of an inter-
mediate degree (Laurea breve). However, this has not changed students’
self-management of their own university careers, the rush to obtain mas-
ter’s degrees and the high drop-out rates. It is true that IFTS, Istruzione e
Formazione Technica Superiore, was introduced in 1999 in response to the
demand from the labour market for highly specialised and skilled profes-
sionals, a move that marked the first tentative steps towards a bipartite
higher education system. However, the Italian system is still very far
from having the balance of the Danish system or, as we shall see, that
found in the Netherlands. In sum, the Italian higher education system
is still strongly academic in character.
The Netherlands has also seen its higher education system expand.
As in Denmark, the system includes both universities (14) and insti-
tutions (HBOs) providing vocational training programmes (mainly at
bachelor’s level but in some cases at master’s level as well). Both seg-
ments have conformed to the Bologna process, are largely publicly
funded and are governed by national regulations. As in the other coun-
tries, the main route to university runs through general secondary
education, with a national system of guidance (and selection for courses
on which numbers are limited). However, the balance between the

academic and vocational pathways is very distinctive (Ministry of Edu-
cation, Science and Culture, 2012). Approximately 50,000 students start
university each year, the vast majority of them coming straight from
school. About 90,000 students enter the HBOs each year, a third of
them from general secondary education. About 41 per cent of the MBO4
are progressing to HBO, which shows the strong connection between
VET and Higher professional education. Nevertheless, many students
enter HBO at a later stage, after experience of work. Thus the Dutch HE
system is truly bipartite and, as in Denmark, the vocational segment
is very much shaped by the social partners and corporatist modes of
regulation.
Lifelong learning: Second chance or placebo?
With 44 per cent of adults enrolled on a lifelong learning programme

in 2006 (a sharp increase since 2000), Denmark is the European leader
in this regard. Its position is based on both very active company
policies (85 per cent of Danish companies’ state that they provide
training for their employees) and a very active public policy, also imple-
mented under the aegis of the Ministry of Education. Some 600,000
individuals, more than 15 per cent of the economically active pop-
ulation, take advantage of this public policy each year. It is targeted
mainly at the unemployed, on the one hand, and at adults ‘with low
levels of training’, on the other (Kvist and Pedersen, 2007). Increas-
ingly, however, it also involves access to high-level continuing training
programmes.
Continuing training is provided mainly by public centres (including
some of the courses funded by companies). These are, firstly, initial
training centres, all of which also offer continuing training programmes,
either in the form of ‘ad hoc’ courses (CVT, approximately 2500 courses,
mostly of short duration and open to employees and/or the unem-
ployed) or of courses equivalent to those provided for young people
but open to adults. Then there are specific centres (AMU) that target the
unemployed more specifically. In both cases, the provision of courses
is under the control of the social partners, at both national (definition
of CVT) and local level. Calculated in terms of full-time equivalents,
approximately 70,000 adults are enrolled on public training courses, a
figure that can be compared, very roughly, with approximately 220,000
pupils in upper secondary school and slightly fewer than 200,000 in
higher education (Danish Ministry of Education, 2010). In 2007, as part
of its lifelong learning strategy and in line with EU recommendations,

Denmark increased public spending on lifelong learning by more than
20 per cent, which would improve its results in this area even further.
In the end, two facts should be underlined. The first is the existence
of a genuine ‘second chance’ that enables a not insignificant share of
adults to obtain a qualification. Following on from this, even though
Denmark is not among the best performing OECD countries in terms of
the share of the population at upper secondary level (53 per cent of the
25–64 age groups, compared with 63 per cent in Austria and 60 per cent
in Germany), largely because of the numbers of pupils leaving at the end
of the compulsory education, it is, nevertheless, one of the countries in
which intergenerational differences are smallest (Westergaard-Nielsen,
2008, p. 63). Thus Denmark’s lifelong learning strategy is consistent
with the universalist principles, even though in the case of ‘initial’
secondary education it also has some of the elements of a corporatist
regime (particularly with regard to the role of the social partners in
determining the supply).
In Italy, according to Geroldi and Paci (1997), 97.8 per cent of edu-
cation and training expenditure is targeted at young people and only
2.2 per cent at adults, a situation that is unique in Europe. All the
international reports are agreed that little importance is attached in
Italy to adult education/training. The concept of lifelong learning was
introduced only very recently and Italy has very little in the way
of coherently linked measures and proposals, whether at national or
regional level (Richini, 2002). Italy is one of the countries that has expe-
rienced the greatest difficulty in meeting its Lisbon Strategy targets in
the area of lifelong learning. Thus a 2009 report notes that a third of
the population of adult age is at serious risk of illiteracy (Florenzano,
2009). In 2006, only 22 per cent of the reference population (25–64-
year-olds) took part in training measures, a figure that puts Italy at the
bottom of the European table. Moreover, in contrast to Denmark and
the Netherlands, where age inequalities are very low, the rate of access
among 45–54-year-olds falls to 13 per cent (Eurostat, 2010).
Despite these figures, various aspects of lifelong learning and related
problems have been at the heart of the debate between providers,
national and local institutions and trade unions and employers’ organ-
isations. Today, around 40 per cent of Italian private companies and
59 per cent of workers subscribe to the Fondo paritetico Interprofessional
(FPI). These national lifelong learning funds are associative organisa-
tions managed by the social partners’ representative organisations. The
FPIs fund training plans for entire industries, companies or groups of
companies. These inter-occupational funds are constituted in a way that

is entirely new in Italy. For the first time, significant public resources
are being allocated to the social partners. Nevertheless, with only
32 per cent of companies providing training for their employees, Italy is
a long way from the European average and even more so from Denmark
(Florenzano, 2009). There are also ‘evening classes’ for workers in Italy.
A number of schemes targeted at adults, mainly those with low levels of
education and training, are being developed at regional level, financed
mainly by European funds. Thus there is a sharp divide in Italy between
a still underdeveloped lifelong learning system, with a low level of state
involvement, and initial education and training. This divide is fairly
characteristic of a meritocratic regime, where often it is in school that
everything is decided.
The Netherlands is a hybrid case. With 44 per cent of adults taking
part in lifelong learning, it is on a par with Denmark, without any
very evident differences between job seekers and those in employment,
which indicates the existence of active employment policies (Ministry
of education, 2011). Two segments co-exist as they do in Denmark,
although the social partners have less power over qualifications. The
first consists of the pathways open to adults within the formal sys-
tem (VAVO). This pathway has changed, as it is now mainly used for
young people trying to obtain an HAVO diploma after failure in the
AVO/general education colleges. However, the MBO track, with about
10 per cent of trainees over the 30 years of age and, as in Denmark, a
high average age for apprentices (50 per cent aged over 24) is still well
alive. Employers are keen to make use of this opportunity to upgrade
their low qualified work force (Meer Van der and Fleur, 2012). Some of
these adults are in receipt of financial assistance, whether from national
or local government; however, they may also be self-funding or financed
by their employer. Taken as a whole, the Dutch system, like that in
Denmark, has the hallmarks of a second chance (or career advancement)
vocational pathway established within the education and training sys-
tem. However, access is lower than in Denmark, and in this respect the
Netherlands is closer to Germany than to the Nordic countries.
Firms are fairly active in developing and implementing employee
training policies (75 per cent of firms provide training). These poli-
cies are based in part on jointly managed industry-wide training funds,
similar to those being developed in Italy and an object of collective bar-
gaining. Firms contribute to the funds through a training levy. There are
140 funds covering about 85 per cent of employees. However, the Dutch
system is a market regime, with a multitude (‘jungle’) of private suppliers

(profit-making or otherwise) and no state intervention. However, some
funds operate an accreditation policy (for their sector only).
Thus the lifelong learning landscape is somewhat fragmented. In line
with Verdier’s hypotheses, the dominant academic regime in Italy is
reflected in a low level of continuing training provision; here, it is in ini-
tial training and education that individual careers are decisively shaped.
Even though recent developments seem to indicate increasing involve-
ment on the part of the social partners through the establishment of
industry-wide training funds similar to those in the Netherlands, their
activities are still not sufficiently well established (and are also too lit-
tle studied) for it to be possible to speak of real change. Denmark is
the country in which the universalist options have been significantly
extended into the field of lifelong learning, albeit in combination with
elements borrowed from the corporatist regime. It is the country where,
from the point of view of education and training policy, internal coher-
ence, based on public action shared by all the actors, is strongest. The
Netherlands are providing the most blurred picture. On the one hand,
a part of the public and firm’s policies linking initial and further educa-
tion and training systems are offering opportunities for a second chance.
On the other hand, a market regime is regulating a second segment of
the further training activities.
Conclusion
Taking up the biological metaphor on species convergence again and

extending it, what can we conclude from this analysis of three cases
(which needs to be expanded), all of which were selected for their
‘fringe’ characteristics and whose value is therefore more experimental
than general?
The first observation to be made is that the measures aimed at
satisfying increasing demand, which is partially ‘self-generated’, have
developed in parallel. The success of virtually universal compulsory
schooling has increased demand from families and pupils for educa-
tional opportunities to be extended into upper secondary school and
then into higher education, regardless of the expected future returns
(at least in strict terms of returns in the labour market). Even where
those returns are not very evident, as in Italy, mass education has had to
be introduced in order to fulfil expectations. Some of these expectations
are undeniably in line with the strategies adopted by employers, who are
looking for better trained workers in order to increase competitiveness,

particularly if they can find them at little direct cost to themselves,
and/or increased opportunities for selection based on the signalling
role of education/training. This trend is driven in part by international
organisations, which have promulgated an ideal notion of the good
society and of competitiveness (‘the knowledge economy’) to which
all developed societies should aspire at the risk of falling behind. They
have established benchmarks that determine the size and direction of
the ‘tunnels’. These are all powerful factors in and evidence of parallel
developments, as attested by the major common indicators.
For all that, have these parallel developments been accompanied by
the development of tools with common functionalities and sufficient
similarities to make it possible to talk of convergence? On the basis
of the three countries investigated here, the answer is very ambigu-
ous. While all three countries have developed pre-primary education,
its significance varies depending on whether it is linked to a totally
comprehensive system (as in Denmark) or a more selective one. Simi-
larly, the propensity to increase the presence of vocational education in
secondary education has not been implemented in the same way. The
differences are also considerable in higher education and even greater
in lifelong learning. The three systems are all hybrid ones (as defined by
Verdier) but their peripheral segments seem to be developing in a way
consistent with their own internal logics rather than borrowing from
other systems (hybridisation within the system seems to be prevailing
over hybridisation between systems).
There remains the question of what might be called convergence.
Although they are all undeniably characterised more by internal
hybridisation (which seems to bring them closer together), the three
systems still belong to different dominant regimes. As far as Italy is con-
cerned, an academic regime continues to prevail, despite some recent
borrowings from the French vocational training system, which has led
only to some mechanical changes. In the Netherlands, the complex
combination of an academic and a corporatist regime, with the intro-
duction of more market-based rules, particularly in lifelong learning as
well as in compulsory schooling, has led to more profound changes.
These are reflected in particular in attainment rates, although they have
not altered the fundamental characteristics of the education and train-
ing system, namely its strongly vocational organising principle and its
strong links with the labour market. Denmark is a more complex case.
Its initial situation placed it among the apprenticeship and occupational
labour market systems. However, there are a number of indications in all
segments and at all levels of the Danish education and training system
that suggest that a fairly profound structural change is taking place,
involving a shift towards a ‘hybrid Nordic’ model in which universalist
principles would be combined with the strength of corporatist rules.
Notes
1. This chapter is based on case studies carried out for the Edesco project. Special
thanks go to Magali Ballatore. The author thanks Anneke Westerhuis, Susanne
Wiborg and Monica Mincu for their useful comments on earlier drafts of this
chapter.
2. The period of compulsory schooling was extended from 16 to 18 years of age
on 1 August 2007 and was made conditional on the obtainment of a certificate
of upper secondary education. Pupils without such a certificate (‘startkwal-
ficatie’) are obliged to stay in education until at least 18 years of age and
preferably until such a diploma is obtained. Diplomas listed as a ‘startkwalfi-
catie’ are: MBO/secondary vocational education at least at level 2, or a diploma
from HAVO or VWO.
3. Primary education starts at the age of four (from four to 12), but compulsory
education starts at the age of five. In other words four-year-olds are not obliged
to attend primary education. However, most of them do.
References
Ballatore M. (2010) Monographie du système italien, Edesco, Lest, mimeo, unpub-
lished.
Bosch, G. and Charest, J. (eds) (2009) Vocational Training, International Perspectives.
Cedefop (2011) Overview of the Vocational Education and Training System, Denmark:
Refernet.
Cedefop (2011) Italy – VET in Europe Country Report, Refernet. http://libserver.
cedefop.europa.eu/vetelib/2011/2011_CR_IT.pdf
Checci, D. and Flabby, L. (2007) Intergenerational Mobility and Schooling Deci-
sions in Germany and Italy: The Impact of Secondary School Tracks, IZA DP,
No. 2876.
Danish Ministry of Education (2008) The Danish Vocation education and Training
System, 2nd edition.
Danish Ministry of Education (2010) Facts and Figures 2009.
Denmark’s strategy for lifelong learning, Education and lifelong skills for all
(2007) Report to the European Commission.
Eurostat (2010) Statistiques par thèmes, http://epp.eurostat.ec.europa.eu/portal/
page/portal/education/data.
Eurydice (2009/2010) Structures of Education and Training Systems in Europe,
Denmark. https://webgate.ec.europa.eu/fpfis/mwikis/eurydice/index.php/
Denmark: Overview
Florenzano, F. (2009) Rapporto sull’apprentimento permanente in Italia 2009.

Dati, analisi, problemi, proposte per ritrovare una chiave per lo sviluppo
economico, Rivista italiana di Lifelong learning n◦ XX.
Geroldi, G. and M. Paci (1997) Le politiche del lavoro, commissione per l’analisi
delle compatibilità macroeconomiche della spesa sociale, Documento di base
n◦ 2, Presidenza del Consiglio dei ministri, Rome.
Grimaldi, E. and R. Serpieri (2010) ‘The Reforming trajectory of the Italian Educa-
tional System. Site-based Management and Decentralization as a Challenge for
Democratic Discourse’, Journal of Educational Administration and History, XLII–I,
75–95.
Iversen, T. and J. Stephens (2008) ‘Partisan Politics, the Welfare State and 3
Worlds of Human Capital Formation’, Comparative Political Studies, XXXXI,
600–637.
Jørgensen, C. H. (2012) Linking the Dual System with Higher Education in Denmark:
When Strength become Weakness. Paper presented at Nordyrk 2012, Roskilde,
Denmark.
Kvist, J. and L. Pedersen (2007) ‘Danish Labour Market Activation Policies’,
National Institute Economic Review, CCII, 99–12.
Meer, Van der M. and E. Fleur (2012) Patronen van bekostigd oderwijs-het mbo,
Utrecht/’s-Hertogenbosch: ecbo.
Mengucci, R. and R. Romano (2006) L’évoluzione dell’istruzione Professionale.
Studi e documenti degli annali della pubblica istruzione: L’istruzione tecnica
e professionale verso un nuovo futuro, Rivista trimestrale del Ministero della
Pubblica istruzione, 115–116.
Ministry of Education (2008) Facts and figure 2007, Key figures in Education,
Statistical publication no 3.
Ministry of Education, Culture and Science (2012) Key Figures 2007–2011, Edu-
cation Culture and Science, Ministry of Education, Culture and Sciences: The
Hague.
Moscati, R. and M. Rostan (2000) Higher Education and Graduate Employment
in Italy, European Journal of Education, XXXV, No. 2.
Mrok, J. and M. Rasmussen (1997) Réformes et nouvelles approches aux problèmes de
l’échec scolaire et de l’exclusion sociale au Danemark, 1985–1995, GRET, Université
de Barcelone.
OECD (1998) Thematic Review of the Transition from Initial Education to Working
Life, Denmark Background Report.
OECD (2009) Learning for Jobs, OECD Policy Review of Vocational Education and
Training, Initial Report.
Richini, P. (2002) Lifelong Learning in Italy. The extent to which vocational edu-
cation and training policy is nurturing lifelong learning in Italy, CEDEFOP
Panorama series, XLIII, Luxembourg: Office for Official Publications of the
European Communities.
Salatin, A. (2011) Modèles de curricula et politiques curriculaires en Italie, Revue
internationale d’éducation de Sèvres: « Le curriculum dans les politiques éducatives »,
LVI, 41–50.
The Ministry of Children and Education (2012) Education and Training in
Denmark, Facts and Key Figures.
Van de Velde, C. (2008) Devenir adulte, Sociologie comparée de la jeunesse en Europe,
Paris: Puf, le lien social.
Vaniscotte, F. (1997) Les écoles de l’Europe, systèmes éducatifs et dimension

européenne, IUFM, INRP.
Vissel, K. (2011) Netherlands Vet in Europe, Country report, Ecbo, Refernet,
Cedefop.
Westergaard-Nielsen, N. (ed) (2008) Low Wage Work in Denmark, New York: Russell
Sage.
Westerhuis, A. (2009) ‘The Meaning of Competence’, in Brockmann, M., Clarke,
L. and C. Winch (eds) Knowledge, Skills, Competence in the European Labour
Market: What’s in a Qualification? 68–84, Abingdon: Routledge.
Wiborg, S. and P. Cort (2009) ‘The VET System in Denmark, Continuity and
Change’, in Bosch, G. and J. Charest (eds) Vocational Training, International
Perspectives, 84–109, New York: Routledge.
Wiborg, S. (2012) Neo-liberalism and Universal State Education: The Cases of
Denmark, Norway and Sweden 1980–2011. Comparative Education. 1–17.
5
Curriculum Patterns in Citizenship
Education: A Cross-National Study
Christine Han, Jan Germen Janmaat, Tom May and Paul Morris
All states are interested in the political education of their young people,
and it is mainly through citizenship education that they attempt directly
to socialise youngsters into the norms, values, and conduct expected
of citizens. We understand ‘citizenship education’ in the broadest pos-
sible sense, that is, as referring to those aspects of school education
that are intended to have socialising effects with respect to citizenship,
including civic education, moral and values education, social studies,
history, geography, literature, physical education, military training, vol-
unteering, service to the wider community and so on. Countries vary
in their conceptions of the ideal citizen and, consequently, citizenship
education can be quite different across countries. However, research
examining these differences shows theoretical and empirical lacunae.
Where theory is concerned, existing research has been dominated by
the influential claim of the World Culture theorists that national cur-
ricula are converging to a transnational normative model emphasising
human rights, democracy and individual autonomy (e.g. Frank et al.,
2000; Soysal and Schissler, 2005; Wiseman et al., 2011). Although World
Culture theory has been criticised by many, few of the critics have pro-
posed an alternative view that could describe and explain cross-national
patterns and trends in citizenship education. This chapter will address
this omission by developing two such perspectives. These perspectives
can be said to rival World Culture theory in that they predict different
outcomes across countries.
From an empirical point of view, existing research investigating the
nature of citizenship education across countries is limited. Apart from
in-depth case studies of individual countries, comparative research
examining a range of countries on specific criteria tend to focus on
superficial features, such as curriculum objectives and structure (e.g.
116
Christine Han, Jan Germen Janmaat, Tom May and Paul Morris 117
Eurydice, 2005), and school subject categories (e.g. Benavot et al., 1991).
The more substantive issue of the values and norms promoted in the
curriculum is usually ignored because of the difficulty in investigating
these. This chapter will explore this values dimension, in addition to
the structural features of citizenship education, across 16 Western and
East Asian states.1 It will use cross-country patterns of both these dimen-
sions to broadly assess the validity of World Culture theory, and the two
alternative perspectives developed below. Thus the current contribution
can be said to enhance our understanding of the area in at least three
ways: (1) it elaborates on two alternative perspectives with regard to
international differences in citizenship education curricula; (2) it uncov-
ers the values promoted in these curricula; and (3) it contributes to the
development of theoretical perspectives in the field.
We start by explaining World Culture theory in greater detail and
presenting the two alternative perspectives. We then discuss the two
dimensions of citizenship education curriculum, and the ways in which
we will measure these. We will also present descriptive statistics of the
two dimensions and offer some preliminary observations regarding the
validity of the three theoretical perspectives. In the final section, we link
the findings more fully to these perspectives.
World Culture theory
World Culture theory enjoys widespread familiarity – even endorsement

in some instances – among scholars interested in citizenship education.
For Boli (2005), the notion of world culture applies if an action ‘invokes
a framework of meaning that (it is believed) should apply everywhere’
(p. 386); world culture elements include ‘general models of purposive
organizations, general theories of the benefits of increasing trade, gen-
eral conceptions of the ultimate purposes of the great, grand human
adventure (progress, justice, self-actualization), and so on’ (p. 385). It is
no accident that conceptions of world culture elements such as Boli’s
often occur at a high level of abstraction and generality. As we shall
see, World Culture theory becomes increasingly difficult to apply as
policy moves along the continuum of implementation. Nonetheless,
proponents of World Culture theory believe not only that world cul-
ture exists, but also that it has become more salient in recent decades,
is more ubiquitous, is becoming more organised and rationalised, and is
indeed leading to the homogenisation of ‘life experience, outlook, and
intensions of individuals everywhere to an extraordinary degree’ (Boli,
2005, p. 388).
Within the field of education, there is a view that modern education

has become ‘a world enterprise, universal and universalistic in aspiration
and in some measures of outcome’ (Meyer and Ramirez, 2003, p. 115).
Astiz et al. (2002) attribute world culture to economic globalisation, and
the ‘hegemonic policy discourse’ created by the spread of neoliberal
thought; they conclude from their study of the mathematics curriculum
in 39 countries that ‘countries have changed their structure of curricular
governance to become more decentralised overall’ (pp. 69, 72). On their
part, Wiseman et al. (2011) believe that there has been a ‘massifica-
tion’ and a ‘normalisation (i.e. isomorphism) of school structure and
content – including curricula’ (p. 562). Rauner’s (1998) cross-national
study of trends in civics education between 1955 and 1995 has been
cited as demonstrating a shift ‘away from more particular national civics
material to material that presents citizenship to students as a global pro-
cess’ (Wiseman et al., 2011, p. 565, original emphasis). More recently,
using data from the 1999 Civic Education Study, the World Bank’s World
Development Indicators and Freedom House’s national political system
indictors, Wiseman et al. (2011) find ‘relatively weak effects of national
characteristics on political socialisation’, and suggest that this is ‘consistent
with an institutional argument regarding the formation of a supra-national
citizenry’ (p. 547, original emphasis).
Wiseman et al. are not the only scholars go beyond the level of
policy discourse and education structure to claim convergence at the
level of curriculum content. McEneaney and Meyer (2000) argue that
there are similar ‘educational and curricular forms’ regardless of the
stage of economic development or traditional culture (p. 195); they
claim, for instance, that citizenship education tends now to reflect the
influence of liberalism, and increasingly focuses on the ‘broader prin-
ciples of human rights and responsibility, (and) . . . democratic political
structures’(p. 197). Ramirez et al. (2006) similarly claim that there has
increasingly been the inclusion of human rights in the curriculum due
to the influence of models from world society and that traditional forms
of citizenship education that give ‘heavy attention to local and national
politics and their values and procedures’ now appear ‘limited or even
jingoistic’ (p. 38). A further claim by World Culture theorists is that the
rise of a neoliberal discourse in the 1980s and 1990s has led, among
other things, to the emergence of ‘a model of decentralised educational
governance’; the result is that, notwithstanding variations in adapta-
tion and implementation at the local level, ‘countries have changed
their structure of curricular governance to become more decentralised
overall’ (Astiz et al., 2002, pp. 70, 72).
The view of the convergence of education systems and curriculum

content has been contested by a number of scholars. Carney et al. (2012)
make the pertinent point that World Culture research tends to focus on
policy development and assign a high degree of consensus to this, while
downplaying ‘local change processes’; they also note that opponents of
World Culture theory tend to focus precisely on these ‘local variations’.
On their part, Jakobi and Tletemann (2011) demonstrate in their study
of policy using data from the OECD and UNESCO that whether there is
convergence depends on the level of analysis: while there may be a con-
vergence at macro-level, education policies closer to implementation
efforts tend to ‘show strict path dependencies and continuing persis-
tence’ (p. 591). Green and Mostafa, in their contribution to the current
volume, make a similar point in stating that the ‘elevated point of view’
of World Culture theorists prevents them from seeing significant cross-
national variation in policies and practices on the ground. Indeed, in
their examination of the social studies curriculum in the light of claims
of the global spread of social studies, Morris et al. (1997) observe that:
Worldwide trends can provide both rhetoric and models for specific
sorts of policy changes. At a micro level, however, conflict or compe-
tition among subgoups can modify or transform proposed changes,
and the adoption and implementation of the changes are determined
by a range of pragmatic considerations within schools. (p. 43)
Whatever the case may be, if the proponents of World Culture theory are
correct, we would expect to find patterns of curriculum with distinctive
supra-national characteristics.
Alternative perspectives on system change
The modernisation perspective

There are two main alternative perspectives to World Culture theory.
The first of these can be labelled the ‘modernisation’ perspective because
its central notion is that, as countries modernise, they follow a singular
and linear path of development, passing through several stages in this
process – agricultural, industrial and post-industrial. Variation across
states – in the structure of the economy and the nature of the polit-
ical system – is predicted and interpreted as variation along this path
of development. Thus countries with different economic and political
systems are seen to be at different stages of development. The fac-
tor that scholars working within the modernisation tradition – such
as Dalton (2004) and Inglehart (1990, 1997) – have in common with

World Culture theorists is that they discern a broad trend away from
authoritarian government towards more democratic forms. However,
the former differ from the latter in ignoring the role of transnational
agents, processes and discourses. They see a country’s shift towards
an ever more effective and participatory democracy exclusively as a
function of domestic socio-economic development.
For modernisation theorists, it would be unthinkable for a country
to maintain a system of government that is not in line with its stage
of socio-economic development, but for World Culture theorists it is
entirely possible for a country in, say, the agricultural stage of develop-
ment to adopt, at least at a formal level, all the attributes of a liberal
democracy. Among other things, this means that, in the modernisa-
tion perspective, countries do not necessarily converge, at least not in
the short run. If the laggards are moving more slowly than the fron-
trunners, all will drift further apart, even though they all move in
the same direction. In the long run, convergence is predicted by some
strands of modernisation theory, but whether this happens depends on
whether a definite end point of development is postulated – for exam-
ple, by Fukuyama (1993) with his ‘End of History’ argument – and on
whether progress along a singular trajectory of development is seen as
being inevitable. Only if a final stage is assumed to exist, and coun-
tries are seen as necessarily progressing and not standing still or going
into reverse, is convergence towards some ultimate outcome (e.g. a
post-industrial, affluent, liberal-democratic market economy) the logical
theoretical expectation.
Due to the work of Inglehart on the rise of post-materialism, the mod-
ernisation perspective has been influential in research on social and
political values. However, it has not received the same attention among
scholars examining public policy, despite some of the latter explicitly
mentioning the possibility of linear development without convergence.
Describing broad trends in education policies, for instance, Green et al.
(1999) remark that, while policies may head in the same direction, these
can start from ‘different points and/or change at different rates, so that
there is little actual convergence between two points in time’ (p. 60).
Similarly, Mahnig and Wimmer (2000), exploring immigration policy,
explain that confusion often occurs between the concept of conver-
gence and that of parallel development: the latter implies ‘that individual
countries go through identical stages in developing their immigration
policies without reaching the same point’ (p. 179). The disregard for the
modernisation perspective in policy studies is all the more surprising in
view of the strong empirical evidence that Koopmans et al. (2005) find
for it in their study of citizenship policies. Examining developments of
these policies, they find that, although all countries in their five country
study move in the same direction (towards more recognition of cultural
difference and greater individual equality of access), they had different
points of departure and did not converge in the process.
There is one perspective that does apply the key modernisation notion
of linear development very explicitly to an area of public policy. This is
the evolutionary perceptive advocated by Taras Kuzio (2002). Accord-
ing to Kuzio, states will gradually abandon ethnic exclusionary policies
on immigration and citizenship, and exchange these for civic inclusive
policies, as they mature and their democracies consolidate. Thus, while
Kuzio differs from Dalton and Inglehart in assigning prime causal sig-
nificance to the age of the state and democratic tradition instead of
socio-economic development, his view could be considered to belong
to the modernisation perspective in that it is inspired by the same logic
of linear and staged evolution.
For present purposes, an implication of the modernisation perspective
is that the education system of countries at different stages of economic
development, state formation or democratic consolidation would pro-
mote different views and values in order to ensure that their population
have the qualities suited for that – and perhaps the next – stage of
development. Conversely, countries at the same stage of development
would tend to have similar education systems generally, and curricula
specifically.
The qualitative differences perspective
The second alternative perspective that we introduce here is the ‘qual-

itative differences’ perspective. The term is formulated in the plural
because the scholars that we associate with this perspective agree on
little more than the basic proposition that countries are neither converg-
ing nor moving in the same direction, but show important qualitative
differences in their education systems. Thus, we can identify a cul-
turalist school of thought that sees cultures as completely exogenous,
almost immutable, entities fundamentally shaping values, attitudes and
behaviour (e.g. Huntington, 1996; Harrison, 2000). In this view, educa-
tion is seen as a core institution of society and a prime expression of
culture, and education systems as being distinctive and enduring. Edu-
cation systems therefore differ across countries where there are uniquely
evolved cultures rooted in different historical and religious traditions;
by the same token, these would be similar across societies that share the
same culture.
Within the qualitative differences perspective, a second school of
thought has been proposed by a group of scholars who, while sharing
the assumption of qualitative differences with the culturalists, argue that
a country’s education system should be understood as part of that coun-
try’s wider societal ‘regime’ of social and institutional characteristics.
As noted in the introduction, such a regime is seen as the product of the
struggle between various interest groups and the particular political sys-
tem of a country and is generally considered ‘mid-range’ in terms of its
persistence. The concept of regimes is used to describe qualitatively dif-
ferent forms of social cohesion (Green et al., 2009; Green and Janmaat,
2011) and welfare states (Esping-Andersen, 1990). According to Green
et al. (2009), regimes of social cohesion are ‘relatively durable (but not
immutable) configurations of social attitudes and behaviours contribut-
ing to society-wide social bonding that are underpinned by particular
institutional arrangements’ (p. 63). Using a combination of a qualita-
tive analysis of the institutions and social attributes in the literature,
and statistical analysis of cross-national data on institutional character-
istics and social behaviours and attitudes, Green et al. (2009) identify
three regimes of social cohesion – the liberal regime associated with
the English-speaking countries, the social market regime which prevails
in the original six European Union countries and the social democratic
regime which is found in the Scandinavian countries. They also accept
the possibility of a Confucian social cohesion regime. They further pos-
tulate that these regimes would be undergirded by particular sets of
values and attitudes, as well as institutions such as the education system.
Differences in the field of education have been remarked on by those
working in citizenship education, among others. Kerr (1999) observes
that there is a continuum with respect to values in citizenship educa-
tion. At one end – in countries like Canada, England and the United
States – there is a ‘commitment to pluralism’ and ‘minimal reference
to values in education in educational legislation’; and, at the other
end – in countries like Japan, Korea, and Singapore – ‘national values
(are) expressed in detail’ (Kerr, 1999, p. 6). Kerr (1999) also refers to
the distinction between ‘values-explicit’ approaches that are ‘commonly
criticised for the associated dangers of bias and the indoctrination of stu-
dents’ and ‘values-neutral’ approaches that are ‘attacked for their failure
to help students to deal adequately with real-life, controversial issues’
(p. 7). Indeed, in their study of curricula in East and Southeast Asia,
Morris and Marsh (1992) identify a set of characteristics distinctive to
the region; in addition to a centralised education system, central control

of textbooks and the use of civics to promote national identity and cohe-
sion, they also find the prioritisation of ‘communal and social values
and needs . . . over individual needs or aspirations’ (p. 252).
Kerr’s distinction between the pluralistic and values-neutral approach,
and the national values and values-explicit approach, share similarities
with Crick and Porter’s distinction between procedural and substan-
tive values. For Crick and Porter, the former comprises values like
freedom, toleration, fairness, respect for truth and respect for reason-
ing (Crick, 1978, pp. 66ff), and are ‘rules for civilised procedures’ (Crick
and Porter, 1978, p. 31). In other words, these values are procedural in
that they inform the process of debate, and do not impinge directly
on the value or topic in question. The notion of procedural values was
developed by educationists – and adopted by practitioners – in Western,
liberal societies as a way to avoid the thorny issue as to the substantive
values that schools should promote in an increasingly multicultural and
diverse environment.
It should be noted that procedural values – such as respect for persons
and differences of opinions – are either ultimately substantive or rest
on substantive values (e.g. equality of individuals). However, in many
Western societies, there appears to be a tacit agreement that these val-
ues are sufficiently widely accepted to be bracketed out of values debates
so as to form the basis of the procedure for these debates. It is possible
therefore that such procedural values can be so highly regarded in cer-
tain societies that these become actively promoted as substantive values.
The three perspectives outlined above embody very different views of
the nature of education systems and their constituent curricula. World
Culture theory argues explicitly that systems are converging towards a
global or Western model emphasising human rights and democracy.
The modernisation perspective predicts a high degree of commonal-
ity between the systems of nations at a similar level of economic
development. The qualitative differences perspective suggests significant
variations across countries, or groups of countries, based on immutable
cultures (as argued by culturalists) or certain social cohesion regimes
(as argued by regime theorists). Our study is well placed to address these
perspectives as it examines both the design of the curriculum and the
values promoted in it, and does so for 16 countries. It is distinctive in
both its ability to explore the regional patterns predicted by the regimes
perspective, that is patterns that transcend national boundaries but do
not constitute global patterns, and it looks beyond general, superficial
descriptors at a high level of abstraction.
Methodological approach
Between 2009 and 2011, a team from the ESRC-funded Centre for Learn-
ing and Life Chances in Knowledge Economies and Societies (LLAKES)
carried out a project to investigate whether curriculum patterns could
be identified for citizenship education across countries and, if so, how
these patterns could be described and explained. Our study focuses on
the level between state policy and policy implementation and on the
specific following elements: state control of the curriculum and the val-
ues promoted through the curriculum. In addition, we include countries
from several regions, which allows for comparative analysis.
Scope of the study: Our choice of countries was informed by the four
regimes of social cohesion identified by Green et al. (2009). Thus,
in order to ensure that the selection included representatives of all
four regimes, we chose Denmark, Finland, England, France, Germany,
Hungary, Italy, Spain and Sweden (representing the social cohesion
regimes found in Europe, including the liberal regime); Japan, Singapore
and South Korea (representing the East Asia or Confucian regime) and
Australia, Canada, New Zealand and the United States (representing
the non-European, English-speaking, countries belonging to the liberal
regime). A pragmatic consideration was that comparable data on these
countries was available from sources such as INCA and EURYDICE.2
Years of the study: We chose the years of formal or compulsory school-
ing for the study, as these would be most likely to reflect government
policy where values are concerned. We considered the junior secondary
curriculum (12+–14+) to be more suitable for analysis, in part because
the primary curriculum would lack sophistication in terms of the values
being promoted, and also because the junior secondary level was more
likely than the upper secondary level to be compulsory, de facto if not
de jure.
Time frame: In order to be up to date, we looked at the education
curriculum that was being practised in the selected counties at the time
of the data collection.
Indicators: In view of the claims of the three theoretical perspectives
explained above, the research team decided to examine two dimen-
sions of national citizenship education curricula: (1) state control of the
curriculum, and (2) the type of values promoted. According to World
Culture theory, we would expect to find decentralised forms of educa-
tion governance (with schools having autonomy in curriculum matters),
and a similarity in the curricula (e.g. an emphasis on human rights,
democracy, and individual autonomy). In contrast, the modernisation
perspective would lead us to expect the curriculum on the two dimen-

sions to vary according to the indices of socio-economic development
and democratic consolidation: the more developed a state and the
longer its tradition of democracy, the more decentralised the curriculum
and the greater the emphasis on democracy, universalism and individual
autonomy. On its part, the qualitative differences perspective predicts
major differences across cultures or regimes which are unrelated or not
limited to stages of economic or political development.
The state control composite indicator is composed of five indica-
tors which were populated using data from existing sources, namely
INCA and EURYDICE (see Figure 5.1). It represents a summative scale
with values ranging from 0 (no compulsory status, no hours specified,
no national curriculum, no state control of textbooks) to 10 (compul-
sory status, minimum number of hours specified, a national curriculum,
textbooks produced by the state). The higher the value, the greater the
degree of state control. The value 5 represents the midpoint, denoting
neither strong nor weak state control. Reliability analysis showed the
composite indicator to have a high internal coherence (alpha of .8),
meaning that all five indicators composing the scale are strongly inter-
correlated. The advantage of a summative scale with a clear midpoint
is that it expresses absolute values in addition to relative ones. It is
important that a scale provides information on both types of values.
For instance, if France has the highest value on state control of all 16
states, but its absolute value is only 3 on a scale from 0 to 10, then
clearly the absolute value shows a low degree of state control overall, no
matter where it stands relative to the other states.
The second composite indicator taps the values promoted in the
curriculum (see Figure 5.2). We adopted Crick and Porter’s (1978) dis-
tinction between procedural and substantive values for one of the
constituent indicators. We also made distinctions between the promo-
tion of moral autonomy and critical thinking on the one hand, and
substantive values on the other, between a focus on the individual and
the collective, between the promotion of a civic or an ethnocultural
identity, and between an emphasis on diversity or homogeneity. As the
countries scoring high on these constituent indicators emphasise either
the collective or characteristics of the curriculum associated with pro-
moting this, we used the term ‘collective values indicator’ for ease of
reference.
The constituent values indicators were original and, to our knowl-
edge, had not been used in previous studies. We felt that these would be
helpful in highlighting the differences in emphasis in values education
126
1. Civics/citizenship education: compulsory or non-compulsory (citedcomp)

1 – Civics/citizenship education is not compulsory
2 – Cross-curricular guidelines is provided for civics/citizenship education
3 – Civics/citizenship education is embedded in other subjects
4 – The subject of civics/citizenship education (or social studies) is offered
5 – The subject civics/citizenship education (or social studies) is statutory and
compulsory
A five point Likert scale was used, with 1 indicating the lowest level state control
with respect to civics/citizenship education, and 5 the highest.
2. Moral education: compulsory or non-compulsory (moredcomp)
1 – Moral education is not compulsory
2 – Cross-curricular guidelines are provided for moral education
3 – Moral education is embedded in other subjects
4 – The subject moral education is offered
5 – The subject moral education is statutory and compulsory
with respect to moral education, and 5 the highest.
3. Civics/citizenship education: hours of study specified (citedhours)
1 – There are no hours/days specified
3 – A minimum number of hours/days is suggested
5 – A minimum number of hours/days is specified and mandated
A three point Likert scale was used, with 1 indicating the lowest level of state
control with respect to hours of study for civics/citizenship education, and 5 the
highest.
4. State control of curriculum/syllabus (statecontrcurr)
1 – There is no national curriculum (although there may be a curriculum at state
or province level)
2 – There is state/community level curricula with no common areas
3 – There is state/community level curricula with some common areas
4 – There is a broad national curriculum framework or core curriculum, with
non-mandatory guidelines
5 – There is a national curriculum
with respect to the curriculum/syllabus, and 5 the highest.
5. State approval of textbooks (stateapprtextbooks)
1 – There is no state control of textbooks
2 – A recommended list is produced, but textbooks that are not on the list can
be used
3 – Textbooks are approved by the state
4 – Textbooks are approved, and some produced, by the state
5 – All textbooks are produced by the state
with respect to textbooks, and 5 the highest.
Figure 5.1 The composite state control indicator

127
1. Procedural or substantive values

1. Primarily procedural values, with minimal substantive values
2. Mainly procedural, with some substantive values
3. A mixture of procedural and substantive values
4. Mainly substantive, with some procedural values
5. Primarily substantive (involving a single, clear moral code), with minimal
procedural values
A five point Likert scale was used in which 1 represented the promotion of
mainly procedural values, and 5 the promotion of mainly substantive values.
2. Moral autonomy and critical thinking, or substantive values (personal, social
and political)
1. Primary emphasis on moral autonomy and critical thinking, with minimal
emphasis on substantive values
2. Main emphasis on moral autonomy and critical thinking, with some
emphasis on substantive values
3. No overall emphasis on one type of values over another
4. Main emphasis on substantive values, with some emphasis on moral
autonomy and critical thinking
5. Primary emphasis on substantive values, with minimal emphasis on moral
autonomy and critical thinking
A five point Likert scale was used in which 1 represented a primary emphasis
on moral autonomy and critical thinking, and 5 the primary emphasis on
substantive values.
3. Focus on the individual or the collective
1. Primary emphasis on the individual, with minimal emphasis on the
collective
2. Main emphasis on the individual, with some emphasis on the collective
3. Equal emphasis on the individual and the collective
4. Main emphasis on the collective, with some emphasis on the individual
5. Primary emphasis on the collective, with minimal emphasis on the
individual
on the individual, and 5 the primary emphasis on the collective.
4. Focus on a civic identity or an ethnocultural identity
1. Primary emphasis on civic identity, with minimal emphasis on ethnocultural
identity
2. Main emphasis on civic identity, with some emphasis on ethnocultural
identity
3. Equal emphasis on civic and ethnocultural identity
4. Main emphasis on ethnocultural identity, with some emphasis on civic
identity
5. Primary emphasis on ethnocultural identity, with minimal emphasis on civic
identity
on civic identity, and 5 the primary emphasis on ethnocultural identity.
Figure 5.2 The composite collective values indicator

5. Emphasis on diversity or homogeneity

1. Primary emphasis on diversity and multiculturalism, with minimal
emphasis on a common/homogeneous national identity
2. Main emphasis on diversity and multiculturalism, with some emphasis on a
common/homogeneous national identity
3. Equal/no emphasis on diversity and multiculturalism, and a
common/homogeneous national identity
4. Main emphasis on a common/homogeneous national identity, with some
emphasis on diversity and multiculturalism
5. Primary emphasis on a common/homogeneous national identity, with
minimal emphasis on diversity and multiculturalism
on diversity and multiculturalism, and 5 a primary emphasis on a
common/homogeneous national identity.
Figure 5.2 (Continued)
in different country groupings; they would also cast light on the nature
of such differences. Having said that, the values indicators could not be
readily populated from extant data sources, and a new approach had
to be found. The novelty was that we were using qualitative data and
quantifying these to make them comparable statistically.3
The populating of the values indicators was not straightforward in
that what was required was not objective data, for example a certain
number of subjects, or number of hours of study; rather, the process
of obtaining a figure on a Likert scale for these indicators required the
judgement of experts familiar with the curriculum – specifically that
relating to values education – of the country in question. For this, we
took two approaches. First, we convened two country expert panels
on the countries in the study – one on Europe and the other on Asia.
However, since panels could only be convened of country experts who
happened to live in or near London where the LLAKES research centre
was located, we had to resort to another approach for the other coun-
try experts. The research team identified experts for these countries, and
sent them a copy of the indicators, with an explanation and an exam-
ple of how these could be completed. For reasons of triangulation, we
required at least two country experts for each country (either on the
panels or written to individually).
As with the indicators of state control, the values indicators were
highly interrelated: states that stressed substantive values empha-
sised the collective, ethnocultural identities and homogeneity in the
citizenship curriculum; conversely, states that put emphasis on proce-

dural values stressed critical thinking and autonomy, individual rights,
civic identities and multiculturalism (reliability analysis produced an
alpha of .75 indicating a high level of internal coherence). Thus, as
with the state control indicators, we created a composite values indi-
cator representing a summative scale with values ranging from 0 to 10,
with 5 being the neutral midpoint. The higher the value on the compos-
ite values indicator, the more a country stresses ethnocultural identity,
homogeneity, and substantive and collective values.
Findings
Figure 5.3 shows the country scores on both composite indicators.

We see that there is considerable variation on both indicators. Some
countries have values far below the midpoint (including all the
Japan
8.00
7.00
Italy South Korea
Singapore
6.00
Collective values
Hungary
5.00
Denmark Sweden France
Spain Finland
4.00
Germany
United States
Canada Australia
3.00
New Zealand
England
2.00 R2 Linear = 0.583
0.00 2.00 4.00 6.00 8.00 10.00

State control
Figure 5.3 The relation between state control and collective values
English-speaking countries), while others have values approaching the

maximum of both scales (Japan and, to a lesser extent, Singapore and
South Korea). Thus, the East Asian states exercise firm central control
over the curriculum and promote substantive and collective values, and
homogeneity, while the English-speaking countries do the opposite,
giving schools autonomy in the curriculum, and endorsing moral auton-
omy and critical thinking, individualism and procedural values. We can
further see that the two composite indicators are closely linked: the
higher the degree of state control, the more the emphasis on collective
values (r = .76; p = .001; N = 16). The East Asian countries cluster quite
tightly at the top of both indicators (i.e. in the upper-right corner of
Figure 5.3), while the English-speaking states do so at the bottom (in the
lower-left corner). The continental European states assume middling
positions around the mid-point of both indicators, although France
shows quite a high value for state control, while Italy does so for collec-
tive values. Roughly speaking, we can therefore identify three clusters of
countries: an English-speaking one with low values on both dimensions,
an East Asian one with high values on both dimensions and a continen-
tal European cluster with medium positions on the two dimensions. The
last cluster forms the least cohesive group.
We also ran a series of correlations to test whether state control and
collective values were related to socio-economic development, the age of
the state and democratic consolidation. This would verify whether states
demonstrated the patterns expected by the two schools of thought –
as expressed by Inglehart and Kuzio respectively – that are associated
with the modernisation perspective. Socio-economic development was
measured with GDP per capita Purchasing Power Parity, taken from the
CIA World Fact Book 2000. The age of the state was measured as the
number of years of state independence until 2000. Democratic consol-
idation was measured as number of years of uninterrupted democracy
from the introduction of universal suffrage until 2000.
Table 5.1 shows that the two curriculum indicators are strongly related
with democratic consolidation, but not with the age of the state, or the
level of socio-economic development. The correlation with democratic
consolidation is negative, indicating that states with a long history of
uninterrupted democratic rule tend to have a decentralised curriculum
and to promote individualism rather than collectivism. The scatter plots
of Figures 5.4 and 5.5 provide more detailed information regarding the
country positions on this negative correlation. We can see the familiar
clusters of countries appearing: the East Asian states group (excepting
Japan to some extent) gather in the upper left corner, combining low
Table 5.1 Correlations of state control and collective values with indicators of
modernisation.
State control Collective values N
Age of the state .031 −.019 16

Socio-economic development −.111 −.083 16
Democratic consolidation −.597∗ −.650∗∗ 16
∗, Correlation is significant at the 0.05 level (2-tailed).

∗∗ , Correlation is significant at the 0.01 level (2 tailed).
10.00 Japan
South Korea
Singapore
8.00 France
Finland
State control
6.00
Hungary
Sweden
Spain Germany
4.00
Italy England
Denmark New Zealand
2.00 United States Australia
Canada
0.00 R2 Linear = 0.352
0.00 20.00 40.00 60.00 80.00 100.00 120.00

Democratic consolidation
Figure 5.4 The relation between democratic consolidation and state control
values on democratic consolidation with high values on state control

and collective values, and the English-speaking countries converge in
the lower right corner showing the exact opposite pattern. Japan is
somewhat of an outlier demonstrating ‘unexpectedly high’ values on
the two curriculum indicators from the point of view of Kuzio’s theory.4
Discussion
As noted earlier, World Culture theory predicts convergence towards

decentralised education structures, and a curriculum emphasising
8.00 Japan
7.00
Singapore
Italy
South Korea
Collective values
6.00
Hungary
5.00
France Denmark
Spain Germany Sweden Finland

4.00
United States
Canada New Zealand

3.00
England Australia
2.00 R2 Linear = 0.403
0.00 20.00 40.00 60.00 80.00 100.00 120.00

Democratic consolidation
Figure 5.5 The relation between democratic consolidation and collective values
democracy, human rights and individual autonomy. As Jakobi and

Tletemann (2011) and Morris et al. (1997) suggest, it is entirely possi-
ble that at the most general and abstract level – at the level of policy
pronouncements, for instance – there might be convergence where edu-
cation is concerned; however, there may not be convergence at the level
of adoption and implementation. It is at this level that our data and
analyses are situated. We recognise that the demonstration of conver-
gence would require data over time, whereas our data merely represents
a snapshot. However, our data can nonetheless address the World Cul-
ture claim of convergence by exposing the sheer variation in the two
curriculum indicators in absolute terms. If World Culture theory were
correct, we should not see countries – particularly those which are well
integrated in the world economy and in transnational organisations –
with high degrees of state control, and a heavy emphasis on collective
values. All such states should have values below the midpoint of the
two scales. Yet we did find countries with such high absolute values –
the East Asian states – and these countries contrasted markedly with the
English-speaking countries. In addition, we found two European states
displaying quite high values on state control (France) and collective
values (Italy). These findings are thus difficult to reconcile with the
World Culture perspective.
The modernisation perspective predicts that countries develop inde-
pendently but that, while they may be at different stages of devel-
opment, they are all heading in the same direction, towards more
decentralised structures, and a curriculum endorsing individualism and
universal values. Of our three indicators of development, we found only
democratic consolidation to show the expected relationship with the
two curriculum indicators: states with a long tradition of democracy
tend to have a decentralised curriculum and to promote individual-
ism. The other two indicators – the level of prosperity, and the age of
the state – did not show any relationship with our two outcomes of
interest. The support for the modernisation thesis is thus only very par-
tial. In particular, the fact that the two curriculum indicators do not
appear to be related to socio-economic development (as indicated by
the level of prosperity) is striking, as this condition is seen as so signif-
icant in shaping the world of ideas by mainstream scholars working in
the modernisation tradition.
This non-relationship is nicely illustrated by the case of Japan. Japan’s
place in economic development would, it is reasonable to say, be ranked
alongside those of many post-Second World War European economies.
Yet, unlike those countries which display low state control and high lib-
eral values, Japan is high on both state control and collective values.
Indeed, where our data is interesting, and persuasive, is that it shows
countries at different points of development clustering together. Japan
is located within the East Asian cluster with Singapore and South Korea,
all of which are high on state control and collective values. However, the
common understanding is that these countries are at different stages
of economic development, with Japan being at the forefront in the
post-war boom, followed a decade or two later by the ‘first tier newly
industrialising economies’, of which Singapore and South Korea are con-
sidered members (Kasahara, 2004, p. 1). This differing rate of economic
development even has been captured in the flying geese paradigm (see,
for instance, Kasahara, 2004). According to the modernisation perspec-
tive, Japan should be further along with respect to modernisation in
the form of social and economic development and would therefore be
expected to be closer to other developed countries, such as those in the
West. However, it can be seen in Figure 5.1 that Japan occupies an even
more extreme position than Singapore and South Korea on both the
state control and collective values axes. This would suggest that neither
the modernisation nor World Culture theories can adequately explain

the patterns identified in our study.
Part of the problem with the modernisation perspective is the lack
of clarity and consensus with respect to what this involves. Does it just
involve levels of prosperity as measured by per capita income, and stages
of economic development (agrarian, industrial, and post-industrial), or
should it also include democratisation? If the latter, does it have to be
the kind of liberal democracy associated with the West? Indeed, the
monolithic view of democracy has been challenged by Chan (1993) and
Huntington (1993) who suggest that there are not only other forms of
democracy, but also other democratic ideals. Chan (1993) has argued
for Asian forms of democracy, and noted the characteristics that such
democracies – including Japan, Singapore and South Korea – have in
common, namely communitarian values, respect for authority and hier-
archy, a dominant political party, a centralised bureaucracy and a strong
state. If this is the case, then the clustering of Japan, Singapore and
South Korea would be indicative of an Asian form of democracy which
would, in turn, be underpinned by institutions such as an education
system that is designed to support it. With regard to the English-
speaking states, the argument can be made that their democracies reflect
a particular kind of political culture, one that emphasises individual
opportunity, pluralism, civil society and minimal state intervention. It is
perhaps only because of the dominance in political discourse of their
model of liberal democracy that the cultural specificity of its origin is
obscured.
Of the three theories put forward at the beginning of the chapter
therefore, the qualitative differences perspective would appear to have
the strongest explanatory power, with the differences in the curricu-
lum being explained by the East Asian cultural heritage shared by these
societies. And, to the extent that democratic tradition is indicative of a
particular culture or regime, the results suggest that variations in state
control and collective values are not so much a function of modernisa-
tion and socio-economic conditions as a reflection of distinct cultures,
traditions and/or regimes.
Finally, it should be noted that a clear limitation of the current study
is that it has not explored the effects of the two dimensions of curricu-
lum differentiation on student attitudes and behaviours. Obviously, if
it is demonstrated that the impact of the curriculum is marginal, the
question would arise as to the relevance of examining cross-national
differences in the citizenship curriculum. However, the contributions in
the current volume by Mons, Duru-Bellat and Savina and by Janmaat
have identified clear links between the content of the curriculum and
the structure of schooling more broadly, on the one hand, and a range of
attitudinal outcomes, on the other These links at the very least suggest a
distinct influence of schooling, in addition to that of other socialisation
agents, and thus highlight the importance of examining cross-national
variation in and dynamics of citizenship education.
Acknowledgements
The authors would like to thank the following country experts and
organisations for their contribution and assistance:
Hermann Josef Abs, Ildikó Bárdossy, Gerard Bonnet, John J. Cogan,
Ann Doyle, Daniel Faas, Yvonne Hebert, Mary Hill, Bryony Hoskins,
Laura Johnson, Kaori Kitagawa, Henry Maitles, Concepción Maiztegui
Oñate, Colin Marsh, Ference Marton, Marcella Milana, Park Joo Hyun,
Lynne Parmenter, David Pepper, Nelli Piattoeva, T. J. Scott, Alan
Sears, Jasmine Sim, Judit Szivák, Susanne Wiborg, and John Williamson,
and staff at the Finnish National Board of Education, the Japanese
Ministry of Education, Culture, Sports, Science and Technology, the
Korea Institute for Curriculum and Evaluation, Skolverket (the Swedish
National Agency for Education), the National Institute for Education
Policy Research of Japan, the New Zealand Ministry of Education,
and (what was then) the Qualifications and Curriculum Development
Agency (UK).
We are further grateful to the ESRC for providing the funding that
made the research for this paper possible (Grant RES-594–28–0001). The
research for this paper is part of the wider research agenda of LLAKES
(the Centre for Learning and Life Chances in Knowledge Economies and
Societies).
Notes
1. We collected data for this purpose between 2009 and 2011. This data collec-
tion was part of a larger project on regional patterns in citizenship education,
carried out by a team from the ESRC-funded Centre for Learning and Life
Chances in Knowledge Economies and Societies (LLAKES).
2. The sources are the International Review of Curriculum and Assessment
Frameworks Internet Archive (INCA) database (then managed by the Qualifi-
cations and Curriculum Development Agency, and currently by the Depart-
ment for Education, http://www.inca.org.uk/ (home page, date accessed
9 August 2012), and the Information on Education Systems and Poli-
cies in Europe (Eurydice Network on education systems and policies in
Europe (http://eacea.ec.europa.eu/education/eurydice/index_en.php (home

page, date accessed 9 August 2012).
3. In the area of migration studies, the researchers responsible for the Migrant
Integration Policy Index (MIPEX) have used a methodological approach that
is similar to the one we devised (see Migrant Integration Policy Index, 2012).
4. We recognise that democracy in Japan may represent a special case as it has
largely been a one party state since the Second World War. The indicator used
here is merely about length of democracy and does not address the form of
democracy.
References
Astiz, M.F., Wiseman, A.W., and Baker, D.P. (2002) ‘Slouching towards Decen-
tralisation: Consequences of Globalisation for Curricular Control in National
Education Systems’. Comparative Education Review, 46, 1, pp. 66–88.
Benavot, A., Cha, Y.-K., Kamens, D., Meyer, J.W., and Wong, S.Y. (1991) ‘Knowl-
edge for the Masses: World Models and National Curricula, 1920–1986’.
American Sociological Review, 56, 1, pp. 85–100.
Boli, J. (2005) ‘Contemporary Developments in World Culture’. International
Journal of Comparative Sociology, 46, pp. 383–404.
Crick, B. (1978) ‘Procedural Values in Political Education’ in B. Crick and A. Porter
(eds) Political Education and Political Literacy. London: Longman, pp. 63–72.
Crick, B. and Porter, A. (1978) ‘The Programme’s Working Papers’ in B. Crick
and A. Porter (eds) Political Education and Political Literacy. London: Longman,
pp. 30–36.
Culture Theory and Comparative Education’. Comparative Education Review, 56,
3, pp. 366–393.
Chan, H.C. (1993) ‘Democracy: Evolution and Implementation’ in R. Bartley,
H.C. Chan, S. Huntington, and S. Ogata (eds) Democracy and Capitalism:
Asian and American Perspectives. Singapore: Institute of Southeast Asian Studies,
pp. 1–26.
Dalton, R.J. (2004) Democratic Challenges, Democratic Choices: The Erosion of
Political Support in Advanced Industrial Democracies. Oxford: Oxford University
Press.
Esping-Andersen, G. (1990) The Three Worlds of Welfare Capitalism, Princeton. New
Jersey: Princeton University Press.
Eurydice (2005) Citizenship Education at School in Europe. Brussels: Eurydice.
Frank, D.J., Wong, S.Y., Meyer, J.W., and Ramirez, F.O. (2000) ‘What Counts as
History: A Cross-National and Longitudinal Study of University Curricula’.
Comparative Education Review, 44, 1, pp. 29–53.
Fukuyama, F. (1993). The End of History and the Last Man. London: Penguin.
Green, A. and Janmaat, J.G. (2011) Regimes of Social Cohesion: Societies and the
Crisis of Globalisation. Basingstoke: Palgrave Macmillan.
Green, A., Janmaat, J.G. and Han, C. (2009) Regimes of Social Cohe-
sion, London: Centre for Learning and Life Chances in Knowledge Economies
and Societies, http://www.llakes.org/wp-content/uploads/2010/08/Z.-Regimes-
of-Social-Cohesion.pdf, date accessed 2 August 2012.
Green, A., Wolf, A. and Leney, T. (1999) Convergences and Divergences in European
Education and Training Systems. London: Bedford Way Papers, Institute of
Education.
Harrison, L.E. (2000). ‘Why Culture Matters’ in L.E. Harrison and S. Huntington
(eds) Culture Matters: How Values Shape Human Progress. New York: Basic Books,
xvii–xxxiv.
Huntington, S. (1993) ‘American Democracy in Relation to Asia’ in R. Bartley,
H.C. Chan, S. Huntington, and S. Ogata (eds) Democracy and Capitalism:
Asian and American Perspectives. Singapore: Institute of Southeast Asian Studies,
pp. 27–43.
Huntington, S. (1996) The Clash of Civilizations and the Remaking of the World
Order. New York: Simon and Schuster.
Inglehart, R. (1990) Culture Shift in Advanced Industrial Society. Princeton, NJ:
Inglehart, R. (1997) Modernization and Postmodernization: Cultural, Economic
and Political Change in 43 Societies. Princeton, NJ: Princeton University
Press.
Jakobi, A.P. and Tletemann, J. (2011) ‘Convergence in Education Policy? A Quan-
titative Analysis of Policy Change and Stability in OECD Countries’ Compare,
41, 5, pp. 579–595.
Kasahara, S. (2004) The Flying Geese Paradigm: A Critical Study of its Application
to East Asian Regional Development, Discussion Paper No. 169. Geneva: United
Nations Conference on Trade and Development.
Kerr, D. (1999) ‘Citizenship Education: An International Comparison’. http://
www.inca.org.uk/pdf/citizenship_no_intro.pdf, date accessed 2 August 2012.
Koopmans, R., Statham, P., Giugni, M., and Passy, F. (2005) Contested Citizen-
ship: Immigration and Cultural Diversity in Europe. Minneapolis: University of
Minnesota Press.
Kuzio, T. (2002) ‘The Myth of the Civic State: A Critical Survey of Hans Kohn’s
Framework for Understanding Nationalism’. Ethnic and Racial Studies, 25, 1,
pp. 20–39.
Mahnig, H. and Wimmer, A. (2000) ‘Country Specific or Country Convergent?
A Typology of Immigrant Policies in Western Europe’. Journal of International
Migration and Integration, 21, 2, pp. 177–204.
McEneaney, E. and Meyer, J.W. (2000) ‘The Content of the Curriculum: an
institutionalist perspective’ in M.T. Hallinan (ed) Handbook of the Sociology of
Education. New York, London: Kluwer Academic, Plenum, pp. 189–211.
Meyer, J.W. and Ramirez, F.O. (2003) ‘The World Institutionalization of Educa-
tion’ in J. Schriewer (ed) Discourse Formation in Comparative Education, Second
Revised Edition. Frankfurt am Main, Oxford: Peter Lang, pp. 111–132.
Migrant Integration Policy Index (2012) ‘MIPEX Methodology: How Does
MIPEX Decide the Scores?’. http://www.mipex.eu/methodology, date accessed
2 August 2012.
Morris, P. and Marsh, C. (1992) ‘Curriculum Patterns and Issues in East Asia:
A Comparative Survey of Seven East Asian Societies’. Journal of Education Policy,
7, 3, pp. 251–66.
Morris, P., McClelland, G., and Wong, P.M. (1997) ‘Explaining Curriculum
Change: Social Studies in Hong Kong’. Comparative Education Review, 41, 1,
pp. 27–43.
Ramirez, F.O., Suárez, D., and Meyer, J.W. (2006) ‘The Worldwide Rise of Human
Rights Education’ in A. Benavot and C. Braslavsky (eds) School Knowledge
in Comparative and Historical Perspective: Changing Curricula in Primary and
Secondary Education. Hong Kong: Comparative Education Research Centre,
University of Hong Kong, pp. 35–52.
Rauner, M.H. (1998) The Worldwide Globalization of Civics Education Topics from
1955 to 1995, Unpublished Doctoral Dissertation. Stanford, CA: Stanford
University.
Soysal, Y.N. and Schissler, H. (2005) ‘Introduction: Teaching Beyond the National
Narrative’ in Y.N. Soysal and H. Schissler (eds) The Nation, Europe and the World:
Textbooks and Curricula in Transition. New York: Berghahn Books, pp. 1–12.
World: the Political Socialization of Youth in Formal Mass Education Systems’.
Compare, 41, 5, pp. 561–77.
Part II
The Social Outcomes of
Education Systems
6
Social Cohesion as Paradigm
Francois Dubet
The notion of social cohesion has asserted itself in the vocabulary of

social sciences as well as in that of political and institutional actors.
Beyond being fuzzy or limited to empirical indicators, this notion
reveals a profound transformation of the way in which the coherence
and unity of social life is explained. Rather than attempting to define
social cohesion precisely, this chapter tries to highlight the changes of
paradigms that have accompanied the notion’s rise to prominence.
Over the last three decades, the notion of social integration has grad-
ually been replaced by the concept of social cohesion in sociological
terminology. The idea of social cohesion has also replaced the notion
of solidarity in political terminology (or at least in French political
terminology). There are two ways of interpreting these changes in ter-
minology: either as new ways of describing old concepts or as the sign
of a more fundamental shift in social and political thought. Despite
the inherent vagueness of the new terminology, the second hypoth-
esis appears to be more plausible. While the paradigm of integration
and solidarity defines society as a ‘functional’ system bound together
by common values in which institutions perform an essential role, the
notion of cohesion implies that society is the product of the strate-
gies, actions and decisions of individuals and of the interactions and
networks developed by individuals.
Community, solidarity and cohesion
In seeking to answer the question, ‘how are societies bound together’,

sociological theory has provided three different answers.
1. Community. Community is defined as the traditional foundation of

social integration based on Gemeinschaft or mechanical solidarity: in
141
142 The Social Outcomes of Education Systems
this paradigm, societies are united by a belief in common values

and (above all) by a belief in a common bond extending beyond
the division of labour. All national societies may therefore be said
to be partly community-based insofar as they share a set of common
beliefs and values – for example religion, patriotism, historical myths,
support for football teams. Various indicators can be used to measure
community-based integration, including the religious, linguistic or
cultural homogeneity of a given society or the strength of its histor-
ical myths. The intensity of community feelings (such as patriotism,
national rites, the strength of class-based communities or ‘races’, the
density of family ties, etc.) may also be measured. Ultimately, soci-
ety is bound together (i.e. is cohesive) because it conceives itself as a
community that has developed into a nation state (or vice versa).
2. Solidarity, in the modern sense of the term, is based on the premise
of an increase in the division of labour or in the inequalities and
differentiations generated by capitalism. Society is considered to be
divided, conflict-ridden and resolutely individualistic. Solidarity is
deemed to be constructed by institutions, political representation,
and more generally by the welfare state, an entity that seeks to restore
the fundamental sense of solidarity that is exposed to the constant
threat of capitalism. As we know, the welfare state can take a variety
of forms. In this model, solidarity involves the political and institu-
tional development of a social contract. A whole range of objective
indicators can be defined, including the degree of social, cultural or
regional inequality, tax rates and the redistribution of wealth, the
degree of institutionalization of labour conflicts and so on. Follow-
ing Polanyi (2010), the solidarity paradigm was developed on the
basis of the ‘social question’: solidarity is the set of institutions and
practices that serve to embed capitalism and the pure market within
the rules of social integration.
3. Social cohesion is based on the premise that there are can be no funda-
mental conflict or tension between capitalism and social integration
since economic dynamism is a component of social cohesion and
social cohesion is generally conducive to development. This is the
central postulate of the British Enlightenment (especially Locke and
Smith). Rather than governing society, the state is expected to create
virtuous rules. This has clearly been the dominant model in the inter-
national literature on social cohesion, which has tended to empha-
sise social mobility at the expense of inequality, empowerment at
the expense of social protection and social capital in the sense of
Putnam and Coleman at the expense of community. Unlike the
Francois Dubet 143
two previous models, the cohesion paradigm posits that society is

constructed and developed by individuals who forge and constitute
society and by public policies that enable individuals to be mobilised
and active. The social cohesion vision is the ‘official philosophy’ of
the OECD, the European community and the World Bank, which
conduct assessments of societies based on the level of trust and social
capital.
Rather than an ‘ideal type’ in the Weber sense of the word, social cohe-
sion needs to be seen as a series of shifts in the various conceptions of
the mechanisms that bind individuals together and constitute a soci-
ety. Social cohesion is a set of mechanisms that forge and constitute
societies. The notion of cohesion is borrowed from Donzelot (2007)
and (more importantly) from the vocabulary of experts working at the
OECD and the World Bank and in ministries and think tanks involved
in developing key indicators that govern how we perceive and inter-
pret societies. While these indicators and their frameworks are clearly
debatable, it is important to recognise that the notion of cohesion has
become sufficiently established to be taken seriously – or at least to be
taken as a serious indicator of recent changes in the nature of soci-
eties. Whereas integration, in the manner of Durkheim and Bourdieu, is
defined as a cultural and social order governing the practices of actors,
cohesion implies an opposite mechanism governing the production of
society – specifically the mechanism of the agreements and coordina-
tion produced by social practices. Whereas integration is imposed from
above, cohesion emerges from below as an effect or product of ‘virtuous’
practices generating cohesion. In contrast to an earlier vision of society
conceived as the realisation of nature and of a stable system, society in
the cohesion paradigm is viewed as a continuous process of production.
In contrast to the ‘evolutionary’ presentation of the three broad ways
of making society provided above, various conceptions of the underly-
ing social contract binding societies together have emerged in recent
years. Abandoning a historical and evolutionist narrative in favour of
a comparison of European societies, the typology developed by Esping-
Andersen (1990) involves a distinction between three general models:
the model of the corporate state, which posits that society is the prod-
uct of public intervention; the model of the liberal state, which defines
a safety net for the poorest in society while positing a harmonious bal-
ance between self-interest and the strength of moral sentiments (what
we would refer to today as trust and social capital); and the model of
social-democratic arrangements. The advantage of this typology is its
emphasis on the idea that societies are bound together in a variety of

ways and that the various ways of producing a society tend to over-
lap. There may be effects of scale and degree and different ways of
producing a society according to whether the emphasis is on family,
local territories, nations and perhaps even wider levels, that is conti-
nents and universal religions. There is certainly no evidence to suggest
that societies are bound together in just one way. One of the ambiguities
of the notion of cohesion resides in the fact that it characterises not only
general social processes, but also particular conceptions of social life
close to those described by Esping-Andersen in the liberal model. Since
social cohesion has yet to become an established and stable concept, it
is perhaps preferable to search for indicators of cohesion rather than a
substantive definition. Four indicators will be used in this chapter: the
role assigned to individuals, the conception of public intervention, the
conception of social justice and the emergence of a concept as broad as
it is ‘miraculous’: social capital.
A new emphasis on the individual
Social integration presupposes that the action of an individual is embed-

ded in the system through socialisation, social roles and social control.
The idea of cohesion emerges when these mechanisms are undone,
causing a ‘transfer of responsibility’ to the individual. Individuals are
therefore required to do what society no longer does in their stead.
The long history of individualism can be interpreted as an injunction
to become the master of one’s fate and to be responsible for one’s deci-
sions, and is not a liberation so much as an imperative. The shift or
transfer implied by individualism is linked to the development of cul-
tural rights shaping the identity of individuals and to the capacity of
individuals to define themselves and to ensure that their unique self-
definitions are recognised and acknowledged by others and by society –
thereby protecting themselves from a whole range of discriminations,
but also acquiring a position within the public sphere.
Cohesion necessarily implies that societies are individualistic. Rather
than imposing positions and roles on individuals, cohesion imposes
challenges that must be overcome by individuals by acting in the
interests of their own autonomy. When the weight of traditions and
ascriptions declines, the onus is on individuals to construct, define and
‘motivate’ themselves (and their own selves) (Giddens, 1991; Beck,
1992). A similar point is often made in the contemporary sociol-
ogy of family: in describing how individuals produce idiosyncratic
Francois Dubet 145
interpretations of social roles that offer several ways of articulating their

passions and interests, the economy of feelings is defined as the foun-
dation of stable relations (de Singly, 2000). Because of the opportunities
provided by consumption and the proliferation of markets and scales
of taste, every individual is now required to develop their own style
and their own specific (i.e. idiosyncratic) combination of practices and
frameworks – even if this means creating a perfectly ‘kitsch’ identity and
establishing distinction and distinctiveness as refined (and somewhat
paradoxical) forms of conformism (Lahire, 2002).
The more we depart from institutional frameworks that serve to pro-
vide actors with systems of meaning and vertical and indisputable
motivations, the greater the pressure exerted on individuals to act and to
develop their own frameworks of motivation. Habermas argued that the
separation of the system and the experienced world causes an endemic
crisis of motivation. The institutions of control and socialisation have
been subject to profound transformations in recent times (Dubet, 2002).
As a result, roles have become misaligned, common beliefs have become
uncertain, situations have become unstable and individuals are now
constantly required to motivate themselves and to adjust their iden-
tity and their behaviours. Teachers and students in modern schools are
thus constantly involved in processes of motivation and adjustment to
produce something that resembles a working classroom. This is why the
type of activity is perceived as being increasingly wearisome and diffi-
cult insofar as it requires searching within the very depths of the self
for reasons to act (whereas mere ascription to a role had long been suf-
ficient). Studies of activism have emphasised similar tendencies while
deploring the perverse effects of individualism, construing individual-
ism as follows: I only become involved in causes that concern me deeply
and personally, and it is inconceivable that my involvement in the same
cause could last a lifetime.
It is important to note that individualism is not merely expressive (or
self-expressive) and is not limited to the private world of feelings and
emotions. Boltanski and Chiapello (2005) showed that signs or manifes-
tations of an expressive and artistic ethics can be found in a whole range
of contemporary managerial doctrines. The main objectives of the new
aesthetics can be defined as follows: make sure you achieve your tar-
gets, be imaginative, organise your work time, make contacts and build
networks and make sure you get involved! Individuals are enjoined to
take responsibility – particularly since the line of authority or chain
of command is far narrower in a system in which every individual is
responsible for their own results and for the development of their own
career. Even in a world as bureaucratic as academia, each individual is

encouraged to be free, creative and enterprising, even if they are ulti-
mately judged on tangible evidence. It would be wrong to assume that
the privileges of individual autonomy imply a form of freedom without
principles since the greater the range of possibilities available to us and
the freer we become, the more we are required to control ourselves. How
could we fail to see that at a time when almost all sexualities appear to be
allowed by moral freedom and pornography, sexual crime has become
the greatest of all crimes?
Whatever our specific judgements or anxieties, it would be wrong to
assume that the transfer of responsibility from society to the individual
is a mere trend or fashion. This transfer has in fact a far deeper sig-
nificance, since it implies that individuals are required to create social
cohesion by shaping and forging themselves and by acting, by creating
wealth, by mobilizing themselves and by developing links with oth-
ers through networks. The assumption is that in order to ensure social
cohesion, a society needs first of all to have been produced (manufac-
tured) by individuals, who must define an acceptable moral code, to
develop interpersonal relationships, to build links and connections that
can no longer be taken for granted, to become involved for their own
and for others’ sakes and to show both self-love and sympathy for oth-
ers. In this respect, it is perhaps worth returning to The Wealth of Nations
and The Theory of Moral Sentiments in order to reconcile the two conflict-
ing and complementary faces of individualism – namely selfishness and
sympathy, or calculation and generosity. Whereas social integration is a
system, cohesion is the result of the engagement of individuals. A good
society is an active society, that is a society that allows for the greatest
amount of individual action and autonomy.
If it is assumed that any sociological theory is based (more or less
implicitly) on a normative framework that defines the basic contours of
the good society (for the most optimistic) or the worst of all societies (for
the most sceptical), the shift from integration to cohesion also implies a
normative and critical shift. Here, the focus is less on the perfection of
the social organisation and democratic transparency than on what hap-
pens to the individuals involved in producing or manufacturing social
cohesion. A good society is a society that enables individuals to act as
much as possible and to enjoy the greatest degree of autonomy. It is a
society that enables every individual to be a subject. This normative shift
was perfectly described by Honneth in his examination of the history
of Frankfurt School critical theory. From Adorno to Habermas, criticism
gradually moved away from the perspective of history and the system in
Francois Dubet 147
favour of democracy. Honneth defined the ‘social pathologies’ that pre-

vent individuals from becoming the agents of their own actions. ‘What
needs to prevail and to form the very heart of the normality of a soci-
ety independently of any culture are the conditions that guarantee an
unaltered form of self-realization to the members of society’ (Honneth,
2007, p. 88; translated from the French). While social cohesion empha-
sises individuals and their capacity to act, it is also important to note
that the burden of moral outrage also shifts away from history and the
system and towards the individual.
From government to management
After parliamentary and representative democracy (which had previ-

ously held a monopoly over the definition of the common good), a
new form of democracy has emerged in recent times as a result of
the development of whole range of procedures granting individuals,
local communities and affinity groups the right to be heard and some-
times even the duty to make themselves heard (Rui, 2004): participatory
democracy. The purpose of participatory democracy is not to shape or to
weigh on projects so much as to secure popular support and to respond
to a democratic imperative that compels every individual to be involved
and to participate in the democratic process. Even when these proce-
dures are deemed to be relatively marginal and are suspected of being
manifestations of the selfishness of NIMBYs (Not in My Back Yard) or
ways of securing consent, it remains that they subvert a specific con-
ception of the construction or manufacture of society. Representative
politics is no longer required to tear society away from itself in order
to construct the general interest and to achieve transcendence in order
to impose the political. Instead, society can be founded and built on the
basis of the agreements, compromises and arrangements that establish
the rule of politics (Gauchet, 2005).
In France (and indeed in many other countries), there is little evi-
dence to suggest that the state is withdrawing – if the level of social
charges (which remains high) and the wealth devoted to common goods
such as education, public infrastructures and healthcare are anything
to go by. However, the specific ways in which the state intervenes and
the ways in which public policies are developed have changed radically
in recent times (Duran, 1999; Muller, 2003). The ‘top down’ model of
policies defined and implemented by a central state and trickling down
through the administrative chain identified with the general interest
has been gradually replaced by a system founded on joint action and
involving decentralised administrations, networks of elected represen-

tatives and ‘civil society’. In the new system, the aim is not to impose
a superior rationality on society so much as to encourage administra-
tions, associations and concerned actors to act together (i.e. in unison).
The assumption is that the common definitions of public issues must be
followed by targeted actions aimed at addressing the relevant issues, and
these actions need to be evaluated. The rationality of a plan is less impor-
tant than the level and degree of public mobilisation and involvement.
As a result, public policies have become more local, more ephemeral
and more collective – at the risk of being perceived as more uncertain
and less reassuring. Civil servants, association members, professionals,
elected representatives and activists are now constantly forced to coor-
dinate their actions, to develop projects and to be accountable – which
may exhaust them and even drive them to despair when policies mul-
tiply their targets and superimpose their own measures and systems.
Ministries of education are increasingly prone to introducing a whole
range of measures aimed at mobilising actors around a targeted issue
(e.g. reading, early school failure, incivility, nutrition, school dropout
rates, etc.). In short, the notion of a political program originating from
the centre of the state and imposing its rationality on society in the
name of social unity and integration has been replaced by a series of
political, administrative and social initiatives in which experts, elected
representatives, professionals and moral entrepreneurs of various causes
learn to work together and to create cohesion.
While the model of social integration aims to establish sustainable
public norms (if not eternal and universal norms), because it is assumed
that ‘common values’ command public action, cohesion implies gover-
nance and management based on indicators used as key instruments:
rational choice overrides adhesion to values (Lascoumes and Le Galès,
2004; Zimmerman, 2004). This is what Max Weber called the ‘disen-
chantment of the world’. This technical change represents a fundamen-
tal shift. The emphasis has shifted to the development of a system of
measurements and comparisons that enable societies and other systems
to conduct assessments, evaluations, comparisons and corrections on
a continuous and permanent basis. In Europe, the Lisbon strategy of
March 2000 (at the time under the influence of social-democratic gov-
ernments) defined the new rules for the creation of social cohesion.
Through benchmarking, a whole range of systems (social protection,
education, healthcare, research) are now constantly required to conduct
comparative evaluations and to improve their performance levels. Every
system is now required to use common indicators in order to measure its
Francois Dubet 149
position or rank within the group of countries concerned and to develop

good practices that will improve its competitiveness and increase its
positive impact on social cohesion. Public policy norms are thus no
longer legal or ‘moral’ but statistical and technical and are no longer
aimed at creating a stable state but are designed to adjust or alter sys-
tems based on comparative assessments. Evaluation is the new order of
the day. Comparisons are required to ensure progress and improvement,
and the assumption is that societies and systems are naturally in com-
petition and that cohesion is just one of the various dimensions of this
competition.
Consider the following example taken from the PISA surveys used
by the OECD to compare the efficiency and equity of educational sys-
tems. All other things being equal, educational systems can be ranked
based on their performance levels, that is the average attainment and
achievement level of students based on international surveys. Educa-
tional systems can also be ranked based on their level of equity, that
is the distribution of performance within the student population. The
best educational system is a system in which students achieve the high-
est level of skills and the gap between the highest-achieving students
and the lowest-achieving students is the narrowest. Everyone is required
to closely monitor everyone else, and a good system is a system that
has the best equity and efficiency results. A whole range of other fac-
tors can be used to refine the general picture produced by comparative
assessments, including student welfare, the level of self-confidence, the
general educational climate and the social efficiency of qualifications
on the labour market. When a country distinguishes itself by achieving
excellence, the factors that have served to shape its success are identified
and assessed, with other countries attempting, if not to imitate it, at least
to seek inspiration from it and to emulate it. In recent years, Finland has
become the paragon of education at primary and middle school levels.
As a result, there are very few recent reforms of educational systems
that have not sought inspiration from the Finnish model. In asking ‘has
France made the right choice?’, Natalie Mons (2007) was not seeking
to determine the principles and values that must be used to evalu-
ate education so much as to define an optimal form of management
based on a detailed comparative assessment of different educational
systems – specifically a comparative assessment of the indicators pro-
duced by different educational systems.
What is true of the automotive industry, schools and universities
(based on the Shanghai academic ranking of world universities) also
applies in other areas. Consider the case of security policies with the
case of New York and the broken windows theory. Consider also the
case of health policies and indeed many other areas – sometimes even
the most unexpected areas. In the 1980s, Robert Parker, an American
lawyer and wine lover, developed a rating system for wine vintages
that has since compelled wine producers throughout the world to sub-
scribe to a universal comparative system placing Bordeaux Grand Crus,
California wines and Australian wines on a par and within the same
market. Wine growers and wine merchants now seek to pre-empt or
second-guess Parker’s ratings, a practice that has resulted in increasingly
homogeneous wine productions. However, it is also in their interest to
demarcate and to differentiate themselves and to encourage other expert
assessments.
It is important to note that despite the many criticisms that have
been levelled against Parker’s rating system, the type of benchmarking
his system illustrates is not necessarily damaging to equality, well-being
and security. There is after all such a thing as left-wing benchmarking
and right-wing benchmarking – or in other words experts pitted against
other experts, think tanks against other think tanks, and so on. The
new way of thinking and acting radically alters the rhetoric and modes
of legitimation of public action. There is perhaps no better evidence of
this tendency than a recent book by Esping-Andersen (2008). In addi-
tion to seeking to defend equality, solidarity and social redistribution
on the grounds that they are good values, that is good for individ-
uals and for society as such, Esping-Andersen argues that cohesion is
conducive to economic competitiveness and efficiency. For example, he
uses the following argument to defend the creation of day-care centres:
(1) Day-care centres free women up for work and create service jobs.
(2) Female labour increases the level of household income and con-
sumption demand. (3) Salaried women pay taxes that serve to fund
day-care centres and other common goods. In the long term, a vir-
tuous circle is created that will ultimately benefit everybody – that is
women (who gain in autonomy), companies (which have a market and
female workers) and the wider community (which becomes more egali-
tarian and more cohesive as the population increases and pensions are
better funded). While disgruntled people might ask if children bene-
fit from such changes, it is important to note that Esping-Andersen’s
social-democratic argument is not based on any moral foundations, and
that rather than seeking to resist capitalism, it aims to socialise and to
tame it (based on the Scandinavian social model). Of course, there are
other types of indicators: consider the idea of a ‘Gross National Happi-
ness’ index that has been suggested by a number of economists inspired
Francois Dubet 151
by Sen, Stiglitz and the Kingdom of Bhutan. However, the same rea-
soning applies: assessments are conducted, comparisons are made and
adjustments performed.
There have been countless arguments of a similar kind that have
governed recent reforms of labour law and social policies. All of these
arguments would merit further examination since it is not necessar-
ily the case that the individual goods created by these ways of acting
automatically become common goods. There is always the possibility of
an ecological paradox rearing its head: the fact that educated individu-
als are more liberal, more peaceful and better paid than less educated
individuals does not necessarily mean that a higher level of educa-
tion implies a lower level of crime, intolerance and suicide, as shown
by the case of the United States, where the excellence of universities
appears not to be associated with good cohesion indicators (Green et al.,
2006).
From the equality of places to the equality of chances
If society is conceived as an integrated functional system, it follows that

social justice aims to reduce inequalities between different social posi-
tions, between the richest and the poorest, between bosses and workers
and so on. In this model, individuals are not necessarily intent on
changing their social status, nor do they insist that their children should
have the right to move up the social ladder or hierarchy. Rather, their
aim is to improve their living and working conditions (Dubet, 2009).
It is important to remember that the labour movement was for many
years largely unconcerned about aspirations for individual social mobil-
ity. It never claimed that the children of workers should rise above the
working class, preferring instead to focus on improving the situation of
workers. In this respect, the ‘elevation’ of the working class does not
mean the elevation of each individual in the social structure, but the
elevation of the class itself. The desire for equality implies the idea of
a fair division of labour. It also implies that everybody must be remu-
nerated and rewarded fairly according to their social utility and their
adherence to the core values of society.1 Everyone has a right to a recog-
nised and stable ‘rank’ and ‘honour’ in a stable (and therefore fair) social
hierarchy. Social justice creates solidarity, and universal social rights are
not incompatible with particular social rights. As shown by Iribarne
(1989), the legacy of the French Revolution is peculiarly ambivalent in
this respect: while it implies a rejection or refusal of aristocratic barriers,
it also implies the need to defend one’s rank in a fair order operating as a
kind of aristocratic structure open to all. However, beyond the matter of

specific national historical legacies, there is evidence of the emergence
of a model of justice that combines equality and fair functional hierar-
chies, that is a model in which people fight for universal equality and to
maintain differences determining positions based on their contributions
to the ‘functions’ of the system.
In recent years, the dominant conception of social justice has shifted
to outsiders (i.e. women, young people, migrants, etc.), who mobilise an
altogether different conception of social justice. Outsiders in contempo-
rary societies are not intent on denouncing inequalities between social
positions so much as criticising disparities in access to unequal posi-
tions. Rather than being levelled against the established hierarchy of the
social order, their criticisms tend to be directed against discriminations.
In other words: it is because we are fundamentally equal individuals that
we have the right to occupy unequal positions based on merit. Rather
than the wage differentials between executives and blue-collar workers,
the issue is the unequal access to executive positions based on birth,
gender, cultural origins or region of origin (or indeed any other factor).
The analysis of inequalities has changed radically in recent years since
its focus is no longer on social positions but on unequal opportunities
across a whole range of areas. The analysis of inequalities is thus no
longer based on an analysis of the social structure but on the analy-
sis of the obstacles and opportunities to which individuals are subject.
Social castes and classes have been superseded by competition between
individuals yearning for a society that has increasingly come to resem-
ble a sports competition. If we like sport so much, it is perhaps because
the drama of sport is founded on this particular conception of justice.
Through a careful enforcement of rules and refereeing, sports compe-
titions aim to determine a clear hierarchy between competitors viewed
as fundamentally equal participants. In short, I owe my victory or my
defeat solely to myself and sometimes to chance. The drama is made
all the more ‘pure’ by the fact that it is renewed and reiterated every
week-end, when anything is once again possible.
Equal opportunity is deemed to produce social cohesion insofar as
every talent can be expressed, the energy and dynamism exhibited
by every individual benefits everybody else, and fair competition pre-
supposes trust in rules and in others. The liberal equality model has
a number of practical and legal implications. The first is the fight
against discrimination. The first step involves identifying areas in which
discriminations are likely to arise (i.e. employment, education, housing,
Francois Dubet 153
leisure, rights, etc). The next step involves identifying the relevant
groups or publics (i.e. women, disabled people, groups stigmatized
on account of their origins or ‘race’, various cultural, religious, and
sexual minorities, etc). The combination of both factors – areas and
publics – results in bills and mechanisms for legal complaints and con-
trol designed to promote equal opportunities. A ‘veil of ignorance’ is
lifted since the tools and instruments used to describe social phenomena
become increasingly more refined and meticulous to allow for the fact
that every individual is subject potentially to a whole range of different
kinds of inequality. The more the veil is lifted, the easier it is to assign
causes and responsibilities (as illustrated by statisticians and insurance
companies: it is now easy to determine the topography of risks based on
lifestyle, the environment, genetic potential, etc). Alongside the indi-
vidualisation of inequalities and risks, liberal equality also reinforces
identity, membership and belonging based on common discriminations.
The veil may be so comprehensively lifted that every group may have
an interest in presenting itself as a victim of discrimination at the risk of
competing with other victims (Chaumont, 2002).
The perception of social justice based on discriminations has a num-
ber of practical implications for public policies, which are increasingly
designed (as noted above) to target specific audiences or groups with a
view counteracting the effects of handicaps that may affect the promo-
tion of equal opportunities. Education is the first area concerned by the
proliferation of positive discrimination policies throughout the world.
Beyond the sphere of education, a whole range of groups and areas have
begun to request specifically targeted aid and support in order to be able
to compete on an equal footing with others. In short, the political capac-
ity to exhibit handicaps has become an essential resource, as shown by
Lorrain (2006) in a comparative analysis of poor urban populations and
poor rural populations; since they protest and create disorder, the former
tend to attract far more public resources than the latter.
A key concept: Social capital
The success of the notion of social capital is indicative of the gradual

shift of social thought from the paradigm of integration to the paradigm
of cohesion. The ambivalence of the concept is part and parcel of its
impact, and as a result it has become the most widely used concept in
global sociological research (Bevort and Lallement, 2006). Social cap-
ital operates as a mixed concept sometimes defining the resources of
actors and sometimes defining the characteristics of the system, and

sometimes even cultural attitudes – and therein lies its charm and
effectiveness.
For example, Bourdieu argued that social capital involves the net-
works and relations that individuals develop with a view to optimis-
ing their interests. Bourdieu conceived social capital as a useful but
unequally distributed good – not unlike money, prestige or power.
‘Social capital is the sum of the resources, actual or virtual, that accrue
to an individual or a group by virtue of possessing a durable network of
more or less institutionalized relationships of mutual acquaintance and
recognition’ (Bourdieu, 1982). We ‘network’ just as we might dabble in
stocks and shares, and relations are fostered and maintained as a form
of capital that has the virtue of converting values into utilities (and vice
versa) – that is relations converted into influence, influence converted
into prestige, prestige converted into money and so on. Despite being
a far more utilitarian theorist than Bourdieu, Coleman (1988) defined
social capital as a quality of the system that enables rational actors to
act in a coordinated manner while pursuing fundamentally utilitarian
objectives: ‘Social capital is defined by its function. It is not a single
entity, but a variety of different entities having two characteristics in
common: they all consist of some aspect of social structure, and they
facilitate certain actions of individuals who are within the structure’
(Coleman, 1988, p. 302). The quality of the road network or of education
is thus deemed to constitute a form of social capital. In short, Bourdieu
the ‘culturalist’ provides an individualistic and utilitarian definition of
social capital, while Coleman the ‘utilitarian’ opts for a collective defi-
nition: cohesion means that everyone is prepared to give because they
know the collective will eventually return the gesture. Parsons’ institu-
tional individualism – the individual interiorising norms – is replaced
by rational choice individualism, an actor guided by their preferences
and involved in social relations according to their interests; in this case,
norms are practical compromises and no longer the expression of val-
ues. Confidence and authority are required to prevent the conflict of
egos from descending into a war of all against all. Though starting
out from the problem laid out by Hobbes, Coleman does not arrive at
the Durkheimian solution of shared values and internalised roles. His
answer is closer to the solution devised by Smith involving the idea of
conflicting and complementary egos and sympathy. Social capital has its
own significance and substance independently of its use by actors and
might involve the peaceful life of a village, the quality of a network of
communications or a legal system. Social capital is thus conceived as the
Francois Dubet 155
set of ‘social-structural resources [operating] as a capital asset’ (Coleman,

1988, p. 302).
Putnam (2000), the most well-known scholar in the field, construes
social capital as a civic virtue, drawing on the analysis of American com-
munity democracy conducted by Tocqueville via Almond and Verba
(1963) and the idea of cultural integration developed by Durkheim.
Here, social capital is conceived as a quality of social relations and of
the trust developed between citizens – implying that one might leave
one’s door open, rely on others, hold one’s word and so on. The notion
has been particularly successful insofar as Putnam (1993) showed that
social capital could contribute to economic development. Some regions
of Italy that have high levels of social capital are for instance sig-
nificantly more developed than regions in the south of the country,
where social capital tends to be limited to the family and the clan
and where it is impossible to mobilize resources that create collective
goods. Here, capital closes doors more than it serves to build bridges.
Because of a harmonious conjunction of strong bonds enclosing the
individual within the membership group and weak bonds opening the
individual out towards others, the unemployed are able to find work
(Granovetter, 2011). We may therefore say that there is such a thing as
good capital and bad capital. However, social capital generally fosters
trust in others, and trust is necessary for democracy and business. When
networks proliferate and are easily accessible, social capital favours ini-
tiative and economic activity: we allow credit and we trade; we know
that debts will be repaid; we know that everyone will benefit from
public infrastructures and that everyone will pay for their bus ticket.
In the same spirit, Sen explained that democracy and communication
explain the disappearance of famines since everyone feels a sense of
attachment to everyone else and free information and open critique
make famine intolerable. Despite being poor and relatively undemo-
cratic, India was able to eradicate famines before totalitarian China, a
country that had the capacity to conceal the issue. Trust is ultimately
turned into economic goods – and therein lies the ‘miracle’.
The notion of social capital is at the heart of a latent theory of social
cohesion insofar as capital is intimately connected with social networks
(Lin, 1999). It owes its success not to its originality and theoretical force
so much as its charm and its capacity to solve problems and contradic-
tions. Social capital is miraculous – as miraculous as the programmed
adjustments of social integration. It is both a utilitarian and a moral
concept demonstrating that social virtues (e.g. the meaning of rela-
tions, trust, tolerance, honesty, etc.) can be converted into economic
dynamism and vitality. It is not merely that it is good to have friends and
to love them. It is also useful for others and for oneself. Political econ-
omy becomes a moral economy, and the old conflict between capitalism
and social cohesion, selfishness and generosity, or individualism and the
community is thus overcome. Society is unable to resist the market – and
indeed never seeks to resist it. Rather, it embraces it and fosters its devel-
opment at the same time as the market manufactures society. In the long
term, the market and society are not in conflict since the market actively
needs society. Networks, open communities and voluntary associations
serve to redefine society as the emerging effect of social relations. This
may be seen as a new and sophisticated version of what Montesquieu
referred to as ‘doux commerce’: ‘It is fortunate for men to be in a situ-
ation in which, while their passions inspire the thought of being evil,
yet they have an interest in not being evil’ (L’Esprit des Lois, Book XXI,
ch.20). The notion of trust merely serves to expand this new integra-
tion of values and the economy around social cohesion. ‘Gross national
trust’ establishes a generalised reciprocity that is merely an immanent
conception of social life. The common good, that is cohesion, is thus
not a conscious design or purpose, but an effect.
Cohesion has another interest. It can be easily broken down into a
series of indicators generally inspired by Scandinavian countries, which
are generally perceived as having the most cohesive societies (Jenson,
1998). Social capital and trust are embodied politically by public poli-
cies, international bodies and social engineering aimed at associating
individual action, the withdrawal of the state, rational public action and
cohesion (Berger, 1998). These notions are all sufficiently soft to be eas-
ily submitted to indicators and benchmarking while serving as a viable
theoretical framework.
It would be all too easy to dismiss these notions and analyses with
the stroke of a pen. First of all, they have solidly established facts to put
forward. Social capital and trust are clearly good for economic develop-
ment and for the control of cultural, ecological, technical and scientific
problems, which are invariably construed as social problems. These per-
spectives also demonstrate that a purely market-based society is a fiction
and sometimes even a barbarity. There are clearly many different forms
of capitalism. Unless we were to adopt the point of view of Sirius, to
confuse two equally liberal societies would be to show a serious lack of
insight and understanding. Fundamentally, the notions of social cohe-
sion and social capital are no more and no less magical and miraculous
than the notions of integration and solidarity when they attributed the
Francois Dubet 157
division of labour to solidarity by the grace of common values and

institutions.
Concluding remarks
The shift from the paradigm of integration to the paradigm of cohesion

has been accompanied by a significant change in the general concep-
tion of education. In the integration model, education is expected to
transmit the key values and skills that are considered to be essential
in society. An educational system is required to reproduce a common
social life from one generation to the next. Education is thus conceived
as a cog in the social machine. In the social cohesion model, education
is expected to produce individuals and subjects capable of producing
or creating social cohesion. The production of autonomous and united
individuals is not merely a moral imperative since education is expected
to develop values that will promote social cohesion. In this sense, it is
less important to ask what society does to school than to ask what school
does to society – not merely in terms of qualification and human capital,
but also in terms of social cohesion. This raises a number of questions:
To what extent can it be said that different societies make the same
choices in terms of social cohesion? To what extent can it be said that
schools foster the same values as their society? Above all, to what extent
can it be said that the forms of cohesion developed in schools have
the predicted impact on social life? To what extent do ‘good’ schools
necessarily lead to ‘good’ societies? A comparative analysis of three key
components is required to answer these (and indeed many other) ques-
tions: education systems, the societies in which they are embedded, and
the ways in which societies ‘use’ education.
Note
1. Parsons attempted unconvincingly to develop a theory of stratification based
on the following argument: stratification is explained by proximity to the
central values of the system.
References
Almond, G. and S. Verba (1963) The Civic Culture: Political Attitudes and Democracy
in Five Nations. Princeton: Princeton University Press.
Beck, U. (1992) Risk Society: Towards a New Modernity. New Delhi: Sage.
Berger, P. (ed.) (1998) The Limits of Social Cohesion: Conflict and Mediations in
Pluralist Societies. Boulder, CO: Westview.
Bevort, A. and M. Lallement (eds) (2006) Le Capital Social. Paris: La découverte,
Mauss.
Boltanski, L. and E. Chiapello (2005) The New Spirit of Capitalism. London-
New York: Verso. (1999).
Bourdieu, P. (1982) ‘Le Capital Social. Notes Provisoires’ Actes de la recherche en
Sciences sociales, 31, pp. 2–3.
Chaumont, J.M. (2002) La concurrence des victimes. Génocides, identité, reconnais-
sance. Paris: La Découverte.
Collins, R. (1998) The Sociology of Philosophies: A Global Theory of Intellectual
Change. Cambridge, MA: The Bleknap Press.
Coleman, J. (1988) ‘Social Capital and the Creation of Human Capital’ American
Journal of Sociology, 94, pp. 95–120.
Donzelot, J. (2007) ‘Un Etat qui rend capable’ in S. Paugam (ed.) Repenser la
solidarité, pp. 87–109. Paris: PUF.
Dubet, F. (2002) Le déclin de l’institution. Paris: Seuil.
Dubet, F. (2009) Injustice at Work. Boulder-London: The Yale Cultural Sociology
Series, Paradigm Publishers.
Duran, P. (1999) Penser l’action publique. Paris: L.G.D.J.
Esping-Andersen, G. (1990) The Three Worlds of Welfare Capitalism. Cambridge:
Polity Press & Princeton: Princeton University Press
Esping-Andersen, G. and B. Palier (2008) Trois leçons sur l’Etat-providence. Paris:
Seuil.
Gauchet, M. (2005) La condition politique. Paris: Gallimard.
Giddens, A. (1991) Modernity and Self-Identity. Self and Society in the Late Modern
Age. Cambridge: Polity.
Granovetter, M. and R. Swedberg (eds) (2011) The Sociology of Economic Life, 3rd
edition. Boulder, CO: Westview Press.
Green, A., G. Janmaat and J. Preston (2006) Education, Equality and Social Cohesion.
A Comparative Analysis. New York: Palgrave Macmillan.
Honneth, A. (2007) Disrespect: The Normative Foundations of Critical Theory. Polity
Press.
d’Iribarne, P. (1989) La logique de l’honneur. Paris: Seuil.
Jenson, J. (1998) Mapping Social Cohesion: The State of Canadian Research, CPRN
Study N◦ F-03. Ottawa: Renouf Publishing Co.
Lahire, B. (2002) Portraits sociologiques. Paris: Nathan.
Lascoumes, P. and P. Le Galès (ed.) (2004) Gouverner par les Instruments. Paris:
Presses de Sciences Po.
Lin, N. (1999) ‘Building Network Theory and Social Capital’ Connexions, 22, 1,
pp. 28–51.
Lorrain, D. (2006) ‘La dérive des instruments : les indicateurs de la politique de la
ville et l’action publique’. Revue française de science politique, 56, 3.
Mons, N. (2007) Les nouvelles politiques éducatives. Paris: PUF.
Muller, P. (2003) Les politiques publiques. Paris: PUF.
Polanyi, K. (2010) The Great Transformation: The Political and Economic Origins of
Our Time. Boston, MA: Beacon Press. 1944.
Putnam, R.D. (2000) Bowling Alone: The Collapse and Revival of American Commu-
nity. New York: Simon and Schuster.
Francois Dubet 159
Putnam, R.D., R. Leonardi, and R. Nanetti (1993) Making Democracy Work. Civic
Traditions in Modern Italy. New York: Simon and Schuster.
Rui, S. (2004) La démocratie en débat. Paris: Armand Colin.
de Singly, F. (2000) Libres ensembles. L’individualisme dans la vie commune. Paris:
Nathan.
Zimmerman, B. (2004) Les sciences sociales à l’épreuve. Paris: Editions de la MSH.
7
Pre-School Education and Care:
A ‘Win-Win’ Policy?
Tarek Mostafa and Andy Green
Debates about policy frequently involve identifying social and

economic trade-offs. A policy which is designed to boost economic
competitiveness may have negative social consequences or, conversely,
policies designed to enhance social cohesion may come at a high eco-
nomic cost. The tensions between economic and social goals seem
particularly evident in times of economic austerity. However, social
scientists occasionally identify macro-policies which they claim would
have clear benefits, both on the macro-economic and macro-social side.
Publicly-funded pre-primary education and care (PSEC) is one such
policy area. Esping-Andersen (2009), in particular, has claimed that it
represents a clear ‘win-win’ policy for developed countries.
The argument, put simply, is that increasing the availability and
take-up of good quality pre-school education and care brings major eco-
nomic and social benefits at the societal level. Affordable and accessible
pre-school provision frees up mothers of young children to undertake
paid work and is thus likely to increase the employment rates of mothers
and female employment rates in general. Raising rates of employment
makes an important contribution to raising GDP per capita and improv-
ing living standards and has been a central goal for European economies
since the Kok Report (Kok, 2003). At the same time participation in
pre-school education and external care is held to improve the cognitive
abilities of young children which will help them to achieve more in their
school education. As this feeds through to higher levels of skills and
qualifications in the workforce this will benefit the economy through
enhanced productivity.
Esping-Andersen (2009) also claims that widespread participation
equalises educational outcomes in the longer term, pointing in particu-
lar, to the effects of near universal provision in the Nordic countries. He
160
Tarek Mostafa and Andy Green 161
claims that social gaps in the quality of parenting and the time invest-
ments made by parents in their children are growing, and that this may
be increasing the unequal effects of social inheritance on life chances
and reducing social mobility in many countries. High levels of partici-
pation in pre-school education, he says, can mitigate against this trend
and reduce inequalities in school performance, since the gains to disad-
vantaged children, he supposes, are greater than to other children. He
offers the example of the Nordic countries where, between the 1960s
and 1990s, participation in pre-school education and external care and
maternal employment rose in tandem, at the same time as inequality in
education outcomes decreased.
The simultaneous social and economic benefits of widespread pre-
school education and care, particularly as exemplified in the Nordic
states, would seem to be highly impressive, yet few countries have
invested sufficiently to universalise pre-school participation. Is the case
in favour as clear cut as it would seem? Or do the benefits vary across
countries, according to other contextual factors. In this chapter we seek
to test the robustness of the claims using time-series international data
on participation and female employment (full and part-time) and the
cross-sectional international data from PISA 2009 on pre-school partici-
pation and inequalities in educational outcomes at the age of 15. More
precisely, we are trying to answer the following questions:
1. Do higher PSEC participation rates at the national level lead to higher

levels of female employment?
2. What are the returns to PSEC in terms of literacy performance scores,
and do these returns vary according to economic, social and cultural
status (ESCS)?
3. Does universalising PSEC contribute to equalising literacy outcomes
within a country?
4. When PSEC is progressively universalised starting with the lowest
social groups, what are the average gains in terms of educational
outcomes that each group can benefit from?
5. Does universalising PSEC increase the average educational outcomes
of a country and enhance its international ranking?
Our findings indicate that the impact of PSEC on female employment

rates is positive and significant both statistically and in magnitude.
Moreover, increasing free PSEC provision to the most disadvantaged
social groups does increase their cognitive prospects at the age of 15 and
simultaneously reduces inequalities in performance scores. However,
inequalities drop until reaching a minimum, and then start to rise again
when free PSEC is provided to all children of the top social groups.
We also found that increased PSEC provisions enhance the national
average PISA scores of a country and improve its ranking in the OECD
league table.
Pre-school education and care and female employment
In the first part of this chapter we assess the impact of aggregate PSEC
participation on female employment at the macro-level. Our approach
is conceptually different from the existing literature. First, most of the
previous studies are at the micro level; the explanatory variables they
utilise are related to the characteristics of the household and to those
of the individual (e.g. mother’s education, single motherhood, the age
of the child, the availability and cost of PSEC, household income, etc).
In contrast, in our analysis, the control variables are institutional and
national in their nature. More precisely, the objective is to measure the
impact of institutional and country-level characteristics on total and
female employment rates. Such characteristics include expenditure on
unemployment, union coverage, wage coordination, rigidity of employ-
ment, benefit replacement rate and duration and, of course, aggregate
PSEC participation rates. To put it differently, micro-level analyses are
concerned with individual behaviour while macro-level analyses are
concerned with labour market institutions.
Secondly, the objective of micro-level analyses is to measure the
impact of the availability of PSEC on the likelihood of women under-
taking paid employment. Even though such analyses indicate that
PSEC does improve the probability of women being employed, they do
not measure the impact of increased aggregate PSEC participation on
employment rates. In other words, such studies do not account for the
bigger picture. Conversely, our objective is to test whether increased par-
ticipation in PSEC leads to higher employment rates. It is also worth not-
ing that when policymakers decide on increasing free PSEC places, they
would like to know by how much employment rates would increase.
This can be answered using macro-analyses, while micro-analyses can
only report by how much the probability of being employed will
increase at the individual level. One may think that since the relation
holds at the individual level then it should hold at the macro-level.
This is not necessarily true. As we will see, the magnitude of the effect
of aggregate PSEC participation on female employment varies between
countries. Thirdly, micro-level analyses, in general, utilise national or
regional data and do not allow for international comparisons. In con-

trast, our study uses macro-data covering the period from 1980 to 2008
for 16 OECD countries.
In what follows we review some of the literature on the impact of
PSEC participation on female employment. Most of the studies focus on
the micro-level in order to estimate the effects of variations in the afford-
ability and availability of PSEC on the likelihood of female employ-
ment. The evidence points to a positive effect of factors pre-disposing
participation and actual participation on female employment rates.
A number of studies in the United States (Heckman, 1974; Connelly,
1992; Ribar, 1995; Kimmel, 1998; Powell, 2002) show that costly child
care acts as a barrier to employment for women. Mason and Kuhlthau
(1992, p. 915) found that up to 30 per cent of mothers of pre-school
age children in the United States felt constrained in their employment
due to childcare problems. Chevalier and Viitanen (2002) report, on
the basis of the 1988 Family Resources Survey, that over 20 per cent
of British women aged 18–44 stated that childcare obligations restricted
their employment possibilities.
Simonsen’s (2010) analysis of Denmark uses regional variations in cost
and availability of PSEC up to 2001. She finds that an increase in the cost
to parents of childcare by 1 Euro per month will decrease the probability
of a women being employed, during the first 12 months after childbirth,
by around 0.08 per cent.
Another study (Kreyenfeld and Hank, 2000) analyses the effect of the
costs and availability of PSEC in western Germany where most provision
is provided by the municipal authorities. Kreyenfeld and Hank model
availability of PSEC places (slots per 1000 children) and female like-
lihood of employment, controlling for the age of the youngest child,
mother’s education, single parenthood, partner’s wage and regional
employment rates. They find that the availability of child care has more
impact than the cost, but only has a small effect on the mother’s propen-
sity to employment. This, they argue, is probably because the limited
hours of day care restrict the mothers’ ability to work anyway.
In addition to the effects of availability and costs of PSEC, the proba-
bility of maternal employment is likely to be influenced by the mothers’
preferences for either staying at home to look after young children or
returning to work soon after childbirth and making use of external
child care. Evans and Kelly (2001) analyse women’s preferences across
countries using International Social Survey Program (ISSP) data. They
find considerable variation with, for instance, Australian mothers gen-
erally preferring to stay at home with young children, and mothers
in ex-communist countries in Central and Eastern Europe being more

likely to want to return early to work. Better educated women, they
find, are more likely to want to return to work, perhaps because their
higher qualifications give them access to better paid employment which
is more likely to provide them a net income gain after they have met the
costs of child care. In what follows we present our analyses and the find-
ings on the relationship between PSEC and female employment at the
macro-level.
Data, estimation and findings
Data
In this section we are using a compiled macro-dataset for the
following countries: Australia, Austria, Belgium, Canada, Denmark,
France, Germany, Ireland, Italy, New Zealand, Norway, Spain, Sweden,
Switzerland, the United Kingdom and the United States. These countries
represent the major economic regimes (i.e. the liberal, social market and
social democratic regimes). Such regimes are known to be different in
terms of their education systems and labour market institutions (Esping-
Andersen, 1990; Hutton, 2002; Green et al., 2006; Iversen and Stephens,
2008). The liberal regime includes the English-speaking countries, the
social market regime includes the German-speaking countries, the social
democratic regime includes the Nordic states, and the Mediterranean
regime includes countries in southern Europe. One should also note that
some countries, like France and Denmark, have ambiguous positions
(see Verdier’s chapter in this book on the hybrid system in Denmark).
For instance, France belongs to the social market regime when it comes
to labour market institutions while it belongs to the Mediterranean
regime when it comes to its education system (Green et al., 2006). Pos-
sibly these different regimes modify the effect that PSEC participation
has on outcomes such as (female) employment and achievement at the
age of 15.
Our data cover a period of 29 years from 1980 to 2008.1 The depen-
dent variables of the model are logempf and logempt, the logarithm of
female and total employment rates respectively. The time-varying inde-
pendent variables are: Logpresch, the logarithm of the gross enrolment
ratio in pre-school education and care; Expunemp, social expenditure on
unemployment as percentage of GDP (including active labour market
policies); and Prison, prison population rate measured as the number of
inmates per 100,000 inhabitants. The time-constant independent vari-
ables are: Rigemp, the index of rigidity of employment (0 = less rigid;
100 = more rigid) and Unioncov, the index of union coverage measured
by the percentage of employees covered by collective trade agreements.
This latter variable is categorical: 1 is under 25 per cent, 2 between
25 per cent and 70 per cent, and 3 over 70 per cent); category 1 is
considered to be the reference group. Coordin is the index of the level
of wage coordination and bargaining. This variable is equal to the sum
of union and employer coordination indices used in Nickell and Layard
(1998). Benfrep is the benefit replacement rate which is equal to the share
of income replaced by unemployment benefits. Benfdur is the duration
of unemployment benefits in number of years. Owner is the percent-
age of individuals who own their houses. This variable is supposed to act
as a brake on labour force mobility.2 One should note that the choice of
the controls was based on the work of Nickell and Layard (1998) and on
the availability of data over the (1980–2008) period.
Estimation
We estimated four different variants of the base model. The first is a

pooled OLS regression, the second is a fixed effects model, the third
is a random effects model, and finally the last is a random coefficient
model (the random coefficient is on Logpresch). Using the Hausman and
the Likelihood-Ratio tests, we found that the random coefficient model
is the most efficient and most robust. Note that a fixed effects model
treats country heterogeneity as fixed and estimates an intercept for each
of the countries, while random effects models treat this specific effect
as random country-specific effect. Further, even though we are inter-
ested in the effect of PSEC participation on female employment, we
are not denying the possibility of the existence of a reverse causality
where female labour participation would lead to the expansion of PSEC
provision. In fact, the relation between the two is likely to be reciprocal.
Findings
In what follows we present the findings on the random coefficient

model while limiting the discussion to the impact of PSEC participation
on total and female employment rates. Note that Green and Mostafa
(2011) provide a full interpretation of the results.
Table 7.1 shows that the effect of 1 per cent increase in PSEC atten-
dance on female employment is of 0.61 per cent. In other words, if PSEC
is made more universal (an increase of 10 per cent) this would lead to
an increase of female employment of 6.1 per cent, which is substantial.
Table 7.1 The impact of PSEC participation rates on female and total
employment.
Female employment Total employment
Logempf Logempt
Logpresch 0.618 0.202

(0.000) (0.001)
Rigemp −0.00247 −0.00193
(0.059) (0.054)
Unioncov2 −0.0332 −0.146
(0.199) (0.200)
Unioncov3 −0.573 −0.276
(0.000) (0.000)
Coordin 0.187 0.0929
(0.000) (0.000)
Expunemp −0.0356 −0.0345
(0.000) (0.000)
Benfrep 0.00568 −0.00382
(0.864) (0.759)
Benfdur 0.103 −0.00655
(0.799) (0.966)
Prison 0.000559 0.000192
(0.000) (0.002)
Owner 0.0885 0.0297
(0.064) (0.100)
Constant −8.955 −0.0248
(0.039) (0.988)
Between Variance 3.2386 0.4321
Within Variance 0.0029 0.0008
Between Variance on 0.4116 0.0594
Logpresch
N 463 463
Note: p-values in parentheses.
When compared with the results for total employment, it is obvious

that PSEC is more effective (three times more effective) in boosting
female employment rather than total employment. This confirms the
argument that PSEC helps women get back to work after childbirth.
Hence, countries with higher levels of PSEC attendance are more likely
to have higher levels of female employment. In Table 7.2, we present
the country-specific effects on the coefficient of Logpresch. Note that
country-specific effects measure by how much each country deviates
from the average effect of 0.61 per cent. If the specific effect is significant
Table 7.2 Country-specific effects on the relation between Logpresch and

Logempf.
Country Average Specific p-Value on Total

effect effect specific effect effect
Australia 0.618 −0.138 0.425 0.618

Austria 0.618 −0.377 0.039 0.241
Belgium 0.618 1.130 0.000 1.747
Canada 0.618 0.112 0.563 0.618
Denmark 0.618 −0.353 0.105 0.618
France 0.618 0.239 0.315 0.618
Germany 0.618 0.269 0.189 0.618
Ireland 0.618 1.648 0.000 2.266
Italy 0.618 0.200 0.287 0.618
New Zealand 0.618 −0.572 0.001 0.046
Norway 0.618 −0.449 0.007 0.168
Spain 0.618 0.156 0.347 0.618
Sweden 0.618 −0.597 0.000 0.021
Switzerland 0.618 −0.308 0.109 0.618
UK 0.618 −0.569 0.001 0.049
USA 0.618 −0.289 0.129 0.618
it adds up to the average effect. In Table 7.2, Column 1 gives the average
effect of Logpresch, the second column gives each country’s departure
from the average effect, the third gives the significance level of such a
departure, and the last one is the total effect which is equal to the sum
of the average and specific effects.
The countries where the effect of PSEC does not deviate from that
of the overall regression coefficient are: Australia, Canada, France,
Germany, Italy, Spain, Switzerland and the United States. The countries
where the effect of PSEC is above the overall regression coefficient are
Belgium and Ireland. In contrast, the countries where the effect of PSEC
is below the overall regression coefficient are Austria, Denmark, New
Zealand, Norway, Sweden and the United Kingdom. From these find-
ings we can conclude that PSEC attendance works differently across the
selected countries.
The reasons why the strength of the effect of PSEC on employment
varies across countries is clearly very complex since a number of dif-
ferent variables are involved, including, for instance, cultural factors
and mothers’ preferences. The strong effect of PSEC participation on
employment in Ireland may be high, for instance, because mothers of
young children have a relatively strong preference for working (Evans
and Kelly, 2001, p. 31). The weaker effects of PSEC on employment
in the Scandinavian countries may be due to the fact that both PSEC
participation rates and female employment rates are very high, or
because high female employment rates are being driven by factors other
than PSEC participation (such as child-friendly employment practises at
work). In the United Kingdom, the effects may be weaker because much
of the PSEC participation is part-time, which allows fewer opportunities
for mothers to work (at least in full-time jobs), or because preference
among mothers for staying at home is still relatively high compared
with other countries (19th highest out of 23 countries according to
Evans and Kelly’s (ibid) data for 1994).
Pre-school education and care and literacy achievements
In the second part of this chapter we assess the impact of PSEC par-
ticipation on reading performance scores using PISA 2009 data. The
importance of early learning for children’s cognitive development and
future learning has been emphasised in many recent studies and the
research that demonstrates this has been taken very seriously by pol-
icymakers in a number of countries (Waldfogel, 2004). Recent studies
based on the analysis of longitudinal data in the United Kingdom
suggest that up to half of the gap in children’s cognitive abilities is
already established by the age of 11 years (Clegg et al., 2009) or ear-
lier. Hence, learning during the early years is highly important to a
child’s cognitive and broader educational development and different
experiences of parenting and early years education and care do appear
to contribute substantially to social inequalities in educational perfor-
mance. As Esping-Andersen (2009) writes, ‘If the race is already half
run before the child begins school, then we clearly need to exam-
ine what happens in the early years.’ These early years affect much
that happens in the child’s schooling later on. ‘Like it or not,’ says
Esping-Andersen, ‘the most important mental and behavioural pat-
terns, once established, are difficult to change once children enter
school.’ (p. 81). Further, PSEC also facilitates female employment, as
we have seen in the previous section, and hence leads to an increase
in the household’s disposable income. The rise in income is said to
have a positive effect on children’s cognition and on their future
prospects.
Waldfogel (2004) provides a review of recent international research on
early cognitive development. She acknowledges that the research shows
that there are multiple influences on development in the early years
and classifies these into three types which include child endowments,
parenting and home environment and pre-school education and care.

She says that what parents do generally matters more than what early
schooling does, but that the latter can be effective. Research in the
United Kingdom and the United States, based both on experimental
PSEC programmes and using large-scale longitudinal surveys, gener-
ally suggests that there are significant cognitive gains to children over
the age of one from receiving high-quality PSEC. In some studies the
benefits are particularly marked for children from more disadvantaged
homes. PSEC also improves performance in skills measured at 15 years
(OECD, 2010a, p. 98). About 70 per cent of those tested at 15 years in the
PISA 2009 study had received one or more years of pre-primary educa-
tion. After controlling for social background, attending a year or more of
PSEC was associated, on average across the OECD, with a 33 point gain
in test scores at 15 years. In all countries, children who participated for
more than a year in PSEC got, on average, higher scores at 15 than those
who didn’t.
There is a fairly widespread consensus that high-quality PSEC brings
educational advantages to children after the first year. However, what
is not so clear is whether high levels of participation tend to lead to
the equalisation of educational outcomes. More equal distributions of
skills and qualifications among adults are associated cross-nationally
with more equal income distributions and these, in turn, are associ-
ated with a wide range of social goods, including better public health,
lower rates of crime and higher levels of trust (see Wilkinson, 1996;
Green et al., 2006; Wilkinson and Pickett, 2009; Green and Janmaat,
2011).
Esping-Andersen (2009) has recently argued that universalising PSEC
does contribute to equalising educational outcomes. In particular, he
argues that where there is near universal participation in consistently
high-quality PSEC for young children, as in the Scandinavian coun-
tries, this is contributing towards reducing social gaps in educational
achievements at the end of compulsory schooling. In much of Europe
attendance at Kindergarten for children aged 3–6 is near universal
already, but attendance among 1–3–year-olds is often much less so –
at about 30 per cent in Belgium, the Netherlands and the United
States, and at only around 10 per cent in Austria, Germany and south-
ern European countries (p. 93). What distinguishes the Scandinavian
countries – and what most contributes towards their relatively equal
educational outcomes at 15 – says Esping-Andersen, is that PSEC for
1–3-year-olds is also very widespread and of consistently good qual-
ity. What is the evidence that Scandinavian PSEC contributes towards
equalising educational outcomes. Esping-Andersen provides two main

arguments:
First, he argues that the over-time evidence for Scandinavia suggests
that PSEC has contributed to the amelioration of educational inequal-
ity. According to his figures, the decline in social inheritance effects in
Nordic countries between the 1960s and 1990s coincided with increases
in participation in PSEC and rising levels of maternal employment.
‘Indirectly,’ he writes, ‘there is evidence to suggest that the arrival of uni-
versal pre-school attendance is associated with a significant equalisation
of school attainment and, one can argue, also links with the compa-
rably quite homogenous performance in PISA . . . tests.’ The decline is
most evident, he says, among the younger cohorts, who were the first to
enjoy near universal participation in PSEC (p. 135). By contrast, in coun-
tries which have done less to universalise PSEC, such as Germany, the
United Kingdom and the United States, there was no equivalent decline
in social inheritance effects over the last half century.
Secondly, Esping-Andersen cites the evidence from some studies that
PSEC is particularly beneficial for children from disadvantaged families
who benefit disproportionally from attendance. Since PSEC is near-
universal in Scandinavian countries he claims that this would mean
that a larger proportion of those most prone to educational under-
achievement were receiving benefits which will serve to close the social
gaps in attainment generally. His general argument is made as follows:
If early child care were to compensate for unequal cultural capital,

we would expect that the latter’s explanatory importance would be
systematically weaker in the Nordic countries than elsewhere. The
reasoning is that participation in child centres that are similar in
quality across the board, so to speak, help cancel out the stimulus
gap that children from low-educated and culturally weak homes suf-
fer. Utilising again the PISA data this is in fact what we find. The
influence of parents’ cultural capital (and socio-economic status) is
systematically lower in Scandinavia than elsewhere. (p. 136)
The logic of Esping-Andersen’s case is certainly very compelling. How-

ever, the evidence for it is rather speculative, as his tentative tone
implies. It is certainly the case that educational outcomes in Nordic
countries are relatively equal, and that this is borne out by the rela-
tively low social gradients for PISA scores in these countries. However,
Esping-Andersen is not able to prove that this is due to the effects of
near-universal PSEC rather than, say, the relatively egalitarian nature of
the compulsory school systems (Green et al., 2006). The associational
evidence he provides comes some way short of proving causality.

Whether PSEC provision is equalising education outcomes in any given
country will therefore depend on how it is distributed. In fact, as the
OECD (2010a) shows, participation in PSEC in most countries is skewed
towards children of higher social class families. This is even true in
Scandinavia, despite near-universal provision, since, as Esping-Andersen
admits, non-attendance in PSEC, particularly during the crucial earliest
years, is most common among immigrant and poorer families. It may
be true that universal participation among 1–6-year-olds would equalise
educational outcomes. However, the fact is that participation in PSEC is
not equally spread around children from different social groups, even in
Scandinavia.
Data, estimation and findings
Data
In this part of the chapter, the OECD Program for International Student
Assessment (PISA, 2009) dataset is used. The dependent variable of the
model is reading performance scores, and the independent variables are
the following: ESCS Economic, social and cultural status of every student
and PSEC A binary variable taking the value of 1 if a student attended
pre-school education for more than one year. Otherwise, PSEC is equal
to zero (one or less years of PSEC attendance). ESCS∗ PSEC is an interac-
tion term between ESCS and PSEC. It measures the additional returns
from attending PSEC for higher ESCS students. MALE is a binary vari-
able taking the value of 1 is the student is a male. NONNAT is a binary
variable taking the value of 1 if a student is a first- or second-generation
immigrant. Note that all the selected controls are exogenous to PSEC
participation (these were also used in Mostafa, 2011). In other words,
they are not affected by PSEC. This choice was made because we are sim-
ulating the implications of a policy, and thus we are interested in the
gross effects of PSEC. For this reason we omitted school-level controls
(e.g. school quality) which might have been affected by PSEC.
Estimation
In this chapter we utilise a simulation approach using PISA 2009 data to

shed light on how PSEC attendance affects literacy performance scores
and educational inequalities. We retain only one country: the United
Kingdom, which was chosen because of the recent policy debate revolv-
ing around PSEC provision to the most disadvantaged families. We start
by estimating a fixed effects multilevel model in order to determine
the magnitude of the returns to PSEC attendance. Then we simulate

the effects of universal PSEC provision using counterfactual data. More
precisely, after estimating the multilevel model, we progressively univer-
salise PSEC participation starting with the lowest economic, cultural and
social status (ESCS) decile and moving up to reach the top decile. At each
stage of the universalisation process we compute the average predicted
performance scores for each ESCS decile, for the United Kingdom as a
whole and we compute their dispersions. This allows us to measure the
change in average predicted literacy scores and the change in the levels
of inequality.
One should note that since the sampled PISA students are 15 years
of age in 2009, they must have attended PSEC in the second half of
the 1990s. Hence, in this analysis we are answering the question of
what could have happened to literacy attainments in 2009 if these stu-
dents had universal PSEC provision back then. The time lag between the
two events (i.e. PSEC attendance and standardised testing) is inevitable,
since they are separated by more than 10 years. However, this time lag
is essential to justify the causal relation between PSEC attendance and
improved literacy scores. Further, the United Kingdom has been chosen
because the within-country variation in PSEC attendance (in the late
1990s as measured in PISA, 2009) is high enough to justify the analysis.
In what follows, we present some descriptive statistics before presenting
the regression results.
Table 7.3 gives PSEC participation rates in the late 1990s for students
in the United Kingdom for each social group. As expected, the rates
increase the higher the social group. Table 7.4 provides the figures for
the simulations. Each step in the simulation (e.g. step 1) gives the total
participation rate when PSEC is universalised up to a particular social
level. Step 1 gives the total participation rate when we simulate and
increase participation for the bottom decile to 100 per cent. Step 2 gives
the total participation rate when we simulate and increase participation
for the bottom two deciles to 100 per cent and so on. The total PSEC
Participation rate starts at 64.7 per cent (the real figure) and increases
progressively in the simulation until reaching 100 per cent when PSEC
is provided to all students (i.e. all ten deciles).
Findings
In this section, we present the different findings from our analyses start-
ing with the regression results. We limit the interpretation to the main
variables of interest.
Table 7.3 PSEC participation (more than one year)

before universalisation.
ESCS groups UK
Group 1 54.0
Group 2 58.5
Group 3 61.5
Group 4 62.5
Group 5 64.4
Group 6 66.0
Group 7 67.2
Group 8 67.6
Group 9 70.8
Group 10 72.6
All groups 64.7
Table 7.4 Cumulative total PSEC participation after

progressive universalisation for each group.
Universal PSEC UK
Before universalisation 64.7

Step 1 69.1
Step 2 73.1
Step 3 76.8
Step 4 80.3
Step 5 84.0
Step 6 87.3
Step 7 90.6
Step 8 94.0
Step 9 97.2
Step 10 100
In the United Kingdom, students attending PSEC for more than one
year benefit from 20 additional points on their literacy performance
scores in comparison with pupils getting one or less years of PSEC atten-
dance (see Table 7.5). Further, an increase in ESCS of one unit leads to a
21-point increase in test scores. Similarly, the interaction term between
PSEC attendance and ESCS has a significant and positive impact on liter-
acy test scores. Students with one additional unit of ESCS are expected to
have five additional points on their test scores if they are attending PSEC
for more than one year.3 However, one should note that this coefficient
is only significant at the level of 5 per cent and it lost significance when
Table 7.5 Regression results.
Student performance
PSEC 20.029
(0.000)
ESCS 21.049
(0.000)
ESCS∗ PSEC 5.3661
(0.046)
Male –23.172
(0.000)
Non-native 1.7961
(0.712)
Constant 465.38
(0.000)
Nb of schools 482
Nb of students 12179
Note: p-values are in parenthesis.
school-level variables are added to the model. This loss of significance

indicates that the interaction term might have absorbed some effects
that transit through school characteristics. In other words, higher ESCS
students attending PSEC for more than one year are likely to be in bet-
ter quality schools.4 It is also worth noting that in the United Kingdom,
attending PSEC for more than one year does compensate the lack of
one unit of ESCS as the effects of both variables are almost of the same
magnitude. Hence, providing free PSEC for one or more years to a lower
ESCS student will help to bridge the performance gap between him and
his upper ESCS counterparts.
In Table 7.6 we present the average predicted performance scores for
each ESCS group at each stage of the universalisation process. As stated
in the introduction, after estimating the multilevel model we compute
the average and the dispersion of predicted performance scores using
counterfactual data which simulates the progressive universalisation of
PSEC. Thus, the columns represent the ESCS deciles which acquired uni-
versal PSEC. Column 1 (0 per cent) presents the results with the data as
it is, the following column (10 per cent) presents the results when PSEC
was provided universally to the lowest ESCS decile (group 1), and hence-
forth until reaching (100 per cent) where PSEC is provided universally to
all ESCS groups. The last column presents the changes between columns
(0 per cent) and (100 per cent).
Table 7.6 Average predicted performance scores for each ESCS group before and after universalisation in the UK.
ESCS groups 0% 10 % 20 % 30 % 40 % 50 % 60 % 70 % 80 % 90 % 100 % Change
Group 1 436.3 445.5 445.5 445.5 445.5 445.5 445.5 445.5 445.5 445.5 445.5 9.2
Group 2 458.1 458.1 466.4 466.4 466.4 466.4 466.4 466.4 466.4 466.4 466.4 8.3
Group 3 467.2 467.2 467.2 475.0 475.0 475.0 475.0 475.0 475.0 475.0 475.0 7.7
Group 4 477.6 477.6 477.6 477.6 485.1 485.1 485.1 485.1 485.1 485.1 485.1 7.5
Group 5 487.2 487.2 487.2 487.2 487.2 494.4 494.4 494.4 494.4 494.4 494.4 7.1
Group 6 496.5 496.5 496.5 496.5 496.5 496.5 503.3 503.3 503.3 503.3 503.3 6.8
Group 7 502.7 502.7 502.7 502.7 502.7 502.7 502.7 509.3 509.3 509.3 509.3 6.6
Group 8 518.6 518.6 518.6 518.6 518.6 518.6 518.6 518.6 525.1 525.1 525.1 6.5
Group 9 532.1 532.1 532.1 532.1 532.1 532.1 532.1 532.1 532.1 537.9 537.9 5.8
Group 10 553.2 553.2 553.2 553.2 553.2 553.2 553.2 553.2 553.2 553.2 558.7 5.5
UK average 494.2 495.1 495.9 496.6 497.3 498.0 498.7 499.4 500.0 500.7 501.3 7.1
UK std 53.6 52.6 52.0 51.5 51.1 50.9 50.8 50.7 50.8 51.1 51.5 –2.1
UK variance 2871 2768 2704 2651 2616 2590 2577 2570 2583 2607 2647 −224
Coef of variation 0.1084 0.1063 0.1049 0.1037 0.1028 0.1022 0.1018 0.1015 0.1016 0.1020 0.1026 −0. 0058
175
The cells highlighted in dark grey present the average of predicted

performance scores for each ESCS group before they have been given
universal PSEC. The cells highlighted in light grey present the average of
predicted performance scores for each ESCS group after they have been
given universal PSEC (e.g. for group one, in the (0 per cent) column the
PSEC participation rate was 53.5 per cent, in column (10 per cent) it was
increased to 100 per cent). The findings can be summarised as follows.
In the United Kingdom, all groups benefit from universal PSEC provi-
sion with the lowest groups benefiting the most. Students in the lowest
decile benefit from an increase of 9.2 points while those in the top decile
benefit from an increase of 5.5 points. This hierarchy of gains reflects
the hierarchy of participation in PSEC before the universalisation pro-
cess. In other words, students in the lowest ESCS deciles are the most
affected by non-attendance in PSEC. Therefore, they benefit the most
from universalisation. Further, the universalisation of PSEC allows each
ESCS decile to catch up with the next ESCS decile in terms of average
performance scores (knowing that students in the next decile are not
getting universal PSEC). However, in general the gain of each group is
not enough to equalise average performances of two consecutive ESCS
deciles.
After universalising PSEC, the national average of the United Kingdom
increases by seven points. This increase corresponds with a rise of the UK
ranking in the b-PISA league table of 12 positions (OECD, 2010b; p. 15).
When it comes to the dispersion of the predicted performance scores,
Table 7.6 presents three measures, the standard deviation, the variance
and the coefficient of variation. For the interpretation we use the latter
because it is standardised to the mean (i.e. it is the ratio of the standard
deviation to the mean).
Figure 7.1 shows that the coefficient of variation of predicted perfor-
mance scores drops with the universalisation process until reaching a
minimum when PSEC is provided to the lowest seven deciles of ESCS
(i.e. lowest 70 per cent), then it starts to increase again. This indicates
the existence of a tipping point after which inequalities in literacy per-
formance scores start to increase. This finding reflects the results of the
regressions. PSEC participation has a strong positive effect on perfor-
mance scores irrespective of ESCS. Hence, anyone who gets PSEC will
achieve higher test scores. Hence, providing PSEC to the lowest seven
deciles of ESCS pushes their average test scores up while those of the
top three deciles are still the same. This is why the dispersion of test
scores drops. However, if PSEC is provided universally to all ten deciles,
the test scores of the upper three ESCS deciles will increase and the
0.1090
0.1080
0.1070
0.1060
0.1050
0.1040
0.1030
0.1020
0.1010
0.1000
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
UK
Figure 7.1 The coefficient of variation
dispersion will expand. It is worth noting, that even though the disper-
sion increases after the threshold of 70 per cent it never reaches its initial
level. Hence, we can conclude that universalising PSEC does equalise lit-
eracy performance scores while increasing their average levels for each
social group and for the United Kingdom as a whole. However, the max-
imum equalisation of literacy performance would be achieved not by
universalisation among all groups, but by restricting universalisation to
the lower seven deciles.
This finding provides a nuance to Esping-Andersen’s (2009) theory
that educational inequality in PISA test scores is strictly decreasing when
PSEC is progressively universalised. In spite of this, one can still argue
that PSEC should be universally provided to everyone irrespective of
their ESCS. First, universal provision helps maintain a sense of solidarity
among the different social groups and gives more legitimacy to such
policy. Secondly, even though the coefficient of variation is convex and
it increases after the tipping point of 70 per cent, it never reaches its
initial level. Further, the national average is strictly increasing which
justifies universal PSEC provision.
Conclusion
In this chapter, we measured the effects of PSEC participation rates on

aggregate female employment rates and the effect of PSEC participa-
tion at the micro-level on literacy performance scores at the age of 15.
Our findings confirm the positive impact of widening PSEC participa-
tion on female employment. The effect of PSEC on total employment is
relatively small, but that on female employment is sufficiently strong to

consider PSEC as a major instrument in boosting female employment.
Further, increasing free PSEC provision to the most disadvantaged social
groups will increase their cognitive prospects at the age of 15. Further,
providing PSEC free of charge to these groups will compress the distribu-
tion of performance scores within the United Kingdom at a later stage.
However, one has to keep in mind that inequalities start to rise again
when free PSEC is provided to the top social groups (top 30 per cent).
Our conclusions do support the idea that high-level PSEC provision
is, a ‘win-win’ policy, having both economic and social benefits. They
also support the policy undertaken by the British Coalition Govern-
ment according to which PSEC provision will be extended to 260,000
two-year-olds from the most disadvantaged families. This number rep-
resents about 30 per cent of the entire cohort of two-year-olds. Given
our findings, we expect that this extension of free PSEC provision to the
most disadvantaged 30 per cent will increase their literacy attainments
at the age of 15 and will reduce inequalities in performance scores by
compressing their distribution at the UK level.
Notes
1. Note that the period spanning 2009 to 2011 was not included because the
high levels of unemployment caused by the economic recession might distort
the relations we are trying to study.
2. The data were compiled from the OECD, the World Bank and Nickel and
Layard (1998). Total employment was included for comparative purposes.
3. One should note that the interaction term measures the additional returns to
PSEC attendance for higher ESCS students.
4. Our regression results differ from those published in the PISA 2010 report
(volume II, p. 193) for two reasons. First, in the PISA report the two categories
of PSEC attendance, ‘one or less years’ and ‘more than one year’ were merged,
and the reference group was ‘no PSEC at all’. In our analysis, given the fact
that most students were already attending one or less years of PSEC in the
United Kingdom and Sweden, it made more sense from a policy point of view
to model PSEC attendance for ‘more than one year’ against ‘one year or less’
and ‘no PSEC at all’ combined. Secondly, the PISA report used a number of
school variables as controls since it is interested in measuring the net effect of
PSEC on literacy performance scores. This is not the case in this chapter, since
we are interested in the gross effect.
References
Chevalier, A. and Viitanen, T. (2002) ‘The Causality between Female Labour
Force Participation and Availability of Child Care’ Applied Economics, 2–9,
pp. 915–918.
Clegg, J., Stackhouse, J. Finch, K. Murphy, C. and Nichols, S. (2009) ‘Language

Abilities of Secondary Age Pupils at Risk of School Exclusion: A Preliminary
Report’ Child Language Teaching and Therapy, 24, 1, pp. 99–115.
Connelly, R. (1992) ‘Self-employment and Providing Child Care’ Demography, 29,
1, pp. 17–29.
Esping-Andersen, G. (1990) The Three Worlds of Welfare Capitalism. Princeton, NJ:
Esping-Andersen, G. (2009) The Incomplete Revolution: Adapting to Women’s New
Roles. Cambridge, UK: Polity Press.
Evans, M. and Kelly, J. (2001) ‘Employment for Mothers of Pre-School Children:
Evidence from Australia and 23 Other Nations’ People and Place, 9, 3, pp. 23–40.
Green, A. and Janmaat, J.G. (2011) Regimes of Social Cohesion: Societies and the
Crisis of Globalisation. Basingstoke: Palgrave Macmillan.
Green, A. and Mostafa, T. (2011) ‘Pre-School Education and Care – A “Win-Win”
Policy?’ LLAKES Research Paper No. 32, University of London: Institute of
Education.
Green, A., Preston, J. and Janmaat, J.G. (2006) Educational, Equality and Social
Heckman, J, (1974) ‘Effects of Child-Care Programs on Women’s Work Effort’
Journal of Political Economy, 82, pp. S136–S169.
Hutton, W. (2002) The World We’re In. London: Little, Brown.
Iversen, T and Stephens, J.D. (2008) ‘Partisan Politics, the Welfare State and
Three Worlds of Human Capital Formation’ Comparative Political Studies, 41,
pp. 600–637.
Kimmel, J. (1998) ‘Childcare Costs as a Barrier to Employment for Single and
Married Mothers’ Review of Economics and Statistics, 80, pp. 287–299.
Kok, W. (2003) Enlarging the European Union: Achievements and Challenges. Report
of Wim Kok to the European Commission. European University Institute,
Florence Italy (26 March 2006).
Kreyenfeld, M. and Hank, K. (2000) ‘Does the Availability of Child Care Influ-
ence the Employment of Mothers? Findings from Western Germany’ Population
Research Policy Review, 19, pp. 317–337.
Mason K. O. and Kuhlthau, K. (1992) ‘The Perceived Impact of Child Care Costs
on Women’s Labour Supply and Fertility’ Demography, 29, pp. 523–543.
Mostafa, T. (2011) ‘Decomposing Inequalities in Performance Scores: The Role
of Student Background, Peer Effects and School Characteristics’ International
Review of Education, 56, 5, pp. 567–589.
Nickell, S. and Layard, R. (1998) Institutions and Economic Performance, LSE
Discussion Paper. London: LSE.
OECD (2010a), PISA 2009 Results: Overcoming Social Background – Equity in
Learning Opportunities and Outcomes (Volume II - Annex B1, Table II.5.7)
http://dx.doi.org/10.1787/9789264091504-en
OECD (2010b), PISA 2009 Results: What Students Know and Can Do – Student
Perf Romance in Reading, Mathematics And Science (Volume I) http://dx.doi.
org/10.1787/9789264091450-en
Powell, L. (2002) ‘Joint Labor Supply and Childcare Choice Decisions of Married
Mothers’ The Journal of Human Resources, 37, 1, pp. 106–128.
Ribar, D. (1995) ‘A Structural Model of Child Care and the Labour Supply of
Married Women’ Journal of Labour Economics, 13, pp. 558–597.
Simonsen, M. (2010) ‘Price of High Quality Day-care and Female Employment’

Scandinavian Journal of Economics, 112, 3, pp. 570–594.
Waldfogel, J. (2004) ‘Social Mobility, Life Chances, and the Early Years’ CASE
Paper 88, London School of Economics: Centre for the Analysis of Social
Exclusion.
Wilkinson, R.G. (1996) Unhealthy Societies. The Affliction of Inequality. London:
Routledge.
Wilkinson, R.G. and Pickett, K. (2009) The Spirit Level: Why Equality is Better for
Everyone. London: Penguin.
8
Educational Models and Their
Impact on Student Attitudes
Nathalie Mons, Marie Duru-Bellat and Yannick Savina
Researchers working in the field of comparative education have

provided extensive descriptions of national educational systems.
Research aimed at identifying the links between the structural character-
istics of educational systems and the academic performance of students
aged 15 has also developed in recent years, especially since the intro-
duction of PISA surveys. Yet, despite these new developments, very
little empirical research has been conducted on curricula and student
attitudes. This represents a significant gap in the field since the core
function of education is to transmit intellectual skills and knowledge
and to develop attitudes that will enable students to enter adult life.
This chapter presents the results of a comparative research aimed at
examining how curricula are structured in different countries and at
determining their impact on student attitudes. It aims to make three
contributions to sociological research:
a) It examines schooling and education through the key dimensions of

curricula and learning content;
b) It focuses on a largely overlooked aspect of the outcomes of educa-
tion: student attitudes;
c) In doing so, the study examines curricula from a cultural integration
perspective – what unifies a whole generation – and not from the
perspective of the reproduction of social divisions.
Curricula: An underexplored field of the sociology

of education
In examining education, sociologists may choose to examine either

the strictly cultural dimensions of the skills, knowledge or values
181
transmitted by different educational systems or the social functions per-

formed by education in governing the reproduction of social groups.
This was precisely the aim of the research programme initiated in the
late 1960s by the sociologist Michael Young: with Knowledge and Con-
trol (1971), he aimed to analyse what counts as knowledge at a given
time in a given society and to identify the factors that determine the
skills and knowledge that are deemed worthy of being transmitted
in formal education. The hypothesis is that educational curricula, the
values incorporated into school knowledge, the organisation of educa-
tional practices and daily classroom life contribute all together to the
reproduction of social hierarchies.
At the same period, the French sociologist Pierre Bourdieu published
a key text entitled ‘Systèmes d’enseignement et systèmes de pensée’
(1967; ‘Systems of education and systems of thought’). He argued that
educational curricula are the reflection of ‘systems of thought’ and are
designed to model how people read and treat reality. So, the educational
institution aims to inculcate an intellectual practice enabling commu-
nication on the basis of a ‘common sense’ that forms the ‘cultivated
habitus’ of a given era. Bourdieu was clearly operating from a cultural
integration perspective rarely adopted in his subsequent research, which
focused instead on the social function of education.
Whatever its intrinsic merits or value, the research conducted by
Young, Bourdieu and others is faced with a number of difficulties. One
major difficulty is the extreme vagueness of the notion of curriculum.
As well as covering educational curricula in the accepted meaning of
the term, the term curriculum also refers to anything experienced by stu-
dents in the course of their education and the resulting socialisation.
The analysis of a curriculum will therefore include prescribed or ‘official’
curricula. For instance, some national curricula include civic education,
while others do not. Beyond official programmes and learning content,
the curriculum also covers (by extensions) the structural organisation of
education. Educational structures include the varying degrees of local
differentiations (based on the degree of autonomy granted to individual
schools), or the practices governing student grouping (notably setting
and mixed-ability grouping), indicative of the prevailing cultural models
and values within a given society.
The notion of learning content or curriculum also implies the global
tone of the educational experience, that is the general atmosphere or cli-
mate that prevails in schools: which are the norms, values or qualities
learnt by students that are likely to facilitate or hinder their integration
in society? In classroom life, a key value such as individualism may be
Nathalie Mons, Marie Duru-Bellat and Yannick Savina 183
fostered through the range of available options or choices or the pro-

vision of individualised education. By contrast, communitarian values
may be fostered by factors such as regular group-learning situations,
a shared learning environment and the elimination of competition.
The nature of teacher–student relations is a key factor in this respect,
as well as the management of learning and discipline by teachers.
As educational organizations, schools may foster an ‘esprit de corps’ with
varying degrees of emphasis, as illustrated by the British and American
educational systems (Alexander, 2000).
The links between educational curricula and the general organisation
of educational systems can be structured and formalised, as suggested
by Bernstein (1971, 1975), who defined two broad types of ‘education
knowledge codes’. The collection code posits that disciplines are isolated
from one another and that educational knowledge is clearly distinct
from familiar knowledge in a system governed by authoritarian and
hierarchised relations based on strict discipline and a precise learning
pace. The integrated code posits that knowledge is less clearly divided
and compartmentalised and that knowledge is structured by an educa-
tional project defined at a school or teacher level. The integrated code
puts greater emphasis on the experience of learning (i.e. how learning
happens) than on the attainment of a given level at a given point in
time. While what is expected of students in the collection code is made
explicit and ‘visible’, implicit and ‘invisible’ goals tend to be given far
greater emphasis in the integrated code. In the integrated code, stu-
dents are assessed on a far wider range of behaviours and attitudes
extending beyond their academic skills and knowledge. The result is
that the integrated code implies a more encompassing form of social
control.
For Bernstein, the two codes reflect conflicting visions of the world.
The collection code implies a universalistic conception of knowledge
and justice that favours student access to knowledge, conceived as an
instrument of liberation and progress. In the collection code, the dis-
semination of knowledge must be uniform and homogenous through-
out the national territory, and student knowledge and skills must be
assessed by exams to ensure fair competition. Conversely, the integrated
code emphasises individualised learning, and its chief objective is to
ensure that every student develops their full potential and personal-
ity. In the integrated code, justice is not founded on uniformity but
emphasises the particular circumstances of students at a local level. So,
there will be policies aimed at favouring autonomy and differentiation
between schools.
In this chapter, we investigate the relations between educational

curricula – in the broad sense of the term – and a range of student atti-
tudes, with a more particular emphasis on attitudes that are deemed to
favour social integration at school.
Research question and methodology
A description of curricula is necessary, before assessing their possible

impact on students. It is hypothesised here that curricula and learning
content have significant socialising virtues, although it is important not
to posit a mechanical causal link between educational curricula and stu-
dent attitudes. Concerning student attitudes, recall that the PISA surveys
examine a wide range of student attitudes at 15, when students from
the same country are assumed to share a relatively common framework
of cultural values and attitudes. Here, five attitudes were selected among
the range of attitudes assessed in PISA surveys: trust in the utility of edu-
cation, adherence to and support for the behavioural values and norms
of the school (such as collaboration and competition), and the sense of
belonging to a peer group (a detailed description of these variables is
given in the fourth section).
Concerning the prescribed curricula, we gave particular attention to
the importance of vocational education and vocational counselling,
which may be deemed to foster integration when it aims to prepare
for professional integration. Life skills were also examined since they
may foster a shared practical culture. The importance given to regional
languages in education was also used as a key indicator to assess the
integration of different linguistic communities. The specific uses of civic
education were also examined (based on the curricula, but also on
student participation in the democratic life of the school). Finally, reli-
gious education curricula were examined based on the hypothesis that
the importance given to religious education may constitute a relevant
indicator for characterising curricula core educational objectives. These
dimensions were examined in every member state of the OECD based
on a range of categorical variables (see Figure 8.1).
Beyond the issue of content of curricula, we also gave special atten-
tion to their differentiation at a student level (i.e. distinct pathways
or courses of study, or individualized education) and at a school level
(based, for example, on the level of autonomy, within the state system or
of the juxtaposition of public and private education systems). Therefore,
in the multilevel models presented in the fourth section two charac-
teristics of schools were incorporated: autonomy and membership of a
private network.
Unlike the majority of comparative educational studies, generally based on

pre-existing data, this research is based on original categorical curriculum
variables designed to complete the data provided by PISA surveys. The practical
methodology is based on some main principles (see Mons, 2007 and 2008 on the
method).
One of the most important is to collect factual data rather than relying solely on
the opinions of indigenous experts. A wide range of sources of information were
combined (academic papers, international reports, quantitative data provided by
international organizations, but also material used by actors). That leads to
elaborate categorical or second-generation variables, resulting from the
synthetic analysis of a set of simple variables founded on factual data (for
example the teaching time devoted to civic education, the autonomous or
non-autonomous nature of the discipline . . . ). These second-generation variables
were constructed (to some extent arbitrarily, as in any typology) based on a
pre-defined combination of the categories of first-generation variables.
We developed an original database covering 33 countries or political/cultural
communities and including roughly 20 variables, most of which were categorical
and based around the following themes: vocational education, life skills,
vocational education counselling, religious education, the integration of regional
languages and forms of individualised education (these variables and their
categories are explained in Appendix 8.1).
Figure 8.1 An original database about curricula in OECD countries
Beyond the issue of prescribed curricula, it was also important to take

into account the implementation of curricula in class, particularly stu-
dents’ responses concerning teacher–student relations (for example, to
what extent do teachers show interest or provide support to students).
A variable describing the quality of discipline in class from the perspec-
tive of headteachers was used to complete the subjective assessment of
life in the classroom.
A necessary step to test the hypothesis of a relation between cur-
ricula and student attitudes was to develop a curriculum typology,
based on the variables presented above. We expect that an educational
system that puts a significant emphasis on subject areas other than
traditional academic disciplines (i.e. vocational and practical subjects,
vocational counselling), that aims to promote civic participation in
school, and that seeks more generally to individualise learning and
to adapt teaching to the particular circumstances and context of the
school may be said to be relatively close to what Bernstein termed
the integrated mode. The impact on students may include favourably
viewed teacher–student relations and an emphasis on integration in
a peer group. However, we may also make a counter-hypothesis, with
pupils more insecure whenever such a global and may be not explicit
evaluation is the rule.
We should underline, to end, the inherent limitations of the available

data, as far as the test of our hypothesis is concerned. PISA data are useful
insofar as they provide a general overview of a relatively broad field of
attitudes relating to educational integration and are conducted at the
end of compulsory education. However, in such international compar-
ative studies, attitudes are significantly more difficult to measure than
educational skills or knowledge, and the correlative data are less reliable.
Another difficulty is that it is impossible to strictly isolate the specific
impact of educational curricula and learning content, with cross-section
data of the kind produced by PISA surveys, since they merely provide a
picture of the result or outcome of the entire educational career – and
socialisation – of students aged 15. Observations made at a given point
in time cannot be used as a basis for dissociating the impact of school
education and educational curricula from the impact of the joint action
of parents, peers and the media.
To progress in that direction, a prerequisite is to demonstrate the
diversity of curricula and to identify ideal types (or ideal-typical config-
urations) that imply a simplification of reality based on the theoretical
hypotheses presented above. While there can never be a perfect corre-
spondence between an educational system and a particular type of cur-
riculum included in the typology, this study is based on the assumption
that every system will reflect one of the types included in the typology
by sharing most of its characteristics or by differing significantly from
the other types.
Towards a curriculum typology of compulsory education
The first stage involved developing a typology of curriculum models

based on a critical re-examination of the Bernstein’s one: would the
classic binary opposition of collection code and integrated code be con-
firmed (the first one being more widespread in continental Europe, the
second one in countries such as UK or USA)? A selection of formal
curricula variables and variables related to the atmosphere in the class-
room and the context governing the implementation of curricula were
incorporated into the analysis (see Appendix 8.1 for description of these
variables).
A first analysis found strong links between the different curriculum
dimensions, thereby validating the relevance of an approach based on a
typology of curriculum models (see Figure 8.2).
Firstly, the graphic representation of the Cramer’s V matrix with val-
ues above 0.40 indicates strong relations between the variables relating
Civic participation .51/.56 Like skills
.40/.45 .56/.53 .48/.53
Pre-vocational education .82/.87 Pre-vocational counselling .48/.44 Regional

language
.48/.47
.51/.56 .61/.69
Quality of
.45/.48 teacher-student .42/.48
relations
.49/.58
First value: Cramér’s V with pairwise deletion.
Second value: Cramér’s V with listwise deletion.
Indiscipline .54/.55 Religious education
Figure 8.2 Graphic representation of Cramér’s V >0.40 matrix1
to the inclusion (or absence) of vocational pathways in school, that

is the modes of organisation of pre-vocational counselling and pre-
vocational education (Cramer’s V 0.82). Pre-vocational counselling and
pre-vocational education were found to be linked to the integration of
life skills such as consumer economics, manual work and health edu-
cation (Cramer’s V respectively 0.56 and 0.48). A key factor linked to
the quality of teacher–student relations was found, thus confirming that
an analysis of educational curricula needs to extend beyond the formal
content of curricula. Links between curriculum choices and the con-
texts of dissemination of prescribed curricula were also found (Cramer’s
V 0.51 for the relation with vocational education and Cramer’s V 0.61
for the relation with vocational counselling and guidance). While it is
more peripheral, the second dimension of class atmosphere (or learning
environment) – the level of indiscipline as perceived by students – was
also found to have an impact.
Other selected curriculum components were found to have a more
marginal impact: religious education, the inclusion or absence of
regional languages and the level of civic participation. Overall, this first
analysis highlighted strong links between the various curriculum com-
ponents initially selected as part of this research and which may have
seemed somewhat disparate.
However, a second stage was needed, since the typology provides
as yet no indications about the relations between the different cate-
gories of these variables and their possible correlations. For example,
when vocational education is compulsory for all pupils, what are the
life skills that tend to be taught? Is there not a link not only between
each variable but also between specific categories of these variables,
thus pointing to curriculum models that are characteristic of particular
groups of countries?
Several country classifications were produced by varying the data table
(type of variables and coding), the similarity index and the method of
classification. The aim was to obtain a partition with very few varia-
tions between the various classifications. Among the range of reasonably
sized partitions (2 to 12 classes), four-cluster partitions were selected
because they had the highest degree of similarity. Figure 8.3 presents
Belgium french-speaking
Belgium dutch-speaking
Czech Republic
United States
New Zealand
Great Britain
South Korea
Netherlands
Switzerland
Germany
Denmark
Australia
Hungary
Portugal
Sweden
Canada
Norway
Greece
Finland
Mexico
Iceland
Poland
Austria
Ireland
France
Japan
Spain
Italy
Student grouping Tracks
Ability grouping
No grouping
Individualised education Widespread
No or remediation
Elite tracks
Model 1 (cluster 1): Model 2 (cluster 2): Model 3 (cluster 3 and 4):
total education producer academic education
Pre-vocational counselling medium
Pre-vocational education within track
Life skills medium
Quality of teacher-student relations [45% ; 60%]

Pre-vocational counselling high
Life skills high

Pre-vocational education compulsory
Quality of teacher-student relations [60% ; 70%]

Religious education medium
Indiscipline [15% ; 30%]
Religious education no
Pre-vocational counselling weak
Regional languages weak
Pre-vocational education no
Life skills weak
Religious education weak
The category (row) is present in the country (column)

Active category in the classification Supplementary category
Figure 8.3 Country clusters and distinctive categories

the four-cluster partition which gives the best picture of all the different
four-cluster partitions.
As shown in the Figure 8.3, specific groups or clusters of countries
appear to share significant similarities in terms of formal educational
curricula and learning environments. The indicator matrix shown in
the figure presents the characteristics of each country for the most dis-
criminant variables.2 At the top of this matrix (considered to be outside
the curriculum typology), two variables were added based on previous
research (Mons, 2004) conducted on individualised education and stu-
dent grouping (although it is not always a clear-cut dimension). Strong
links were again found between educational curricula in the broad sense
of the term and student grouping policies.
Bernstein’s binary typology distinguishes between two types of cur-
riculum also found in the typology developed here. Model 1 mainly
includes countries from Northern Europe and the English-speaking
world3 based on characteristics linked to the integrated organisation of
the ‘open school’. Model 1 (which can be described as the ‘total educa-
tion’ model) involves a wide-ranging curriculum that integrates subject
areas extending beyond the confines of traditional academic disciplines
and that is open to the outside world. In model 1, curricula generally
include vocational education and guidance as well as counselling for
all students, life skills and a frequently ecumenical religious culture.
Model 1 curricula thus recognise the wide-ranging nature of education –
that is the academic knowledge of future educated adults, the vocational
or professional prospects of the future producer and the spiritual life of
the future individual and citizen.
As noted by Bernstein, the emphasis on guiding children through
the various dimensions of their personality is associated with a marked
emphasis on the individual, characterised here by close relations
between adults and children (the variable ‘high quality of teacher–
student relations’). Model 1 is also characterised by the attempt to
develop an educational system without any hierarchy of courses or
ability levels. Students are generally taught in mixed-ability classes or
ability groups flexibly organized around a specific discipline. Preference
is given to partially individualised learning over whole-class or whole-
group teaching. While clear examples of this ideal-type can be found
in some countries, educational systems in other countries often share
some of its characteristics without ever embodying it altogether, such
as Greece and Spain, which offer developing practical courses (such
as home economics at middle school in Greece). Model 1 reflects the
findings of a study by Osborn et al. (2003): while the British liberal

model is based on individualised student support, the Danish model
is rooted in local communities and a strong popular tradition of local
democracy. However, both models contrast markedly with the French
model of universal education cut off from the individual and designed
as the embodiment of the Republic ideal.
A correspondence analysis on the indicator matrix of Figure 8.3
enriched with country membership of the four clusters was performed
to produce a visual representation of the similarities and differences
between different national models (see Figure 8.4). This analysis is
known as Multiple Correspondence Analysis (MCA).4 The rows (coun-
tries) and columns (categories) of the complete indicator matrix are rep-
resented by points in two clouds: the clouds of categories and those of
countries plotted on two axes generated by the analysis. Category points
that are close together represent categories which are in association (i.e.
are present within the same individuals). Country points that are close
together represent countries where the same categories are present.
In addition to model 1 (‘total education’), the presence of three other
clusters of countries in the typology may cast doubt on the relevance of
the distinction established by Bernstein. Factor 1 and factor 2 of the fac-
tor analysis (graph 2), which account for 32 per cent of the total variance
of the cloud, serve to visualise ‘total education’ countries isolated in the
left-hand quadrant, while the three remaining clusters are grouped in
the right-hand quadrant (clusters 3 and 4 almost overlap).
The four clusters are supplementary categories. This spatial position-
ing suggests that there are two different versions of the collection code
defined by Bernstein: these countries (clusters 2, 3 and 4) illustrate
the main characteristics of the collection code (with some variations
between clusters 3–4 and cluster 2). Concerning the countries in clus-
ter 3, their educational systems are founded on a rigid and hierarchical
curriculum, indicating that education (particularly elite courses and
tracks) is relatively closed to the outside world. So we may speak of
‘academic education’, with no vocational or practical education and
reserving its counselling and guidance services for students wishing to
undertake short courses. The cluster 4 is a variant of this model, with
the exception of some practical classes, either introduced recently and
often against the grain of the national educational culture (such as in
Japan and Korea) or taught in the form of optional classes (such as in
the United States).
The cluster 2 (named as the ‘producer’ model), which developed in
continental Europe (Germany, Switzerland, Austria, Belgium, Hungary,
Eigenvalues
Axis 2 Axis 2
Cluster 3
PreVocCouns
weak
Cluster 4 United Portugal
States South Korea
PreVocEdu no
RegLang France
Qual 70–75% RegEdu weak Iceland
no
weak LifSkil Japan
RegLang no weak
LifSkil high PreVocEdu New Zealand
Indis Italy
optional Soc Stud 15–30% Mexico Poland
high Sweden
Cluster 1 CivPart weak Australia
Indis 35–40% Great Britain
PreVocEdu compulsory
Axis 1Greece
CivPart Czech Republic
Qual 50–60% Denmark Ireland
PreVocCouns high high Canada
SocStud SocStud
Qual Norway Finland Hungary
medium Indis 30–35% weak
60–70% RelEdu Qual 40–50%
Spain
medium Important
Very important Germany
RelEdu Netherlands
RegLang high Switzerland
PreVocCouns
LifSkil medium Belgium
medium Austria french-speaking
PreVocEdu Cluster 2
Belgium
within track dutch-speaking
191
Figure 8.4 Correspondence analysis of curriculum variables
Switzerland, the Netherlands and the Czech Republic), is based on a

more open curriculum. However, pre-vocational courses in compulsory
education are provided exclusively to struggling students, while voca-
tional counselling (also targeted at struggling students) is not integrated
in the curriculum. Rather than aiming to ensure that students are open
to the world (the professional world, among others), pre-vocational
education and pre-vocational counselling are designed to support a cur-
riculum hierarchy. Similarly, although some life skills (such as health
and hygiene education and road safety) are taught in cluster 2 sys-
tems, these tend to be limited to primary education and are generally
integrated in broader disciplinary fields. They are also far less visible
at secondary level, either because they are left to the discretion of
schools (as in the different linguistic communities in Belgium, or in the
Netherlands) or because they are provided exclusively to students in less
highly rated pathways or courses. The Austrian hauptschule, devoted to
low-ability students, is required to provide classes in home economics
and nutrition, unlike gymnasiums (for the elites). With some rare excep-
tions, non-academic subjects in the producer model demarcate the least
valued educational pathways, while students destined for prestigious
pathways are integrated in a more focused system strictly confined to
traditional subject areas.
In the two collection models (the academic model and the producer
model), the closure of the curriculum is also reflected in the relative
unimportance of regional languages. Religious education is generally
not included, with the exception of some countries offering religious
education as an option (such as Austria, Germany and Belgium, which
makes specific provision for religious education in its constitution).
The two collection models also involve a hierarchy of courses and
classes. The producer model is characterised by explicit and highly dif-
ferentiated pathways and courses of study from the beginning of middle
school. While this distinction is less pronounced in the academic model
(mainly in Latin European countries), this model is also characterised by
ability grouping from the beginning of middle school (often an illegal
practice). In addition to specific curricula, tracks and ability grouping
practices, the two collection models are also based on vertical teacher–
student relations (i.e. distant relations and strict discipline). Lastly, in
this group of countries, individualised education is not highly devel-
oped as in model 1 (even if the USA are in an intermediary situation
with individualised teaching more developed than in this group but
less than in countries integrated in model 1). The academic model
sometimes implements remedial teaching measures targeting struggling
students, while the producer model has generally sought to develop

enrichment classes aimed at the elites whenever it has ventured into
some individualisation.
In short, there is a marked contrast between two general curricu-
lum models. The first model (‘total education’ model) is mainly found
in Northern and English-speaking countries and is characterised by
the search for a common cultural framework in compulsory education
extending beyond traditional academic disciplines, an emphasis on the
individuality of students, close teacher–student relations, less strict disci-
pline, personalised student support and a broad conception of education
that integrates vocational subjects, spirituality and minority regional
languages alongside mainstream academic disciplines.
The second model is the collection model, a system characterised by a
strict hierarchy of educational curricula, rigid teacher–student relations
governed by a relatively strict discipline and student grouping (ability
grouping and distinct pathways or courses based on ability). This model
can take two forms. The ‘producer’ model promotes close links between
education and the labour market by offering pre-vocational education
and by enforcing a strict hierarchy of schools and courses at lower sec-
ondary level. The ‘academic education’ model conceives education as a
fortress designed to transmit universal skills and knowledge in a system
marked by a strong resistance to religious, regional and linguistic differ-
ences and by an emphasis on traditional encyclopaedic knowledge. The
academic model is relatively closed to the outside world and seeks to
promote abstract universal knowledge isolated from everyday life.
From educational curricula to attitudes?
This section aims to relate the typology of educational curricula to

the prevalence of various attitudes among students aged 15. Table 8.1
presents the variables used.
Two sets of variables were created to explain these attitudes. The first
set relates to student characteristics, while the second set relates to
school characteristics (see Table 8.2). A third set of variables based on
the typology of countries designed in the study was also created. Clus-
ter 4 was eliminated because of missing data for three of the five relevant
countries.
Three-level multilevel models were estimated with a view to explain-
ing the prevalence of certain attitudes at an individual level based on
student characteristics, school characteristics and the characteristics of
educational curricula models.
Table 8.1 Attitude variables.
Name Items
Sense of belonging I feel like an outsider (or left out of things) (rev.)
I make friends easily
I feel like I belong
I feel awkward and out of place (rev.)
Other students seem to like me
I feel lonely (rev.)
Competitive I like to try to be better than other students
learning Trying to be better than others makes me work well
I would like to be the best at something
I learn faster if I’m trying to do better than the others
Instrumental I study to increase my job opportunities
motivation I study to ensure that my future will be financially secure
I study to get a good job
Co-operative I like to work with other students
learning I learn most when I work with other students
I like to help other people do well in a group
It is helpful to put together everyone’s ideas when
working on a project
Note: Items marked ‘rev.’ had their response categories reversed before scaling.
Table 8.2 Explanatory variables by level.
Student level variable Categories or variable type
Gender Girl/Boy
Hiseia Quantitative variable
Language Spoken at Home Language of assessment/ Other language
Reading Ability Estimate Quantitative variable
School level
Gender mixb Girls’ school/Boys’ school/Coeducational
schoolb
School Type Public/Private
School Autonomy Quantitative variable
School Average Reading Ability Quantitative variable
School Average Hisei Quantitative variable
Country level
Educational model total education/producer/academic
a Highest International Socio-Economic Index5 .

b Girls’ school or boys’ school: at least 90% of students are girls or boys.
Table 8.3 Variance component (in percentage) at each level for attitudes and
reading ability.
Level Instrumental Competitive Co-operative Sense of Reading Ability

Motivationa Learninga Learninga Belongingb
15 20
countries countries
Country 2.3 3.9 6.9 2.5 6.0 5.9

School 2.9 2.9 3.6 2.8 29.5 32.9
Student 94.7 93.2 89.6 94.7 64.5 61.1
a
15 countries: AUT, BELd, CHE, CZE, DEU, DNK, FIN, IRL, ISL, ITA, NLD, NZL, PRT,
SCO, SWE.
b 20 countries: AUT, BELf, BELd, CHE, CZE, DEU, DNK, ENG, ESP, FIN, FRA, GRC, IRL, ISL,
ITA, NLD, NZL, PRT, SCO, SWE.
The results of the empty models (i.e. without any explanatory vari-
ables) are shown in Table 8.3.
The table shows that the explained variance at a country level and
at the school level is low. It is interesting to note that as far as perfor-
mances are concerned (here, reading ability), the story is very different:
in that case, the school and the country level do matter much more.
However, coming back to attitudes, while they primarily operate at an
individual level, a country effect and a school effect are found, which
can be explained by including the relevant variables within the mod-
els. Tables 8.4 and 8.5 only present the results for the two attitudes
where the country level has the highest explanatory power (competitive
learning and co-operative learning).
All of the individual variables appear to have an impact on attitudes
towards competitive learning.6 Female students were found to be gen-
erally less favourable to competition than male students. It was also
found that the higher the level of reading skills, the more positive the
attitudes towards competition. In addition, at a given educational and
social level, students with a home language different from the language
of the test are more likely to view competition favourably. The spirit of
competition also tends to increase in line with the socioeconomic level
of students.
Variations in attitudes towards competition are explained signifi-
cantly better by incorporating the characteristics of the school attended.
However, very few school characteristics (among the selected ones) were
found to have an effect. The average reading level has a negative impact:
competition tends to be viewed less favourably in schools with low
average achievement levels (the relation also applies in a model in which
Table 8.4 Competitive Learning – nested multilevel models.
Reference category coef. t-value coef. t-value coef. t-value
Fixed part
constant 55.90 48.5∗∗∗ 68.40 34.8∗∗∗ 65.61 30.2∗∗∗
Student level
girl boy −5.62 −21.7∗∗∗ −5.50 −20.4∗∗∗ −5.49 −20.4∗∗∗
hisei 0.02 2.6∗∗ 0.03 4.1∗∗∗ 0.03 4.1∗∗∗
other language lang. of 3.30 6.8∗∗∗ 3.10 6.4∗∗∗ 3.10 6.4∗∗∗
assessment
reading ability 0.19 18.9∗∗∗ 0.24 21.2∗∗∗ 0.24 21.1∗∗∗
School level
school 0.01 1.1 0.01 1.1
autonomy
boys’ school coeducational 1.14 1.7 1.15 1.8
school
girls’ school coeducational 0.33 0.5 0.34 0.5
school
private school public school 0.24 0.5 0.24 0.5
average reading −0.22 −7.2∗∗∗ −0.22 −7.2∗∗∗
ability
average hisei −0.03 −1.3 −0.03 −1.3
Country level
total education producer 5.46 2.7∗
academic producer 3.21 1.3
variance variance variance
Random part
country 19.3 18.2 12.2
school 12.7 11.3 11.3
student 397.7 397.0 397.0
−2LL 237603 237499 237493
Note: ∗ p <.05, ∗∗ p <.01, ∗∗∗ p <.001.
the social climate of the school, correlated with the average academic
level, is removed). When the ‘social climate variable’ is the only variable
(i.e. when the variable ‘average achievement level’ is removed), social
climate is found to have a negative impact – that is the higher the aver-
age social level, the lower the likelihood of a favourable attitude towards
competition.
The significance of these results is not obvious, and their interpreta-
tion may require reference to phenomena known in social psychology
as the ‘big fish in a little pond’ effect. In other words, students with
high achievement levels and/or in a high-ability academic environment
Table 8.5 Co-operative Learning – nested multilevel models.
Reference category coef. t-value coef. t-value coef. t-value
Fix part
constant 54.75 45.64∗∗∗ 53.92 29.83∗∗∗ 51.35 23.79∗∗∗
Student level
girl boy 4.10 19.14∗∗∗ 3.89 17.47∗∗∗ 3.89 17.47∗∗∗
hisei −0.01 −2.62∗∗ −0.01 −1.52 −0.01 −1.52
other language lang. of −0.09 −0.22 −0.03 −0.07 −0.02 −0.05
assessment
reading ability 0.07 7.94∗∗ 0.06 6.85∗∗∗ 0.06 6.85∗∗∗
School level
school 0.01 1 0.01 0.98
autonomy
boys’ school coeducational −1.80 −3.25∗∗ −1.80 −3.24∗∗
school
girls’ school coeducational 0.63 1.1 0.63 1.1
school
private school public school −1.20 −2.66∗∗ −1.16 −2.58∗∗
average reading 0.05 1.98∗ 0.05 1.98∗
ability
average hisei −0.05 −2.94∗∗ −0.05 −2.95∗∗
Country level
total education producer 4.36 1.84
academic producer 4.22 1.45
variance variance variance
Random part
country 21.1 21.1 16.6
school 9.4 8.9 8.9
student 270.0 270.0 270.0
−2LL 227388 227357 227353
Note: ∗ p <.05, ∗∗ p<.01, ∗∗∗ p<.001.
attending a low-ability school and/or a school with a largely working-

class climate are in a relatively gratifying position and are generally more
favourable to competition (since they often emerge in an advantageous
position).
The effect of the typology proves tenuous. The third model is not sig-
nificantly different from the second model. However, all other things
being equal, student attitudes in the total education model are more
likely to be favourable to competition compared to students in the pro-
ducer model. This can be explained by the fact that these students work
more often in ability groups (or pathways based on ability) and are
therefore less exposed to competition.
The analysis of student attitudes towards cooperation and collabora-
tive work (see Table 8.5) indicates that female students and high-ability
students are more likely to favour group work and collaboration.
In terms of school characteristics, students attending schools with a
largely male intake are marginally less likely to be favourable to coop-
eration (beyond their particular gender, which appears to be indicative
of a specific climate or environment in this type of school). The same
tendency is also found in private schools. Analysis of the social or edu-
cational composition of the school indicates that social climate has a
negative impact: students in schools at the higher end of the social
spectrum are less likely to be favourable to cooperation.
At the level of the typology, students exposed to a ‘total education’
curriculum are on average more likely to be favourable to cooperation
than students in the ‘producer’ model. This can be explained by the
nature of the relations developed in the two educational models (both
peer relations and teacher-student relations) and by the emphasis on
civic participation.
The ‘total education’ model was found to be associated with student
attitudes that are more favourable to both competition and cooperation
(which are therefore not conflicting attitudes). The coexistence of these
attitudes may reflect higher levels of sociability and openness among
students. This may be linked to the human dimension of education and
learning (frequent exchanges and close contact between teachers and
students, and mixed-ability classes or flexible grouping, which tend to
promote a wider network of social experiences) and to the content of
education (active civic education, vocational guidance and counselling
available to all, general ecumenical religious education, life skills, etc.).
These less specialised items of the curriculum are assumed to foster
greater latitude in student exchanges and relations.
Discussion
Between the culturalist vision, which postulates a unity of cases, and

the universalist position, which seeks for laws that transcend national
contexts, this study provides a middle way that extends Bernstein’s find-
ings by highlighting three major curriculum models. Conceived as ideal
types, the three models underline the regularities and coherent sys-
tematisations of educational policies (namely curriculum policies). Our
analysis highlights the existence of educational systems founded on
‘total education’ characterised by a wide spectrum of curriculum sub-

jects, not based on a rigid hierarchy, thus opening education to the
outside world. At the opposite, the analysis also highlighted two cur-
riculum models defined as variations of Bernstein’s collection model and
involving a rigid hierarchy of academic disciplines. While the first cur-
riculum model emphasises academic disciplines, the second curriculum
model focuses on vocational subjects and the world of work (the ‘pro-
ducer’ model). Bernstein’s thesis is thus further refined by a substantial
body of empirical data.
Differences in curriculum and learning content appear to be increas-
ingly linked to different educational and social organisations within
schools. The ‘total education’ curriculum is generally implemented in
an environment governed by close student–teacher relations, a low level
of discipline, flexible grouping and organisations that foster exchanges
between teachers and students and among students. The ‘total educa-
tion’ model also promotes individual expression and the promotion
of individual rights. At the opposite, the collection curriculum (fur-
ther subdivided into the ‘academic education’ model and the ‘producer
education’ model) is based on human relations governed by a rigid edu-
cational system and hierarchy (i.e. both teacher–student relations and
peer relations within the framework of ability grouping or tracks based
on ability), that is relational modes and modes of organisations that
generally serve to further entrench social barriers.
The hypothesised relations between student attitudes on the one
hand and curriculum models and the attendant social organisations
within schools on the other were also confirmed (at least in part), even
if attitudes are largely determined by personal characteristics. The cur-
riculum typology was also found to be linked to variables relating to
attitudes towards collaboration and competition. The total education
seems to be the most favourable to interpersonal exchanges (whether
cooperative or competitive). By decompartmentalising curricula and
opening them to the outside world, by restructuring teacher–student
relations and by fostering more fluid relations between students of
different social backgrounds and ability levels, the ‘total education’
curriculum may thus foster the development of strong interpersonal
relations (i.e. both collaborative and competitive relations).
However, the country level of analysis and the curriculum typology
have limited explanatory power compared to the student level of anal-
ysis. That should not lead to conclude too rapidly that curricula only
have a limited impact on student attitudes. That is because PISA sur-
veys are primarily designed to measure educational performances, and
not attitudes – as shown by the limited number of questions relating

specifically to attitudes. Moreover, the assessment of attitudes is not gen-
erally the main priority of countries involved in the survey, as shown
by the fact that a number of countries opted not to answer to the
questionnaires on attitudes. Beyond the issues raised by PISA surveys,
all international comparative studies of attitudes need to consider the
varying effects of desirability norms in different countries (for exam-
ple attitudes to competition). It is also more difficult to obtain clear-cut
results based on declared variables than with performance test results.
Finally, it is important to recognise that the attitudes examined here,
conceived as the outcome or result of education, may have an ambigu-
ous status insofar as they may be viewed as an integral dimension of
education.
The curriculum typology also needs to be questioned. Insofar as it
is based on a limited number of system characteristics and only par-
tially explains the variance of national educational models, it only
accounts for visible differences between students to a very limited
extent. Therefore, the diversity of attitudes within the various types
identified remains to be explained. One solution might have been to
develop a more refined typology. However, this would have implied a
more limited number of countries per type, resulting ultimately in one
type for every country, which would certainly provide less insight!
More generally, the very possibility of measuring or assessing the
‘impact’ of education may reasonably be questioned – particularly in the
field of attitudes. The attitudes measured among students aged 15 based
on cross-section measurements such as those used in PISA surveys is the
result not only of the entire academic career to date (unlike achievement
and performance levels) but also of the global context of socialisation,
which is not a purely educational (i.e. school-based) process.
Notes
1. Cramér’s V is a measure of association between two categorical variables.
Unlike the chi-squared statistic from which it can be derived, it does not
depend on the sample size. It varies between 0 (independence) and 1 (max-
imum coincidence of categories). All the Cramér’s V values of more than
0.40 have been selected and dispatched in the graph so that the strongest
associations between categorical variables appear in the centre of the graph.
2. The p-value of Fisher’s exact test was used as a guide to select the categories
mostly closely linked to the clusters of the partition. The resulting sub-matrix
has been reordered by row and column permutations to reveal patterns of
associations between categories and the curriculum typology. This gives rise
to the display in Figure 8.3 which is an application of Bertin’s reordered matrix

(Bertin, 1981).
3. The United States is not included in this group for many reasons. Pre-
vocational counselling and religious education are not common place and
pre-vocational education is not compulsory for all students. Individualised
education is not dedicated to all the students but mainly to low and high
performers.
4. For an introduction to the method, see Le Roux and Rouanet (2010).
5. The HISEI index is the highest value of a composite index computed for
the student’s father and mother, including their occupational status, level of
education, family wealth, possessions relative to ‘classical culture’.
6. After treatment of the missing data, the models were estimated with 15 coun-
tries for the variables ‘competitive learning’, ‘instrumental motivation’ and
‘co-operative learning’, and with 20 countries for the variable ‘sense of belong-
ing’. The analyses used the individual weighting variable provided by PISA (see
PISA Data Analysis Manual, 2nd edition 2009). The analyses were conducted
using SAS PROC MIXED and R lme4 software. The attitude variables, HISEI and
reading score variables were expressed as the percentage of maximum possible
score.
References
Alexander, R., 2000. «Pédagogie, Culture et comparaison: Visions et Versions de
l’école élémentaire», Revue française de pédagogie, 142, pp. 5–19.
Bernstein, B., 1971. ‘On the Classification and Framing of Educational Knowl-
edge’ in Young, M. F. D. (ed.), Knowledge and Control: New Directions for the
Sociology of Education, London: Collier-Macmillan, pp. 47–69.
Bernstein, B., 1975. ‘On the Curriculum’ in Bernstein B. (ed.), Towards a The-
ory of Educational Transmissions: Volume III – Class, Codes and Control, London:
Routledge, pp. 79–84.
Bertin, J., 1981. Graphics and Graphic Information Processing, Berlin: De Gruyter.
Bourdieu, P., 1967. «Systèmes d’enseignement et systèmes de pensée», Revue
internationale des sciences sociales, XIX, 3, pp. 367–409.
Le Roux, B. and Rouanet H., 2010. Multiple Correspondence Analysis, London: Sage.
Mons, N., 2004. De l’école unifiée aux écoles plurielles : évaluation internationale
des politiques de différenciation et de diversification de l’offre éducative, Phd in
education science, Dijon, university of Burgundy, France.
Mons, N., 2008. «L’évaluation des politiques éducatives: apports, limites et néces-
saire renouvellement des enquêtes internationales sur les acquis des élèves»,
Revue Internationale de Politique Comparée, 14, 3, pp. 409–423.
Mons, N., 2007. Les nouvelles politiques éducatives, Paris: PUF
Osborn, M., Broadfoot, P., McNess, E. and Planel, C., 2003. A World of Difference?
Comparing Learners Across Europe, Maidenhead: Open University Press.
Young, M. F. D., 1971. Knowledge and Control: New Directions for the Sociology of
Education, London: Collier Macmillan Ltd.
9
Educational Differentiation and
Inequalities of Civic Engagement
Jan Germen Janmaat
Although the literature on political socialisation is vast, there are only

a few studies examining the role of education in mitigating or exacer-
bating inequalities of civic engagement. This is a surprising omission as
there is growing acknowledgement that disparities of participation and
democratic values are at least as harmful for democracy as low aggregate
levels of such behaviours and attitudes. Democracy cannot adequately
serve the needs and interests of all groups in society if there are major
differences in participation and civic attitudes across ethnic and social
groups. In such situations, policymakers will tend to accommodate
the desires of the politically engaged – usually the affluent and well-
educated sections of society – and ignore the interests of the politically
invisible – as a rule the low-income groups and non-native population
(Bartels, 2008; Levinson, 2010). Consequently, the latter may feel that
democracy only serves the rich and powerful and develop distinct anti-
democratic preferences. This is in turn likely to pose a serious threat to
the social cohesion of Western democratic societies.
There is thus every reason to examine how education can dimin-
ish the inequalities of civic engagement resulting from the social and
ethnic stratification of society. In this chapter I explore the influence
of two institutional properties that have been linked to disparities of
civic engagement in the literature: the degree of grouping by ability
and the degree of school autonomy. As the two institutional properties
are characteristics of national education systems I conduct an interna-
tional comparative study using the data of two international surveys
to measure the civic engagement levels of two groups of adolescents
and explore the relations between the two institutional properties and
inequalities of civic engagement among these two groups. According
202
Jan Germen Janmaat 203
to Van de Werfhorst and Mijs (2010), who conducted an extensive

review of the literature on educational differentiation and inequalities
of achievement, there are as yet no studies examining the link between
educational differentiation and inequalities of civic engagement in a
cross-national comparative manner.
The chapter starts with a discussion of the contested concept of
civic engagement. Subsequently it reviews the literature on educa-
tional differentiation and civic engagement. Next the data sources and
methodology are explained. The final two sections present and discuss
the results of the analyses.
Civic engagement
Civic engagement, civic culture, civic competences and active citizen-

ship are all fuzzy terms because they are contested. In the academic
literature on citizenship two schools of thought can broadly be identi-
fied (Janmaat, 2012). Whereas the first sees traditional ways of political
and civic participation, institutional trust, law abidance and a sense of
civic duty as key civic virtues (e.g. Bellah et al., 1985; Putnam, 2000),
the second puts much greater emphasis on alternative ways of par-
ticipation (including protest and participation in social movements),
a critical attitude towards authority, tolerance and gender equality as
aspects of citizenship that underpin modern liberal democracies (e.g.
Inglehart, 1990; Lichterman, 1996). Moreover, these different views on
what makes a good citizen are highly politically charged as the first view
is usually embraced by conservative politicians while the second is more
popular among left-wing parties.
The contested nature of civic engagement obviously does not make it
any easier to investigate the phenomenon. I acknowledge this difficulty,
but as the primary purpose of this chapter is to investigate the relation
between educational differentiation and civic engagement, exploring
this debate any further would undermine the focus of the chapter.
Hence I take a pragmatic approach to defining civic engagement by
selecting components from the two schools of thought that are consid-
ered by many scholars to constitute key indicators of the phenomenon.
These are complemented by indicators drawn from other citizenship
approaches, for example those highlighting the crucial role of civic effi-
cacy (Almond and Verba, 1963) and civic knowledge and skills (Galston,
2001; Hoskins et al., 2008). Following Hoskins et al. (2008) I understand
civic engagement to be a multidimensional phenomenon comprising
a cognitive (knowledge and skills), an affective (values, attitudes) and a
behavioural dimension. I will measure the concept with the following

indicators:
• Civic knowledge and skills

• Civic efficacy
• Intention to vote
• Institutional trust
• Gender equality
• Ethnic tolerance.
Ability grouping and school autonomy
Differentiation in the compulsory education phase is at a minimum

in systems with comprehensive education and a uniform curriculum
sanctioned by the central state. Conversely it is at a maximum level
in systems which practice grouping by ability and allow schools total
freedom in teaching the subjects that matter for the development of
civic competences. It makes good sense to expect disparities of civic
engagement to be most effectively redressed in systems with minimal
differentiation, and this has certainly been well argued for the notion of
comprehensive education (or de-tracking as it is known in the United
States).
Comprehensive education is likely to have an equalising effect for
three reasons. Firstly, all students, whatever their backgrounds, are
exposed to the same curriculum in comprehensive systems (Van de
Werfhorst, 2007). This gives students of disadvantaged backgrounds
who have not been socialised in societal and political affairs by their
parents a chance to catch up with their more privileged peers. By con-
trast, inequalities in civic engagement are likely to become only more
pronounced in systems which select on the basis of ability due to cur-
riculum differences across tracks or schools. There is evidence that the
more prestigious academic tracks offer a different and more advanced
version of citizenship education than the (pre)vocational tracks. For
instance, in a qualitative study in the Netherlands, a country with
an early selection system, Ten Dam and Volman (2003) found that
while academic school types (HAVO, VWO) focused on the transmis-
sion of knowledge and the development of independent and critical
thinking skills, prevocational schools (VMBO) devoted their social stud-
ies lessons on fostering social competences, appropriate behaviour and
self-confidence.
Secondly, comprehensive education can be assumed to equalise civic

engagement by giving children of disadvantaged backgrounds the
opportunity to learn from their more privileged classmates. The impor-
tance of learning from peers (the so-called peer effect) as a way to
compensate for a lack of family resources and support has been well
demonstrated in the literature on student achievement. Many studies
in this field have found that mixed ability schooling reduces the effect
of individual social background on performance and can thus be said
to be conducive for social mobility (Green et al., 2006; Brunello and
Checchi, 2007; Van de Werfhorst and Mijs, 2010). There is no reason
to assume that the peer effect would be any different for civic engage-
ment (Hyland, 2006). Thus, youngsters who have received little input
from their parents in terms of democratic socialisation can acquire civic
competences by learning them from their more advantaged classmates.
It needs to be stressed that this process of learning in mixed ability
classrooms does not only involve a cognitive component. As students
of disadvantaged backgrounds get in contact with children of many
different backgrounds, they learn to understand and appreciate differ-
ent points of view and how to defend their own point of view. Some
scholars have thus asserted that mixed schools are laboratories of demo-
cratic debate and provide the most comprehensive training for teenagers
to participate later on as adults in a diverse democratic society (Rubin,
2003; Hyland, 2006; Rubin, 2007).
In early selection systems youngsters also learn from their classmates
but here the peer effect is likely to only exacerbate inequalities. Group-
ing by ability systems are prone to enhance social and ethnic segregation
because children from disadvantaged and immigrant backgrounds are
disproportionately allocated to the low status (pre)-vocational tracks
(Hallinan, 1994; Loveless, 1999; Crul and Vermeulen, 2003). Thus, these
children have little opportunity to come in contact with children from
middle- and upper-class backgrounds for whom civic engagement was
a self-evident part of their upbringing. The result is that youngsters
in different tracks experience not only different curricula but also dif-
ferent peer effects leading them to develop different life worlds with
diverging norms and values (Van de Werfhorst, 2007; Janmaat and
Mons, 2011). Researching youngsters in Flanders’ early selection sys-
tem, Stevens (2002) indeed found huge cross-track differences in civic
and socio-political attitudes. While students in the academic track dis-
played high levels of support for democratic values, students in the
vocational track stood out for their ethnocentric, authoritarian and
‘tough on crime’ views. Analysing data of the 1999 Civic Education

Study on 14-year-olds, Janmaat and Mons (2011), moreover, found that
disparities in ethnic tolerance between native and immigrant groups
were smallest in states with comprehensive systems.
Given the widely assumed equalising effect of comprehensive educa-
tion, it is not surprising to find many scholars advocating if not mixed
ability classes for all subjects then at least common classes for citizen-
ship and social studies (Hyland, 2006; Rubin, 2007; Van de Werfhorst,
2009). I derive the following hypotheses from the above discussion:
1.a. Differences between classrooms in civic engagement are smaller in

comprehensive systems than in systems which practice grouping
by ability (either between or within schools).
1.b. Social and ethnic background have a smaller effect on civic engage-
ment in comprehensive systems than in systems which practice
some form of ability grouping.1
By comparison to the studies on grouping by ability there are only a few

studies exploring the effect of that other dimension of differentiation –
decision-making structure – on disparities of civic engagement. The
obvious argument to make is that decentralised systems providing
schools with substantial autonomy in curriculum matters are likely to
yield greater inequality in civic engagement than centralised systems.
This inequality may well not only concern greater differences between
schools but also greater differences between social and ethnic groups
in civic engagement if the school system is socially and ethnically segre-
gated. In line with this proposition Wolf and Macedo (2004) have argued
that a national regulatory framework helps prevent school autonomy
from showing its divisive effects. Shared values can be maintained, they
contend, under conditions of school pluralism provided there is some
minimal degree of national regulation applying to all schools.
There are surprisingly few empirical studies verifying these proposi-
tions with regards to civic engagement. This is all the more surprising as
there is quite solid evidence from the field of achievement that central-
isation reduces inequalities in achievement. Mons (2007), for instance,
found that a centralised structure, particularly concerning the curricu-
lum, certification of teachers and teacher recruitment, is associated
with relatively small between-school and between-student disparities in
academic achievement. Moreover, several studies have found standard-
isation (i.e. central examinations, a national curriculum and standard-
ised school resources) to diminish the effect of social background on
achievement (see the extensive review by Van de Werfhorst and Mijs,

2010). To the knowledge of the author, the only study examining the
link between centralisation and inequalities in civic engagement is the
aforementioned study by Janmaat and Mons (2011). They found that
disparities of ethnic tolerance and patriotism across ethnic and social
groups are smaller in systems with some form of national regulation (i.e.
non-federal systems). The current study builds on this research by using
more recent datasets (the 2009 ICCS study and the 2000 CIVED study)
and by examining the effect of school autonomy (i.e. the inverse of cen-
tralisation) on many more indicators of civic engagement. I extract the
following hypotheses:
2.a. Differences between classrooms in civic engagement are larger in

systems with more school autonomy.
2.b. Social and ethnic background have a larger effect on civic engage-
ment in systems with more school autonomy.
Data and methodology
I utilise two sources of survey data to measure the civic engagement of

two groups of adolescents: the 2009 International Civics and Citizen-
ship Education Study (ICCS) conducted among Grade Eight students
(ages 13 and 14) and the 2000 Civic Education Study (CIVED) con-
ducted among upper secondary students. The ICCS study is a survey
with worldwide coverage (38 countries in total; 25 European). In each
participating country at least 3000 students from at least 150 schools
were selected in a nationally representative sample (Schulz et al., 2010).
The data for CIVED were collected in 16 countries, 11 of which were
European, by means of nationally representative samples ranging in size
between 1270 students in Switzerland to 4050 students in Poland. The
average age of the participating students varied from 16.6 in Latvia to
19.4 in Denmark – see Amadeo et al. (2002) for more information on
the data collection. In both surveys one whole classroom per school
was sampled. I selected the European states in both surveys for further
study.
I made use of a number of ready-made scales created by the ICCS
methodological experts to tap the aforementioned indicators of civic
engagement. The scale representing civic knowledge and skills is a com-
posite index comprising the answers to 79 test items on civic society,
principles, participation and identities (Schulz et al., 2010). The other
scales are composites of questionnaire items. They are all internally
coherent at both the pooled level and at the level of individual coun-
tries, which indicates that they are reliable in terms of conceptual
equivalence across Europe (Schulz et al., 2011). Appendix 9.1 shows
these scales and the items included in them. Only the indicator intention
to vote was captured with a single item, phrased as follows: ‘When you
are an adult, what do you expect that you will do?’ ‘Vote in national
elections’ (I will certainly not do this – probably not do this – probably
do this – certainly do this).
To measure social background I developed a scale consisting of the
items (1) number of books at home, (2) education level of mother and
(3) education level of father. Reliability analysis showed this scale to
be internally coherent in each of the participating countries. I further
created the item ‘immigrant status’ with values 0 (native) and 1 (first-
and second-generation immigrant) to indicate ethnic background.
The OECD report on the results of the 2009 edition of the Programme
of International Student Assessment study (PISA) was relied upon to
develop a composite index of comprehensivisation. I used data on the
age of first selection and the percentage of students in schools where
the principal reported no ability grouping in any subject (OECD, 2010,
pp. 203, 212) to construct this index. Although the age when young-
sters are first separated and allocated to different post-primary schools
on the basis of ability is the most obvious indicator of the degree of com-
prehensive education (and has been used as such by many studies), it is
incomplete as it does not take ability grouping inside otherwise compre-
hensive schools into account. It is difficult to know beforehand which
of the two modes of differentiation is more important. If in a compre-
hensive system all the schools practice ability grouping the degree of
segregation by performance may well be as strong as in a system that
selects early but does not practice grouping by ability within schools. For
this reason the two indicators contributed equally to the construction
of the index. The index represents the sum of the standardised scores
(Z-scores) of the indicators. Thus, countries which select early and prac-
tice grouping by ability on top of that have very low values on the index.
Countries with comprehensive systems and minimal grouping by ability
have very high values.
I used data from the ICCS national context file to construct a mea-
sure for centralisation. The data in this file represent the answers of
national experts to questions on the nature of citizenship education and
the characteristics of the education system more broadly in each of the
participating countries (Schulz et al., 2011, pp. 41, 42). I developed a
composite measure based on the following three items:
“How much autonomy do schools have . . .”
1. in curriculum planning?
2. in curriculum delivery?
3. in choice and use of textbooks?
Categories: 0 – no autonomy; 1 – some autonomy; 2 – complete

autonomy.
The measure, labelled ‘school autonomy’, represents the sum of the

answers to these items and consequently has values ranging from 0 –
‘minimum school autonomy’ to 6 – ‘maximum school autonomy’. Reli-
ability analysis showed that it is an internally coherent construct with
an alpha value of 0.63.2
I calculated the Intraclass Correlation Coefficients (ICCC) of the six
civic engagement indicators to assess the between-classroom relative to
the within-classroom variation in these indicators. An ICCC of some
individual-level measure represents the outcome of the between class-
room variance divided by the sum of the between-classroom and the
within-classroom variance. Its values range between 0 and 1 with 1
meaning that all the variation is between classrooms and not within
classrooms (i.e. all students have the same score within classrooms) and
0 indicating that all classrooms have the same mean value and students
vary maximally within classrooms. It thus provides an excellent measure
of cross-classroom inequality in civic engagement. To assess the effect of
social and ethnic background on civic engagement, I used the propor-
tion of the variance in the six indicators of civic engagement explained
by each of these variables.
Results
I begin by reporting the findings for the Grade Eight students. Table 9.1
provides data on the between-classroom differences in civic engagement
for all six indicators. These data represent a multiplication of the ICCC
by 100 and thus denote the percentage of the total variance which is
across classrooms. The last four columns show country rankings on the
degree of comprehensivisation and school autonomy.
European countries appear to vary substantially in the degree of
inequality across classrooms in civic knowledge. While more than
50 per cent of the total variance in civic knowledge is between class-
rooms in Bulgaria and the Netherlands, less than 20 per cent is so in
210
Table 9.1 Between-classroom differences in six indicators of civic engagement.
Civic Civic Intention Institutional Gender Ethnic Comprehen- School

knowledge efficacy to vote trust equality tolerance siviation autonomy
BGR 50.9 SWE 16.0 CHE 18.5 BGR 20.0 IRL 20.0 SWE 18.9 GRC 2.79 ESP 6
NLD 50.3 NLD 13.2 NLD 16.3 CHE 18.6 NLD 19.8 DNK 13.6 LVA 1.79 CZE 5
BFL 38.6 ITA 11.1 IRL 15.5 RUS 14.8 BGR 18.3 ENG 13.2 POL 1.73 NLD 5
RUS 33.5 SVN 10.7 ENG 15.4 POL 14.1 BFL 13.3 ITA 13.2 DNK 1.54 BFL 5
CHE 31.6 BFL 10.6 LVA 14.5 LTU 13.7 EST 13.3 BFL 12.6 FIN 1.14 RUS 5
IRL 31.1 BGR 10.5 SWE 12.9 SVK 12.3 LVA 13.2 EST 12.3 ESP 0.99 ENG 5
ENG 29.7 RUS 9.6 CZE 12.8 NLD 11.6 LTU 13.0 AUT 11.6 EST 0.69 SWE 5
AUT 29.4 ESP 8.9 BGR 12.3 LVA 11.2 ENG 12.3 NLD 11.5 NOR 0.33 NOR 5
ESP 27.2 GRC 8.9 BFL 12.2 ITA 11.0 ESP 12.2 LVA 11.1 SWE 0.29 FIN 5
LTU 27.2 LTU 8.7 LTU 12.0 NOR 10.4 GRC 12.2 CHE 10.5 SVN 0.23 DNK 5
EST 25.6 FIN 8.6 POL 12.0 BFL 10.1 CHE 11.9 ESP 10.4 ITA 0.20 POL 5
SVK 25.6 IRL 8.5 AUT 12.0 DNK 10.0 SVK 11.8 RUS 10.4 BGR −0.26 SVK 4
LVA 24.6 POL 8.5 ESP 11.9 ESP 9.9 AUT 11.7 IRL 9.9 BFL −0.32 ITA 4
POL 24.4 LVA 8.2 RUS 11.4 SVN 9.6 RUS 11.5 BGR 9.9 LTU −0.62 SVN 4
CZE 23.9 AUT 8.0 EST 10.9 IRL 9.2 POL 10.3 NOR 9.6 RUS −0.64 IRL 4
GRC 22.3 SVK 8.0 NOR 10.7 GRC 9.2 ITA 10.0 SVK 9.5 ENG −0.97 LVA 4
ITA 19.2 ENG 7.7 SVK 10.4 SWE 8.8 CZE 9.0 POL 9.5 IRL −1.35 AUT 3
SWE 16.0 NOR 7.7 GRC 9.6 ENG 8.6 SWE 8.5 SVN 9.4 AUT −1.39 BGR 3
DNK 14.7 CHE 7.1 FIN 9.3 CZE 8.2 SVN 8.4 FIN 9.2 SWI −1.80 LTU 3
NOR 12.8 EST 7.1 DNK 9.3 AUT 8.0 DNK 7.9 LTU 9.0 CZE −2.01 EST 3
SVN 10.7 DNK 6.2 ITA 8.8 EST 7.8 FIN 7.5 GRC 8.3 NLD −2.07 GRC 2
FIN 10.6 CZE 5.3 SVN 8.5 FIN 7.6 NOR 6.7 CZE 7.1 SVK −2.24 SWI 0
Note: The figures for the six indicators of civic engagement represent the variance between classrooms as a percentage of the total variance (i.e.
ICC × 100).
Italy, Sweden, Denmark, Norway, Slovenia and Finland. Cross-classroom

disparities are less pronounced on the attitudinal indicators of civic
engagement with percentages ranging between 5 and 20 across these
indicators. Interestingly, the country rankings on inequalities of civic
knowledge and intention to vote are almost the exact opposite of
that of comprehensivisation as the countries topping the league of
comprehensivisation trail the leagues of civic knowledge and intention
to vote and vice versa. The cross-country patterns on these two indica-
tors of civic engagement thus provide strong support for the proposition
that comprehensive education reduces inequalities of civic engagement
across schools (i.e. Hypothesis 1.a). The patterns on the remaining four
attitudinal indicators of civic engagement are less clear. Particularly that
of ethnic tolerance does not conform to the expected regularity. In con-
trast to the prediction of Hypothesis 1a, cross-classroom differences in
ethnic tolerance are most pronounced in Denmark, a country with a
robust comprehensive system, and smallest in the Czech Republic, a
country with an early selection system. Classroom inequalities in ethnic
tolerance may well be more of a function of the salience of immigra-
tion as the countries topping the league have all experienced substantial
immigration over the last two decades (Estonia excepting) while those
trailing the league are emigration or otherwise ethnically homogenous
countries (Czech Republic, Greece, Lithuania, Finland, Slovenia and
Poland). No clear relationship emerges between inequalities of civic
engagement and degree of school autonomy either. For instance, the
Czech Republic is second in the ranking order on school autonomy
but trails the league tables on cross-classroom inequalities in five of
the six indicators of civic engagement. Thus, school autonomy in cur-
riculum and textbook matters does not necessarily translate into larger
cross-classroom gaps in civic engagement.
I performed correlations on the data of Table 9.1 to assess the
connections between educational differentiation and inequalities of
civic engagement more precisely (see Table 9.2). These correlations by
and large confirm the picture emerging from Table 9.1. As expected,
comprehensivisation indeed shows a strong and significant nega-
tive relationship with cross-classroom inequalities of civic knowledge
(r = –.46; p = .03; N = 22) and intention to vote (r = .50; p = .02; N = 22).
It is also negatively related with inequalities of institutional trust and
gender equality, but these relations are not significant. Thus, with
the exception of inequalities in ethnic tolerance (and to some extent
civic efficacy), comprehensivisation is linked in the expected way to
classroom differences in civic engagement but this relation can only
Table 9.2 Educational differentiation and inequalities of civic engagement

among grade eight students (correlations).
Comprehensivisation School
autonomy
Between classroom differences in:

Civic knowledge −.46(∗ ) −.16
Civic efficacy .01 .19
Intention to vote −.50(∗ ) −.26
Institutional trust −.19 −.43(∗ )
Gender equality −.35 −.21
Ethnic tolerance .07 .19
The effect of social background on:
Civic knowledge −.24 −.02
Civic efficacy .47(∗ ) .29
Intention to vote −.31 .12
Institutional trust .09 .08
Gender equality −.06 .16
Ethnic tolerance .09 .13
The effect of ethnic background on:
Civic efficacy −.25 .27
Intention to vote .01 −.33
Institutional trust −.01 −.01
Gender equality .36 −.04
Ethnic tolerance −.34 −.05
Note: ∗ p < .05; ∗∗ p < .01; N = 22 for all correlations.
be convincingly demonstrated for two indicators. Table 9.2 further

confirms the above observation that school autonomy is not related
to cross-classroom inequalities of civic engagement, except for one
indicator – institutional trust. Remarkably, the correlation is negative,
indicating that the more school autonomy a country has the smaller are
the disparities between classrooms in institutional trust. However, fur-
ther analysis showed that this correlation was entirely shaped by outlier
Switzerland. Omitting Switzerland reduces the negative correlation to a
non-significant level and has a mitigating effect on the other correla-
tions as well. Thus, as noted before minimal school autonomy need not
result in greater value homogeneity across schools.
Do comprehensive systems and limited school autonomy help
to diminish the effects of social and ethnic background on civic
engagement? In other words, do they equalise engagement across
social and ethnic groups? Table 9.2 shows correlations between
comprehensivisation and school autonomy on the one hand and the

effects of social and ethnic background on the six civic engagement
indicators on the other.3 I would expect comprehensivisation to be
negatively correlated, and school autonomy to be positively corre-
lated with the social and ethnic background effects. It turns out that
the two indicators of educational differentiation are not significantly
related with any of these effects, except for the positive correlation of
comprehensivisation with the effect of social background on efficacy.
In other words, comprehensivisation does not equalise civic engage-
ment across social and ethnic groups and even seems to magnify cross-
class inequalities in efficacy. Similarly, centralisation does not reduce
any cross-class and cross-ethnic inequalities either.
Why are comprehensivisation and centralisation not doing the
expected job of equalising civic engagement outcomes across social and
ethnic groups? Possibly, the respondents of the ISSC study (Grade Eight
students, meaning 13- and 14-year-olds) were simply too young to have
experienced any effect of the curriculum yet. One of the subjects that is
likely to impact on civic engagement is citizenship education. In coun-
tries where this is taught as a separate subject, the teaching of the
subject usually does not start until Grade Seven (age 12) and the bulk
of it happens between ages 14 and 16 (Eurydice, 2005, p. 20). Instruc-
tion in language and arithmetic, in contrast, starts very early. Thus,
while 13- and 14-year-olds have been exposed to subject matter that is
important for their literacy and numeracy skills from the moment they
entered primary schools, they have hardly experienced relevant curricu-
lum input yet regarding the development of their civic competences.
This may explain why comprehensivisation is linked to more equality
in achievement across social and ethnic groups (as noted before) but not
to more equality in civic engagement.
If comprehensivisation and centralisation simply have not been able
to do their equalising work because of an absence of relevant curricu-
lum content until Grade Eight, one would expect them to do so once
youngsters have experienced several years of relevant curriculum input.
To test this conjecture I performed the same analyses on the CIVED
2000 data, that is the data collected among upper secondary students.
The CIVED study included the same scales as ICCS apart from civic
efficacy and intention to vote. I selected the scale ‘expected political
participation’ to replace these missing scales (see Appendix 1 for its com-
position). I have to note that no states have undifferentiated systems
anymore in upper secondary. Thus, all students have been allocated to
various academic, vocational and professional tracks. Yet, the students
vary substantially in their educational history. While those from states

with full comprehensive systems have only recently been grouped on
the basis of ability, students from states with early selection systems have
a long history of attending tracked schools and classes. Table 9.3 ranks
the countries by cross-classroom inequalities in civic engagement and
by the two measures of educational differentiation. The 11 European
countries participating in the Cived Upper Secondary study appear to
vary substantially in disparities of civic engagement. They also differ
maximally in comprehensivisation as the group includes countries with
early selection systems (Switzerland and Czech Republic) and coun-
tries with pronounced comprehensive systems (Portugal, Latvia, Poland,
Denmark). The variation between countries in terms of school auton-
omy is also quite pronounced, ranging between the values 0 (minimal
school autonomy) and 5 (substantial school autonomy).4
Analogous to Table 9.2, Table 9.4 shows correlations between the
two dimensions of educational differentiation on the one hand and
the three modes of inequalities of civic engagement on the other (i.e.
cross-classroom disparities and the effects of social and ethnic back-
ground).5 We can see that except for two correlations all the others are
not significant. Comprehensivisation seems to mitigate cross-classrooms
disparities as it shows negative correlations with all five indicators of
civic engagement, but none of these correlations are significant. School
autonomy seems to enhance the effect of social background on civic
engagement as it is positively correlated with all five indicators. But yet
again the correlations are not significant, except one. No consistent pat-
terns emerge for the remaining correlations. Possibly, if more countries
had participated in the Cived Upper Secondary Study more relation-
ships would have been significant. Thus, although the findings from
the Upper Secondary study are at least in line with the earlier tentative
conclusion that comprehensivisation indeed helps to reduce classroom
segregation in civic engagement, the lack of statistical significance of
the relationships means that the puzzle of the absence of a link between
educational differentiation and inequalities of civic engagement is still
in place.
Discussion and concluding remarks
We know from the literature on achievement that both comprehen-

sivisation and centralisation reduce inequalities, particularly those
across social groups (Van de Werfhorst and Mijs, 2010). Do they also
equalise disparities of civic engagement? This study, which explored this
Table 9.3 Between-classroom differences in civic engagement (upper secondary students).
Civic Expected Institutional Gender Ethnic Comprehen- School

knowledge political trust equality tolerance siviation autonomy
participation
SVN 47.7 SVN 18.2 SWE 17.9 SWE 22.1 SWE 27.9 POR 1.94 CZE 5
RUS 46.2 RUS 16.3 RUS 16.7 NOR 21.8 EST 26.6 LVA 1.79 RUS 5
POL 44.9 SWE 15.8 SVN 14.1 CHE 20.5 NOR 26.1 POL 1.73 SWE 5
CHE 43.5 NOR 14.9 LVA 13.8 POL 16.9 LVA 24.3 DNK 1.54 NOR 5
SWE 43.5 LVA 11.9 NOR 13.4 CZE 15.0 CHE 21.8 EST 0.69 RUS 5
CZE 40.4 POR 11.6 CZE 10.3 SVN 14.7 DNK 17.7 NOR 0.33 DNK 5
NOR 36.6 CZE 11.1 CHE 10.2 LVA 13.3 RUS 17.3 SWE 0.29 POL 5
LVA 35.8 CHE 9.7 POR 9.4 POR 12.7 CZE 14.7 SVN 0.23 SVN 4
EST 19.8 EST 9.6 EST 9.3 DNK 11.3 POL 12.4 RUS −0.64 LVA 4
POR 19.7 DNK 8.7 POL 9.2 RUS 9.1 SVN 8.1 SWI −1.80 EST 3
DNK 15.2 POL 7.7 DNK 7.2 EST 7.0 POR 6.4 CZE −2.01
215
Table 9.4 Educational differentiation and inequalities of civic engagement

among upper secondary students (correlations).
Comprehensivisation School
autonomy
Between classroom differences in:

Expected political participation −.20 .26
Institutional trust −.20 .25
Gender equality −.25 −.11
Ethnic tolerance −.13 −.18
The effect of social background on:
Civic knowledge .15 .89(∗∗ )
Expected political participation −.16 .50
Gender equality −.09 .46
Ethnic tolerance −.17 .18
The effect of ethnic background on:
Civic knowledge .11 .08
Expected political participation −.35 −.26
Gender equality .00 .34
Ethnic tolerance −.43 −.85(∗∗ )
Note: ∗ p < .05; ∗∗ p < .01; N = 11 for the correlations with comprehensivisation; N = 10 for
the correlations with school autonomy.
relationship among lower and upper secondary students, suggests that

they do so only partially. Countries with strict comprehensive systems
indeed show smaller disparities of civic engagement across classrooms,
but inequalities of civic engagement across social and ethnic groups are
equally pronounced in such countries. Centralisation (as tapped with
the inverse indicator of school autonomy) goes hand in hand with
smaller disparities of civic engagement across social groups, but this
relationship is only visible among upper secondary students and is not
based on significant correlations. Centralisation is not related to any
other inequalities of civic engagement, neither across classrooms nor
across social and ethnic groups. Thus of the four hypothesised rela-
tionships only one is supported wholly (Hypothesis 1.a) while another
(Hypothesis 2.b) is only partially backed up by evidence.
As the expected relationships by and large fail to emerge among the
upper secondary group as well, there is not more than marginal support
for the aforementioned proposition that the absence of relationships
among the younger group of respondents is due to this group not yet
having been exposed to relevant curriculum content in the same way

that the upper secondary group has. This means we are back at Square
One. Discounting the link between school autonomy and inequali-
ties across social groups among upper secondary students, why are
the expected relationships not visible among upper secondary students
either? Four reasons can be proposed. The first two can be labelled pos-
itive in that they suggest that inequalities of civic engagement can be
mitigated with fairly straightforward educational interventions. If the
latter two reasons apply, a much more comprehensive approach going
beyond education would seem to be required as a means to reduce these
inequalities.
The absence of the expected relationships among upper secondary
students could first of all be due to the low status of citizenship educa-
tion and related subjects by comparison to language and mathematics.
As a rule, citizenship education is not given the same weight in the
curriculum and is not assessed as rigorously as the last-named subjects.
Students thus have fewer contact hours and have fewer incentives to
learn the subject matter by comparison to language and mathematics.
This relative neglect of the political socialisation function of public edu-
cation, possibly due to a desire of many schools to remain ideologically
neutral, may thus have the unintended consequence of perpetuating
civic inequalities across social and ethnic groups. That civic education
might well be effective in reducing inequalities as long as enough of
it is provided is confirmed by Netjes et al. (2011). In their study of
15 secondary schools in the Netherlands, they found that inequali-
ties in socially responsible behaviour across students of different social
backgrounds were smaller in schools that paid a great deal of atten-
tion to civic education. However, she did not find such mitigating
effects for democratic behaviour and responsible behaviour. Similarly,
Langton and Jennings (1968) found that students of African American
background benefited much more from civic education in terms of
improving their political knowledge and skills levels than did white
American students. As the former constituted a severely deprived group
in the 1960s (as they still do today), civic education helped to equalise
political knowledge across racial groups. The existing national-level
data on citizenship education, such as that provided by the Eurydice
report on citizenship education across Europe (Eurydice, 2005) and the
national context file of the ISSC study, is not detailed enough in terms
of the information provided on the importance, volume and modes of
assessment of citizenship education to explore the proposed reason fur-
ther in the current cross-national study. Future research could however
very well use the ISSC school and teacher data files to explore intra-
nationally whether schools taking the political socialisation task very
seriously succeed in mitigating social and ethnic inequalities in civic
engagement. Indeed, there might well be more variation in the civic
education provided between schools within countries (particularly so in
countries with federal systems and high degrees of school autonomy)
than between countries.
Another reason why the expected relationships are not observed
among upper secondary students could be that youngsters only become
interested in social and political issues in late adolescence. Psycholo-
gists for instance have argued that teenagers develop political identities,
values and beliefs much later than identities and beliefs relating to gen-
der, social roles, occupation and religion (Archer, 1982; Waterman, 1982;
Goossens, 2001). If this is true, citizenship education in the lower sec-
ondary phase may well not be very effective as youngsters do not pick
up the subject matter yet. By the time they develop an interest in soci-
etal affairs they have already been assigned to different tracks in upper
secondary where they are likely to receive different kinds and volumes
of citizenship education. Thus, at a stage when they are most recep-
tive to citizenship education they are exposed to differential curricula,
which are thus likely to only reinforce rather than mitigate inequali-
ties of civic engagement across schools, and social and ethnic groups.
The implication for policy would be to focus citizenship education on
upper secondary and ensure it is provided in a uniform manner across
different tracks. Again, due to insufficiently detailed data on citizenship
education in upper secondary, this proposition could not be examined
any further in the current study.
Thirdly, civic engagement may after all be very different from achieve-
ment regarding how it is affected by educational differentiation. Possibly
youngsters as long as they are in the education system do not become
aware of the inequalities and processes of exclusion in the wider soci-
ety and consequently have an optimistic outlook on life, expressed as
high levels of civic engagement. Perceptions of equality and fairness
may paradoxically be even higher in states with early selection systems
because the greater homogeneity within schools may lead students in
these states to falsely assume that society at large reflects their school
environment. Possibly, they only start to become more aware of unequal
opportunities once they have made their first strides on the labour mar-
ket. If true, the key stage in terms of influencing their civic engagement
would be the school to work transition and the first few years of working
experience. Education may well not be able to sustain or revive the

civic engagement levels of youngsters who feel excluded or have other
negative experiences in this key period.
Lastly, the absence of the expected effects could be due to the increas-
ing salience of civic and political engagement as an identity marker
distinguishing the well-educated from the poorly educated. The rise
of education in becoming the most important cleavage in society dis-
placing divisions of religion, ethnicity, class and gender has been well
documented in the literature (e.g. Rosanvallon, 1995; Bovens and Wille,
2010) and some have suggested that interest in politics and social
affairs has become a tool for different educational groups to mark their
boundaries and discipline members internally (Stevens, 2002). While
political and civic engagement is the norm for well-educated people, it
is unacceptable for poorly educated groups as it is interpreted as sign
of defection, of rejection of the group and its values. Stevens (2002)
argues that this difference in civic engagement is merely a component
of a more fundamental difference in outlook on life and society. The
poorly educated tend to view education and society as a whole as unfair,
as they are seen as not offering any opportunities for them no matter
what efforts they put in. In response they develop a counter culture
that rejects the status symbols of the dominant culture and has its own
indicators of prestige (Ibid.; Willis, 1997). In contrast, the well-educated
believe that society is meritocratic, that citizens can shape their own
fate both individually and collectively and that political engagement
therefore makes sense.
If civic engagement has indeed become a status symbol for the well-
educated and a no-go area for the poorly educated, citizenship education
(and any other kind of programs specifically intended to enhance
civic competences) by itself is unlikely to be very effective in reducing
inequalities of civic engagement. As it has also been acknowledged that
civic engagement is the product of much more than education alone
(Torney-Purta et al., 2001; Hoskins et al., 2012), a much more compre-
hensive approach targeting the educational divide as a whole might in
that case well be necessary to raise the participation and engagement
levels of the disadvantaged groups in society.
Due to limitations of data and space the four proposed reasons could
not be explored further in the current study. It is up to future research
to investigate whether a robust and uniform program of citizenship
education in lower/upper secondary or a much more comprehensive
intervention is needed to reduce inequalities of civic engagement.
Notes
1. The effects of social and ethnic background on civic engagement can be seen
as a proxies for disparities of civic engagement across social and ethnic groups.
The stronger the effects, the larger the disparities.
2. I realise that schools can have autonomy in other matters than the curricu-
lum and textbooks, for instance in teacher recruitment and finances. I focused
on curriculum and textbooks since they are most directly linked to civic
engagement.
3. These effects (as indicated by proportions of explained variance) are not
provided in the current study. The authors will provide them upon request.
4. This data on comprehensivisation and school autonomy is the same as that
displayed for the Grade Eight students (see Table 9.1).
5. Thus, the correlations have been performed on the data of Table 9.3 concern-
ing the classroom disparities of civic competences. The effects of social and
ethnic background are not shown in this table. These will be provided to the
reader upon request.
References
Almond, G. A. and S. Verba (1963) The Civic Culture. Princeton: Princeton
University Press.
Amadeo, J., J. Torney-Purta, R. Lehmann, V. Husfeldt and R. Nikolova (2002)
Civic Knowledge and Engagement: An IEA Study of Upper Secondary Students in Six-
teen Countries. Amsterdam: The International Association for the Evaluation of
Educational Achievement.
Archer, S. (1982) ‘The Lower Age Boundaries of Identity Development’ Child
Development, LIII, 1331–56.
Bartels, L.M. (2008) Unequal Democracy: The Political Economy of the New Gilded
Age. New York: Russell Sage Foundation.
Bellah, R., R. Madon, W.M. Sullivan, A. Swindler, and S.M. Tipton (1985) Habits
of the Heart. Berkeley, CA: University of California Press.
Bovens, M. and A. Wille (2010) Diplomademocratie. Amsterdam: Bert Bakker.
Brunello, G. and D. Checchi (2007) ‘Does School Tracking Affect Equal-
ity of Opportunity? New International Evidence’ Economic Policy, LII,
781–861.
Crul, M. and H. Vermeulen (2003) ‘The Second Generation in Europe’ Interna-
tional Migration Review, XXXVII, 965–86.
EURYDICE (2005) Citizenship Education at School in Europe. Brussels: Eurydice.
Galston, W. (2001) ‘Political Knowledge, Political Engagement and Civic Educa-
tion’ Annual Review of Political Science, 4, 217–34.
Goossens, L. (2001) ‘Global Versus Domain Specific Statuses in Identity Research:
A Comparison of Two Self-Report Measures’ Journal of Adolescence, XXIV,
681–99.
Green, A., J. Preston and J. G. Janmaat (2006) Education, Equality and Social
Cohesion: A Comparative Analysis. Basingstoke: Palgrave Macmillan.
Hallinan, M. T. (1994) ‘Tracking: From Theory to Practice’ Sociology of Education,
LVII, 79–91.
Hoskins, B., E. Villalba, D. van Nijlen and C. Barber (2008) Measuring

Civic Competence in Europe: A Composite Indicator Based on the IEA Civic
Education Study 1999 for 14 years old in School. Luxembourg: European
Communities.
Hoskins, B., J.G. Janmaat and E. Villalba (2012) ‘Learning Citizenship through
Social Participation outside and inside School: an International, Multilevel
Study of Young People’s Learning of Citizenship,’ British Educational Research
Journal, 38, 3, 419–46.
Hyland, N.E. (2006) ‘Detracking in the Social Studies: A Path to a More Demo-
cratic Education?’ Theory into Practice, XLV, 64–71.
Inglehart, R. (1990) Culture Shift in Advanced Industrial Societies. Princeton, NJ:
Princeton UP.
Janmaat, J.G. and N. Mons (2011) ‘Promoting Ethnic Tolerance and Patriotism:
The Role of Education System Characteristics’ Comparative Education Review, LV,
56–81.
Janmaat, J.G. (2012) ‘Civic Competences: Some Critical Reflections’ in M. Print
and D. Lange (eds.) Civic Education and Competences for Engaging Citizens in
Democracies. Rotterdam, The Netherlands: Sense Publishers, 51–64.
Langton, K. P. and M. K. Jennings (1968) ‘Political Socialisation in the High
School Civic Curriculum in the United States’ American Political Science Review,
LVII, 852–67.
Levinson, M. (2010) ‘The Civic Empowerment Gap: Defining the Problem and
Locating Solutions’ in R. Sherrod, J. Torney-Purta and C.A. Flanagan (eds.)
Handbook of Research on Civic Engagement in Youth. Hoboken, N.J: John Wiley &
Sons, 331–62.
Lichterman, P. (1996) The Search for Political Community: American Activists
Reinventing Commitment. Cambridge: Cambridge UP.
Loveless, T. (1999) ‘Will Tracking Reform Promote Social Equity?’ Educational
Leadership, LVI, 28–32.
Mons, N. (2007) Les Nouvelles Politiques Educatives. Paris: Puf.
OECD (2010) PISA 2009 Results: What Makes a School Successful? Resources, Policies
and Practices (Volume IV), http://www.oecd-ilibrary.org/education/pisa-2009-
results-what-makes-a-school-successful_9789264091559-en, date accessed
2 February 2012.
Netjes, J.E., H. van de Werfhorst, A.B. Dijkstra and E.A.M. Geboers (2011)
‘Eenheid of Verdeeldheid: Burgerschap in een Gedifferentieerd Voortgezet
Onderwijssysteem’ Mens en Maatschappij, LXXXVI, 1, 34–65.
Putnam, R. (2000) Bowling Alone: The Collapse and Revival of American Community.
New York: Simon and Schuster.
Rosanvallon, P. (1995) La Nouvelle Question Sociale: Repenser L’Etat Providence. Paris:
Seuil.
Rubin, B.C. (2003) ‘Unpacking Detracking: When Progressive Pedagogy Meets
Students’ Social World’ American Educational Research Journal, XV, 539–73.
Rubin, B.C. (2007) ‘ “Laboratories of Democracy”: A Situated Perspective on
Learning Social Studies in Detracked Classrooms’ Theory and Research in Social
Education, XXXV, 62–95.
Schulz, W., J. Ainly, J. Fraillon, D. Kerr and B. Losito (2010) ICCS 2009 International
Report: The Civic Knowledge, Attitudes and Engagement Among Lower Secondary
School Students in 38 Countries. Amsterdam: IEA.
Schulz, W., J. Ainly and J. Fraillon (2011) ICCS 2009 Technical Report. Amsterdam:
International Association for the Evaluation of Educational Achievement.
Stevens, F. (2002) ‘Growing up in a Symbolic Society: The Difference
Between “Spirou” and “Tintin” (Gender and Educational Tracking in the
Social and Cultural Practices of Young People in Flanders)’. Conference
Paper, http://www.vub.ac.be/TOR/main-/publicaties/downloads/t2002_38.doc,
date accessed 1 February 2012.
Ten Dam, G.T.M. and M. Volman (2003) ‘Life Jacket and the Art of Living: Social
Competence and the Reproduction of Inequality in Education’ Curriculum
Inquiry, XXXIII, 117–37.
Torney-Purta, J., J.R. Lehmann, H. Oswald and W. Schulz (2001) Citizenship Edu-
cation in Twenty-eight Countries: Civic Knowledge and Engagement at Age Fourteen.
Amsterdam: The International Association for the Evaluation of Educational
Achievement.
Van de Werfhorst, H. (2007) ‘Vocational Education and Active Citizenship
Behavior in Cross-National Perspective’ AIAS Working Paper Number 2007/62,
http://www.uva-aias.net/uploaded_files/publications/WP62.pdf, date accessed
1 February 2012.
Van de Werfhorst, H. (2009) ‘Education, Inequality and Active Citizenship:
Tensions in a Differentiated Schooling System’ AIAS Working Paper 09–73,
1 February 2012.
Van de Werfhorst, H. and J.B. Mijs (2010) ‘Achievement Inequality and the Insti-
tutional Structure of Educational Systems: A Comparative Perspective’ Annual
Review of Sociology, XXXVI, 407–28.
Waterman, A. (1982) ‘Identity Development from Adolescence to Adulthood:
An Extension of Theory and a Review of Research’ Developmental Psychologist,
XVIII, 341–58.
Willis, P. (1997) Learning to Labour: How Working Class Kids Get Working Class Jobs.
Farnham: Ashgate.
Wolf, P. J. and S. Macedo (2004) ‘Introduction: School Choice, Civic Values, and
Problems of Policy Comparison’ in P. J. Wolf, S. Macedo, D. J. Ferrero and
C. Venegoni (eds.) Educating Citizens: International Perspectives on Civic Values
and School Choice. Washington, DC: Brookings Institution Press, 1–27.
10
Lifelong Learning and Social
Cohesion
Audrey Dumas, Philippe Méhaut, Noémie Olympio
and Isabelle Dimeglio
An extensive literature has evolved, particularly in the United States,

around the notion of social cohesion. This literature marks a shift away
from sociological ideas of the past. It reflects the profound economic
and social changes that societies have undergone in the last 30 years
under the impact of successive crises, the development of market rela-
tions in most spheres of social life and the effects of globalisation. As at
other times in the past, these changes call into question what it means
‘to live in society’. The notion of social cohesion has been the subject of
considerable debate (for example Chan et al., 2006; Green et al., 2006;
Dubet et al., 2010). The arguments have been all the more intense since
various countries have put the spotlight on social cohesion as the foun-
dation and/or goal of public policies. Thus the European Union (EU),
for example, has offered member states a set of targets and indicators
linked to social cohesion as part of the ‘European social model’. Simi-
larly, in Canada (Patrimoine Canadien, 2004; Jeannotte, 2000) several
government reports have examined this issue.
The research literature on these questions can be divided into sev-
eral different groups. Some studies focus on the effects of cohesion or
of a loss of cohesion, whether these effects be directly economic (on
growth, on the functioning of markets, and so on in particular Algan
and Cahuc, 2007) or of a different kind (criminality, tensions between
social groups and so on, cf Baudelot and Leclercq, 2005; Sabates, 2010).
Others are more directly conceptual in nature, whether their objective
is to debate the notion and/or test hypotheses concerning the universal-
ism (or otherwise) of a particular model of social cohesion (Green et al.,
2011; Janmaat, 2011). Yet others focus on earlier stages in the process
of establishing social cohesion in order to examine the economic, social
223
and institutional conditions that are assumed to play a role in the con-
struction and consolidation of ‘life in society’. Thus Green et al. (2006),
in their seminal book on education, equality and social cohesion, not
only discuss the concept extensively but also use various indicators and
groups of countries to test the links between education (level, educa-
tional inequalities, and so on) and indicators of cohesion. More recently,
Dubet et al. (2010) have also tested the same relationship in a sample of
OECD countries.
This chapter adopts a similar perspective. It examines the links
between the structural characteristics of education systems in EU mem-
ber states and the main indicators of social cohesion. Do systems with
strong apprenticeship systems, which are often characterised by early
selection, perform less well than more ‘comprehensive’ systems that
are regarded as more egalitarian? Has the advent of mass higher edu-
cation had a visible effect on social cohesion? Is the existence of a
strong continuing training system, which is supposed to compensate
for inequalities in initial education, a factor that encourages social
cohesion? The key question is whether or not educational models and
societal forms of social cohesion can be compared. In the first section,
we briefly review the notions of social cohesion and the hypotheses
of a link with education. In the second part, the methodology used
is outlined. In the third section, the relationships between various
characteristics of European education and training systems and various
indicators of social cohesion are tested.
Social cohesion and education/training
Social cohesion: A hazy concept

A number of attempts have been made to conceptualise the notion of
social cohesion (Jenson, 1998; Berger-Schmitt, 2000; Jeannotte, 2000;
Beauvais and Jenson, 2002; Conseil de L’Europe, 2005; Green et al.,
2006; Green et al., 2011; Dimeglio et al., 2012). Jenson (1998) investi-
gates how social cohesion is perceived in four national and international
organisations, namely the Government of Canada’s Policy Research
Sub-Committee on Social Cohesion, the Commissariat Général au Plan
(State Planning Commission) in France, the Club of Rome and, finally,
the OECD. She notes that only the first two bodies offer an explicit
definition of social cohesion, although they all stress the importance
of the notion. For the Government of Canada’s Policy Research Sub-
Committee, social cohesion is ‘the on-going process of developing a
community of shared values, shared challenges and equal opportunity
Philippe Méhaut, Noémie Olympio and Isabelle Dimeglio 225
within Canada, based on a sense of trust, hope and reciprocity among

all Canadians’ (quoted by Jenson, 1998). For the Commissariat Général
au Plan, it is a set of social processes that help instil in individuals the
sense of belonging to the same community and the feeling that they are
recognised as members of that community. Drawing on these four doc-
uments, Jenson identifies five separate dimensions of social cohesion,
namely: belonging versus isolation (shared values, feeling of identity);
inclusion versus exclusion (equality of opportunity among citizens in
the economic sphere); participation versus non-participation (politi-
cal participation); recognition versus rejection (respect for differences,
tolerance of diversity); legitimacy versus illegitimacy (legitimacy of insti-
tutions as mediators of particular interests). For its part, The Council of
Europe defines social cohesion as a society’s capacity to ensure the well
being of all its members, to minimise disparities and to avoid polarisa-
tion. Despite this infatuation, researchers and politicians, in the absence
of any agreement on a common definition, are at least agreed that no
such definition exists. As Alaluf puts it, ‘However, the notion of social cohe-
sion seems rarely to be defined and this lack of clarity frequently substitutes
for logical reasoning in debates’ (Alaluf, 1999).
Social cohesion regimes?

The distinction between social capital and social cohesion has been cited
several times in order to highlight a common difficulty, that of the
level of analysis. For McCracken (1998), individual social capital con-
stitutes the basic foundation for social cohesion. For Bouchard et al.
(2002), social capital is a sub-set of social cohesion. Janmaat (2011),
Green et al. (2011) and Chan et al. (2006) emphasise that social cohe-
sion should be seen as a macro-level concept. After all, it is not difficult
to imagine a very divided society in which various communities culti-
vate strong intra-community ties but do not maintain links with other
groups. Thus a high level of social capital will not necessarily be reflected
in a high level of social cohesion. This does not mean that social capital
and social cohesion are mutually contradictory notions but rather that
social capital is not in itself sufficient to capture social cohesion in its
entirety.
Social cohesion is a multidimensional concept. Everyone seems to
agree on this point but not on how many dimensions should be
adopted, which dimensions they should be or the linkages between
them. Thus Chan et al. (2006) are opposed to definitions such as that
adopted by Jenson (1998), which they describe as pluralistic in the sense
that it encompasses several dimensions and the question of which one
to focus on will depend on the problem being tackled. For their part,
Green et al. (2011) and Janmaat (2011) note that different ‘regimes’ of
social cohesion exist. Green et al. (2011) identify four different regimes
in OECD countries: a ‘liberal’ regime characterised by a high level of
civic participation, tolerance and cultural pluralism, a ‘social demo-
cratic’ regime that is egalitarian, trusting and pluralistic, a ‘conservative’
regime in which social order and hierarchy are dominant and, finally,
an ‘East Asian’ regime characterised by equality, social order and social
hierarchy. Dimeglio, Janmaat and Méhaut (2012) go down this route by
comparing the expression of certain values at the macro-social level and
then at the level of the labour market (for example, does a high toler-
ance of immigrants go hand in hand with rejection of the notion that
nationals should be given priority when it comes to accessing jobs?).
Thus they show that, in a sample of EU member states, a number of
different profiles exist, which are sometimes unstable depending on the
level of analysis (society as a whole or labour market). In the remain-
der of this section, we examine the hypothesis that ‘societal’ forms’ of
social cohesion exist, each of which can be linked to one of a number of
different educational models.
What are the links between education, training

and social cohesion?
At the end of their book on the links between education, equality and
social cohesion, Green et al. (2006) re-examine their initial model and
schematise these links as follows (Figure 10.1):
Labour market regime
Welfare regime
Skills distribution Income distribution

Status distribution
LLL system Social cohesion regime

Employment rate
Socialization
Figure 10.1 The link between education and social cohesion

Source: Green et al. (2006, p 167).
They formulate and develop the hypothesis that there are different
models of knowledge in the knowledge society. In this respect, they
concur with the varieties of capitalism (Estevez-Abe et al., 2001) and the
various institutionalist schools. They emphasise the dual effect of these
models, which is mediated, on the one hand, through the socialisation
of pupils and students (transmission of values by schools, more or less
collective and cooperative modes of organising education and training,
construction of identities, and so on) and, on the other, through the
impact of education on the distribution of knowledge (between social
groups) and on labour market positions.
In this book, Green et al. (2006, p. 42) stress the difficulties involved
in establishing a firm link between education and cohesion. Drawing
on data from the IALS survey, they do not find any relationship between
the average level of literacy and the indicators of social cohesion. On the
other hand, an indicator of inequalities in literacy levels between social
groups is correlated with inter-individual trust. Similarly, having exam-
ined the links with tolerance of foreigners/immigrants, which some
analysts regard as a dimension of social cohesion, they first conclude
that it is difficult to interpret questions of tolerance in conjunction
with the other dimensions of cohesion. They also discuss the various
possible relationships between education and tolerance, which may be
mediated through the part that education systems play in socialisation,
the forms of cooperation between pupils that they or may not foster and
the impact of the knowledge acquired at the various levels of education.
However, they remain very doubtful about the possibility of establishing
a direct link between education and tolerance.
When it comes to the question of educational inequalities, and in
particular the role of comprehensive schools (see Chapter 5 in Green
et al., 2006), they discuss the various models of education in terms of
whether or not they involve early selection. Generally speaking, edu-
cational inequalities (often measured at the end of compulsory school-
ing) are greater in countries with early selection. However, this broad
statement merits debate and further qualification. Although formally
non-selective, schools in the ‘Anglo-Saxon’ (that is English-speaking)
countries are in fact differentiated because of the existence of a quasi-
market in education. Conversely, the very high levels of inequality
observed at the end of the three streams of compulsory secondary
education in Germany seem to be rectified if the measurement takes
place after apprenticeships have been completed. We share this concern
to go beyond compulsory schooling in any investigation of the links
between educational inequalities and forms of social cohesion and to
take account of the fact that the majority of a given age group will today
have access to various forms of post-compulsory education and even to

higher education.
Dubet et al. (2010) take up this question again with reference to a
group of OECD countries. In sharp contrast to our study, their analysis,
which is based on data from the PISA surveys, is concerned essentially
with compulsory schooling, even though they depart from this some-
what when they examine the question of the ‘ascendancy of qualifica-
tions’. Furthermore, their analysis is more ambitious than ours, since it
seeks to link models of society with models of education. However, their
conclusions concur in part with those reached by Green et al. There is no
strong correspondence between education systems and the characteris-
tics of social models, taking into account questions of social cohesion.
Thus they point to the paradox of the German model (which we extend
to other systems in which apprenticeships play an important part), in
which a firmly entrenched process of early selection does not seem to
give rise to inequalities and tensions in terms of cohesion. They also
point to the limitations and exhaustion of the meritocratic system. Only
the Nordic societies (and to a lesser extent those of ‘small countries’)
seem to exist in symbiosis with their education systems. Finally, and
this is their main contribution, they show, by taking account of progress
through the labour market, how the ‘ascendancy’ of qualifications in the
labour market makes it possible to understand relationships that appear
to be strongly influenced by the functioning of labour markets.
In this chapter, we examine a similar question on the relationships
between education and social cohesion, focusing more particularly on
upper secondary education, on the forms of linkage between vocational
and general academic education and on access to higher education and
continuing training. This decision was taken for several reasons, even
though, as we shall see, the quality of the data on these educational
levels is not as good as the data on compulsory education (Field et al.,
2009).
The first reason is that in most European countries education at this
level has expanded considerably in recent decades, to the point where it
has tended to absorb virtually the whole of any given age cohort. Conse-
quently, even though the initial socialisation in compulsory schooling
is most certainly fundamental, what happens at upper secondary level
cannot be ignored, particularly since the various vocational education
systems place varying degrees of emphasis on the various aspects of
citizenship or group work (cf. in particular Brockmann et al., 2010).
The second reason is that this expansion has taken different forms,
particularly with regard to the link between general and vocational
education. This may affect both inequalities and the formation of iden-
tities. It is known, for example, that occupational identities (the notion
of Beruf in Germany) exert a considerable structuring influence on
apprenticeship systems in German-speaking countries. In other systems,
vocational training is a ‘dominated’ pathway that acts principally as a
destination for people who have not done well in general education.
Furthermore, it is known that the availability of ‘second chance’ oppor-
tunities through continuing education or training varies considerably
across Europe.
In Chapter 2 of this book, we identified a number of different config-
urations for education/training systems, as they existed in 26 European
countries at the beginning of the twenty-first century. The main config-
urations are summarised here and some hypotheses are advanced as to
the possible link with social cohesion.
Three models can be identified in countries with strong vocational
training systems. The first is based on comprehensive schools. Voca-
tional training is integrated into the system, with low failure rates.
Schools in such systems might be described as ‘integrative compre-
hensive schools’, which transmit the same values and have to forge
common identities regardless of the pathway selected. Such an ‘egali-
tarian’ system goes hand in hand with the ‘social democratic’ model of
cohesion identified by Green and Janmaat, 2011). It is also similar to
Verdier’s ‘universalist’ regime, the aim of which is to foster education
for all in accordance with the equal opportunities principle.
The second is characterised by separation but also by integration into
occupations. The countries in question here are those with highly devel-
oped apprenticeship systems, namely Austria, Czech Republic, Slovakia,
Switzerland, and, to a lesser extent, Denmark and Germany. Vocational
training is separated from general academic education, on the basis of
‘early tracking’. There is little prospect of access to higher education.
However, because of the homogeneity and strength of the apprentice-
ship system, this separation does not exclude the construction of strong
occupational identities involving large numbers in each generation.
In this sense, this group can be compared with Verdier’s ‘occupational
regime’, since this education and training system facilitates ‘access to
an occupational community’. Consequently, the foundations of social
cohesion would probably be different from those in the preceding
group.
The third model is also characterised by separation, but is less integra-
tive. Vocational training (which tends to be school-based) is strong but
the pathways are often short. Dropout rates are average or high, while
the opportunities for progressing to higher education tend to be limited.

Such systems involve ‘tracking’ at upper secondary level, which may or
may not follow on from a comprehensive lower secondary system. This
group seems to be a sort of ‘half-way house’ between the ‘separation’
model and Mons’ ‘uniform integration’ model (Mons, 2007), in which
the integrative power of the single type of school (with its common cur-
riculum until the end of the lower secondary stage) is more formal than
real. This should have ‘negative’ consequences for certain dimensions of
social cohesion, such as trust.
In countries with weak vocational training systems, there is greater
heterogeneity. In most cases, a ‘meritocratic’ principle seems to be at
work throughout the system, which makes it akin to Verdier’s ‘academic’
regime (Verdier, 2009). Such systems are characterised by the impor-
tance attached to qualification and the academic meritocracy. However,
these mechanisms may be at work in the compulsory stages of edu-
cation or come into play only after the end of ‘comprehensive’ lower
secondary education. In such systems, general education is separated
from vocational education/training, which is often the destination of
those excluded from general academic education. However, this does
not totally close off the possibility of higher education, particularly
in countries where short pathways exist in higher education (two-year
foundation degrees and so on). It might be assumed that in these
countries interpersonal trust is more difficult to establish because of
the competition in the education system. Systemic trust would depend
heavily on ‘academic success’.
Indicators and methodology
Capturing social cohesion: Source, indicators and method

Following the example of Chan et al. (2006) as closely as possible, we
propose to use a limited number of synthetic indicators that describe
various dimensions of social cohesion. They are all constructed on the
basis of data drawn from the World Value Survey (WVS),1 either from the
2005–2008 wave or, failing that, from the preceding wave (1999–2004),
in order to be as close as possible to the data on education/training
systems. For the main items selected by Chan et al., we use a ques-
tion or set of questions from the WVS in order to construct synthetic
indicators. It is fairly standard practice to use these indicators in stud-
ies of social cohesion. They are, for example, fairly similar to those
used by Dubet et al. (2010). However, they are calculated solely for the
15–30-year-old age group who have left the education system, that is the
group most directly affected by the changes in education and training

systems.
Thus social cohesion is captured through three dimensions, namely
(interpersonal and institutional) trust, civic and political participation
and tolerance of and respect for diversity.
• Participation
In the literature on cohesion, civic participation is typically captured in

terms of the strength of social ties as measured by individuals’ involve-
ment in cultural, sporting, professional and political organisations. Thus
we construct a vertical synthetic indicator of participation (part_prof) on
the basis of declared membership of a political party (part_pol), trade
union (part_syn) or professional association (part_assop). Similarly, hor-
izontal participation (part_civ) is measured on the basis of declared
membership of educational, artistic or musical associations (part_art),
sporting associations (part_sport), religious organisations (part_relg) or
environmental organisations (part_animal).
• Trust
Trust is a key element in the social bond. Consequently, and despite the
difficulty of defining social cohesion, ‘trust’ is a dimension that always
occupies a dominant place in our understanding of this notion; unlike
other dimensions, such as the more economic ones or the sharing of
values, trust is never called into question as an indicator of social cohe-
sion. Horizontal trust is measured by the indicator of interpersonal trust
that is normally used, namely the share of individuals who state that
they generally trust most people.2 Vertical trust is measured by insti-
tutional trust (conf_inst). The indicator is constructed on the basis of
individual declarations of trust in institutions; the strength of that trust
is measured on a scale from 1 to 4 and the institutions in question
are the armed forces (conf_army), parliament (conf_parle), the police
(conf_police) and the justice system (conf_judi).
• Tolerance
Tolerance of and respect for diversity are essential aspects of social

cohesion in societies characterised by increasing diversity (Dimeglio,
Janmaat and Méhaut, 2012). We construct a synthetic indicator of tol-
erance (tol_gen) based on six variables.3 The first four are the share of
individuals who state they are not averse to having a neighbour from
a different race (tol_race) or suffering from AIDS (tol_aides) or who is
homosexual (tol_homo) or an immigrant worker (tol_immig). To these
are added the share of individuals who state that tolerance and respect
for diversity are important qualities for their children (tol_qual) and
those not giving priority to ‘nationals’ in the labour market when jobs
become scarcer (pior_natv) (Dimeglio, Janmaat and Méhaut, 2012).
Methodology
In order to establish the link between the characteristics of the educa-
tion and training system and the various dimensions of social cohe-
sion, we make use once again of the qualitative comparative analysis
(QCA) method and more particularly the second stage of this method
(cf. Chapter 2 of this book for an outline of the first stage of this method
and its use in the construction of education and training profiles). In the
first stage, groups of countries were constituted on the basis of the char-
acteristics of their education systems. In the second stage, efforts will
be made to establish the link between these typologies and an outcome
variable, such as one of the indicators of social cohesion.
In order to establish this link, the QCA method (Ragin, 2008) is based
on Boolean logic. Its aim, therefore, is to determine whether a combi-
nation of indicators is a sufficient condition for obtaining an outcome.
More specifically, if the membership score for a combination of indica-
tors, for example a typology, is a sub-set of the membership score for an
outcome variable, that is a dimension of social cohesion, then the com-
bination of education and training system characteristics is a sufficient
condition for obtaining a particular level (HIGH or LOW) for one of the
indicators of social cohesion. Otherwise, if it is not a sub-set, then the
characteristics emphasised in the typology are not a sufficient condition
for predicting that level.
Thus a degree of consistency is calculated between a typology and
an outcome variable. The measure of consistency is as follows, where
T is each country’s degree of membership in the typology i and Y is
each country’s degree of membership in the variable of social cohesion
within that same typology i.

(Ti ,Yi )= Min(Ti ,Yi )/ Ti
Consistency i i
If Ti ≤ Yi , consistency is equal to 1.
If Ti > Yi but Ti is close to Yi consistency is lower than 1 but close to 1.
If Ti > Yi and Ti is far from Yi then consistency is less than 0.5.
Let us take an example. Let us assume that four typologies of edu-

cation system characteristics have been obtained, each containing two
countries whose membership scores are obtained as set out in Chapter 2.
Let us also assume that our ‘calibrated’ outcome variable Y, that is with a
value between 0 and 1 (see Chapter 2), is an indicator of social cohesion.
We can then see which typologies may be linked with a HIGH level of
social cohesion (Y) or a LOW level of social cohesion (1–Y). The formula
applied to obtain the relationship’s consistency score is based on the
idea of penalising, that is moving away from the value 1, firstly when
there is inconsistency, that is when the degree of membership in the
combination is greater than the degree of membership in the outcome
variable (Typology 1 versus Typology 2 for the variable Y) and, secondly,
penalising all the more the greater the extent of the inconsistency is
(Typology 3 versus Typology 2 for variable Y) (Table 10.1).
The relationship between the typologies and the variable of social
cohesion is considered to exist if the degree of consistency is sufficiently
high. As a general rule, for QCA we have to consider a consistency
threshold of 0.8. For our example, therefore, Typology 1 and Typology
2 may be a condition for strong social cohesion, while Typology 4 is a
condition for a low level of social cohesion.
The tables in the following section give the degree of consistency
between the characteristics of the education and training systems and
the various indicators of social cohesion, when the indicator is consid-
ered for both a HIGH level and a LOW level (the score is underlined in
these cases). For the first typology, a consistency threshold of 0.8 is con-
sidered. For the following typologies, a threshold of 0.9 will be consid-
ered, since a number of links with cohesion are revealed. These are ‘truth
tables’ (rather than correlation tables), as is appropriate in Boolean logic.
Table 10.1 Typologies and scores of social cohesion according to QCA.
Typology Country Membership Y Social (1–Y) Social Consistency Consistency

score in the cohesion cohesion score for Y score for
typology HIGH LOW (1–Y)
Type 1 S 0.6 0.8 0.2 1 0.23

T 0.7 0.9 0.1
Type 2 U 0.6 0.8 0.2 0.92 0.46
V 0.7 0.6 0.4
Type 3 W 0.6 0.8 0.2 0.73 0.47
X 0.9 0.5 0.5
Type 4 Y 0.6 0.3 0.7 0.54 0.92
Z 0.7 0.4 0.6
The link between education systems and social cohesion:

An indeterminate relationship?
In this section, the links between various configurations of access to

initial and continuing training and the main indicators of cohesion will
be tested in succession.
The degree and mode of ‘vocationalisation’

of secondary education
In the first analysis, the relationship between the degree of ‘vocation-
alisation’, the mode of vocationalisation (strong apprenticeship system
or otherwise) and our indicators of cohesion is tested. In the literature,
selective education systems that channel pupils towards the vocational
streams (and particularly towards apprenticeships) at an early stage are
often regarded as inegalitarian systems that might be expected to have
negative effects on the indicators of cohesion. However, we do not
accept this hypothesis. After all, it might be thought, conversely, that
a large-scale apprenticeship system, which both has a strong socialising
impact and contributes to the construction of strong occupational iden-
tities, would in fact have a positive effect that could compensate for the
early selection.
In fact, the ‘high degree of vocationalisation’ condition, whether
it is associated with extensive use of apprenticeships or a school-
based pathway, produces nothing. The same is true of the ‘low degree
of vocationalisation’ condition, whether it is associated with a weak
apprenticeship system or ‘general’ education systems, most of which
could be described as meritocratic. No relationship can be verified at
the consistency threshold of 0.9. However, if the threshold is lowered to
0.80, there is an association at a low level of interpersonal trust for those
countries whose education systems are not highly vocational in nature.
Mass access to (upper) secondary education

Table 10.2 shows the same analyses but adds two variables reflecting
access to and success in upper secondary education, namely the rate of
access (ESPSS) and the difference in success between vocational and gen-
eral pathways (DTSS). Certain relationships become apparent, although
a secondary configuration of this kind cannot be clearly linked to any
particular indicator of cohesion.
Thus social trust is not linked to the various configurations of mass
access to upper secondary education. Only in the case of countries
that have a low level of vocationalisation (normal font in the table),
Table 10.2 Access to upper secondary education and indicators of social

cohesion.
ESPSS DTSS Country Trust Tolerance Participation

social institutional political civic
High High IRL 0.94 0.95 0.90 0.90

SI 0.93 0.93
FIN
CZ
AT
CH
SK
DE
High Low LT 0.91 0.91
PL
LV
EL
EE
SE 0.93 0.96 0.98 0.95
NOR
Low High IT 0.90 0.94 0.94
LUX
NL 0.94 0.94
Low Low ES 0.96 0.91 0.96 0.95
FR
HU
PT
RO 0.90 0.91 0.94 0.92
DK 0.99 0.97
Note: Interpreting the table:

In chapter 2, we obtained three groups of countries: those characterised by a low share of
vocational pathways (font: normal), those with a high share of vocational pathways but not
in the form of apprenticeships (font: italics) and, lastly, those with a high share of vocational
pathways in the form of apprenticeships (font: bold).
The consistency scores are calculated for different typologies (combining these three initial
groups and two other characteristics of education systems: ESPSS and DTSS) with a HIGH
level for our five indicators of social cohesion (score in bold) and a LOW level for our five
indicators of social cohesion (score underlined).
relatively low access (that is lower than average) to upper secondary

education and no difference in success between the various path-
ways can a relationship be discerned with a low level of interpersonal
trust. This tends to concur with the results obtained by Duru-Bellat
and co-authors (Chapter 11 of this book), who note that the most
meritocratic countries are not the best performers.
Trust in institutions does not seem to be unambiguously explained by

education systems either. It may be linked equally well to mass access to
upper secondary education (Ireland, Sweden) and to low – or differing –
access (Italy, Luxembourg, Holland and, particularly, Denmark). On the
other hand, with the exception of Ireland, those countries in which
vocational education is relatively underdeveloped seem to be charac-
terised by a link to low levels of trust in institutions, regardless of how
accessible upper secondary education may be.
Tolerance does not have any obvious link with the various upper sec-
ondary education configurations either. Only three countries (Ireland,
Sweden and Denmark) have a consistency score suggesting a positive
relationship. However, they belong to different educational groupings.
As far as participation in civic and political life is concerned, the
results are more consistent with what is known. Mass access to upper
secondary education may lead to a good participation score. However,
once again with the exception of Ireland, this is verified only for coun-
tries with ‘integrated’ vocationalised systems (Slovenia, Finland, Sweden
and Norway). In contrast, low levels of access to upper secondary edu-
cation tend to go hand in hand with scores suggesting a negative
relationship, particularly for the meritocratic countries.
At this stage of the analysis, therefore, it is not possible to establish a
firm link between the structures of upper secondary education systems
and social cohesion, even for countries that have adopted the ‘Nordic’
model, which are well known for their high cohesion ‘scores’ (cf. also
on this point Dimeglio, Janmaat, Méhaut, 2012) but which are far from
being homogeneous in their educational systems (strong apprenticeship
system in Denmark, for example, but weak apprenticeship systems in
Sweden and Finland).
Access to higher education

In this stage, the rates of access to upper secondary education (ESPSS)
and the rates of access to higher education (ESPSUP) will be combined.
Here too, the hypothesis is that the systems that are more open at all
levels will probably be more egalitarian. The characterisation by degree
and type of vocationalisation is retained (Table 10.3).
Here too, the results are fragmented and there is no general trend.
As before, social trust is not explained by the access characteristics,
apart from a weak and negative link in some of the more merito-
cratic countries. Stronger positive relationships are evident in the case
of institutional trust. Here too, however, that ‘small Nordic country’
effect seems to prevail, regardless of educational consideration. The
Table 10.3 Access to upper secondary and higher education and indicators of
cohesion.
ESPSS ESPSUP Country Trust social Tolerance Participation

institutional political civic
Low Low ES 0.97 0.90

FR
PT
UK
RO 0.90 0.91 0.91 0.94
DK 0.99
High Low IRL
SE 0.96 0.93 0.98 0.93
AT
CH
SK
DE
Low High HU 0.95 0.95 0.95 0.99
IT 0.97 0.93
NL 0.93 0.90 0.93 0.90
High High LT 0.91 0.90
PL
LV
EL
EE
SI 0.94 0.93
FIN
NOR
CZ

In chapter 2, we obtained three groups of countries: those characterised by a low share of
The consistency scores are calculated for different typologies (combining these three initial
groups and two other characteristics of education systems: ESPSS and DTSS) with a HIGH
same applies to tolerance (for example, there is no convincing result:

whereas one might have been expected for countries that combine very
high levels of access to upper secondary education with high levels of
access to higher education). The same observation can be made in the
case of associative or political participation: there is a degree of con-
sistency in the positive or negative relationship in some cases, which
is overdetermined by a country effect. The ‘country’ effect seems to

prevail particularly in the case of small countries which, to quote Dubet
et al. (2010), seem ‘to live in symbiosis with their education and training
systems’, regardless of their configuration.
Continuing training
Using the same approach, we turn now to the indicators of access to con-
tinuing training. While it is not possible to identify a firm link between
the characteristics of access to upper secondary and higher education,
on the one hand, and the indicators of social cohesion, on the other,
what can we say about access to continuing training, a sort of ‘second
chance’, access to which varies from country to country? (Table 10.4).
In order to examine this question of continuing training in greater
detail, an indicator of access to continuing training (CT) and an indica-
tor of access to continuing training leading to a qualification (CTQ) are
combined with the various types of vocationalisation used above.
This time, the results are more significant and more coherent. A high
level of access to continuing training (whether or not it leads to
Table 10.4 Continuing training and social cohesion.
CT CTQ Country Trust social Tolerance Participation

institutional political civic
Low Low PL, PT, 0.92 0.90 0.90

IRL,ES,FR
IT LU 0.92
CZ, SK, DE 0.94 0.92
Low High HU,EL 0.91 0.90 0.94 0.90 0.96
BEL 0.95 0.93 0.92
High Low CH, NL, AT 0.94 0.98 0.90 0.99
High High UK 0.99 0.99 0.98 0.97
FIN, SE, NOR 0.93 0.98 0.98 0.99 0.99
DK 0.93 0.97 0.99 0.99

In chapter 2 we obtained three groups of countries: those characterised by a low share of
The consistency scores are calculated for different typologies (combining these three ini-
tial groups and two other characteristics of education systems: CT and CTQ) with a HIGH
qualifications) is a predictor of high scores for most of our indicators

of cohesion, even with a fairly high consistency threshold. On the other
hand, a low level of access to continuing training seems to predict low
scores for cohesion, or relatively few links. These results are in part
consistent with other studies that emphasise the importance of con-
tinuing training in various areas ‘outside the labour market’ (Preston
and Feinstein, 2004). Furthermore, they cast considerable doubt on the
strategies adopted in certain countries that rely almost exclusively on
initial training and provide support for the notion of lifelong learning
advocated by the EU.
If the countries are integrated into the framework on the basis of their
form of vocationalisation, then it is clear that, with the exception of
the United Kingdom, those countries with a ‘general’ education system
(in normal font in the table) are also those that have least access to
continuing training (and/or, as is the case in France, have very unequal
access to continuing training, in accordance with the ‘Matthew effect’,
whereby further training is available to those who already have train-
ing) and low scores for our indicators of cohesion. The countries with
strong apprenticeship systems (in bold) stand apart from the others.
Switzerland, the Netherlands, Austria and Denmark are characterised
by high levels of access to continuing training (not necessarily lead-
ing to qualifications) and a positive relationship with the indicators of
cohesion.
Conclusions
These results tend to confirm the weakness or complexity of the direct

links between education and social cohesion. Overall, few configura-
tions can be clearly and decisively associated with high or low scores
for our various indicators. If we focus, for example, on the Scandinavian
countries, Denmark, Sweden and Finland have very different education
systems but these differences do not preclude good performances on our
indicators of cohesion.
However, our results do reveal several major phenomena. In accor-
dance with our hypotheses, there is nothing in the countries with
early selection but strong apprenticeships systems to indicate that early
tracking has a negative effect. There is absolutely no doubt that the
socialising power of a system that tightly structures occupational iden-
tities counterbalances the early selection, which does not happen in
systems in which vocational training, undertaken by a minority only,
remains socially undervalued. This can probably be linked, on the one
hand, to the somewhat fluid conditions under which young people

access employment (Shavit et Müller, 2000) and, on the other, to the
fact that the possession of elitist qualifications opens up fewer possi-
bilities for upward mobility (cf. also Di Paola, Moullet, Chapter 12, on
downgrading or downward mobility).
Similarly, some countries in which a strong vocational pathway is
integrated into the secondary school system but that also have signif-
icant levels of access to university and continuing training (Finland, for
example) have positive results on our indicators of cohesion. In this
case, a more detailed analysis would probably reveal that it is the inte-
gration of vocational and general education within a comprehensive
school system in which pupils remain until their secondary educa-
tion is completed, together with low failure rates, that are the decisive
factors.
This finding seems to reflect those VOC studies that indicate that a
certain level of initial vocational training is likely to offer individuals
more opportunities, in particular by motivating academically weaker
pupils and encouraging them not to lose heart (and leave the education
system without any qualifications, for example) and to improve their
performance, if only to obtain high-quality vocational training (Estevez
et al., 2001). From this point of view, vocational training could be seen
as creating opportunities for pupils with difficulties who might thereby
avoid exclusion.
On the other hand, even though they may have comprehensive
schools up to the end of compulsory education (sometimes genuine
ones, as in Italy, and sometimes more heterogeneous ones, such as the
colleges – lower secondary schools – in France), those countries that have
general secondary education systems and/or an underdeveloped system
of vocational education, which is often socially undervalued, tend to
have low scores on our indicators of cohesion. This is consistent with
the Dubet et al.’s findings (Dubet et al., 2010) on the limitations of
meritocratic systems.
In all these cases, it is essential, and certainly in European societies in
which more than half of each generation has or is going to have access
to higher education, to go beyond analyses that often take account only
of compulsory education and fully incorporate the role and status of
vocational education.
Our results for continuing training provide further support for this
argument. These may seem surprising given the relative importance
of initial education and continuing training. However, they may also
accord fairly well with several hypotheses on cohesion. A fluid society, in

which not everything is determined by social origin, education and/or
initial position in the labour market, is, on the face of it, more redis-
tributive of opportunities. Moreover, a number of studies on continuing
training have shown how, above and beyond questions of the trans-
mission of knowledge, such training is linked to forms of personal
mobilisation, membership of social networks and re-socialisation. Thus
societies in which the notion of ‘second chance’ has a certain reality
(notably the Nordic countries but also the UK and some of the German-
speaking countries) may well have greater fluidity than those in which
access to social positions is relatively predetermined.
It remains for us to test the direct link between the structures of edu-
cation systems and the indicators of cohesion. Despite some results, this
link is sometimes difficult to interpret at this stage.
How are we to explain and analyse this result? Four hypotheses can
be formulated. The first concerns the limitations of our data, particu-
larly those on post-initial training systems. Some of the classifications
used are determined by national choices. To take just the example of
Germany, the school-based vocational pathways (apart from apprentice-
ships) are not recorded as vocational. Moreover, there are still gaps or
inconsistencies for a number of countries (which we have tried to rem-
edy by drawing on national information and data). These limitations
are well known (e.g. Unevoc, 2006). The OECD tended in the past to
neglect vocational training but has recently undertaken some studies
on this topic (Field et al., 2009). Eurostat and CEDEFOP have followed
the same path.
The second is linked to the fact that, while we have been concerned
with structural indicators, they tell us little if anything about what hap-
pens within systems, particularly from the point of view of socialisation
and the transmission of values. Green et al. pay particular attention to
this dimension. Dubet et al. try to capture this aspect with their notion
of ‘educational styles’ (p. 115). Chapters 11 and 12 of this book also shed
some light on this question. These indicators do not tell us anything
about the disparities between social groups that have passed though the
various stages and pathways in the systems. We will probably have to
wait for the future OECD survey on adults’ competences before we will
be able to tackle this question properly.
The third is linked – crucially – to Dubet et al.’s hypothesis (as well
as to that underlying Green et al.’s schema), namely that the effects
of educational systems are largely dependent on their links with the
labour market. Whether it be the return to qualifications or inequalities

in individual positions, the effects of education and training systems
can also be interpreted through their links to the functioning of labour
markets. Di Paola and Moullet (Chapter 12) show how the mismatch
between level of education and job qualifications might work. Dimeglio
et al (2012) have shown how ‘general’ civic values may or not withstand
the test of the labour market.
The fourth, finally, relates to the fact that education systems ulti-
mately have little to do directly with social cohesion. It may be that
other, more powerful determinants are at work. Or it may ultimately
be necessary to reverse the proposition and take the view that ‘what
society does to the education system’ is more powerful than ‘what the
education system does to society’ (Dubet et al., 2010).
Notes
1. The World Value Survey data are drawn from a survey of values and attitudes
conducted first at European level (European Value Survey) and then carried
out and repeated regularly at world level. The areas tackled are the family,
work, social relations, religion, the environment, the economy and politics.
Most of the questions are retained from period to the next. The database is
now made up of six waves (from 1981 to 2012) and is available on the WVS
site at http://www.worldvaluessurvey.org/
2. As opposed to those who state that one can never be too careful when dealing
with others.
3. For the variables tol_race, tol_aides, tol_homo et prior_natv, there are no data
for Hungary.
References
Alaluf, M. (1999) Evolutions démographiques et le rôle de la protection sociale: le
concept de cohésion sociale. Université libre de Bruxelles.
Algan, Y. and Cahuc, P. (2007) La société de défiance, coll. du Cepremap, Éd. de la
rue d’Ulm.
Baudelot, C. and Leclercq, F. (sld) (2005) Les effets de l’éducation, La documenta-
tion française, Paris.
Berger-Schmitt, R. (2000) Social cohesion as an aspect of the quality of societies: con-
cept and measurement, Eureporting Working Paper No. 14, Mannheim : Centre
for Survey Research and Methodology.
Beauvais, C. and J. Jenson (2002) Social cohesion: updating the state of the research,
CPRN Discussion Paper No. F|22, Ottawa :Canadian Policy Research Networks.
Bouchard, L., Ray, J.-F., Lemyre L. and Gilbert, A. (2002) Capital Social, CIRCEM
et Institut de recherche pour la santé des populations.
Brockmann, M., Clarke, L. and Winch, C. (eds) (2010) Knowledge, Skills, Com-
petence in the European Labour Market: What’s in a Qualification? London,
Chan, J., To, H-P. and Chan, E. (2006) ‘Reconsidering Social Cohesion: Devel-
oping a Definition and Analytical Framework for Empirical’ Social Indicators
Research, LXXV, 2.
Conseil de l’Europe (2005), Elaboration concertée des indicateurs de la
cohésion sociale, guide méthodologique, http://www.coe.int/t/dg3/socialpolicies/
socialcohesiondev/source/GUIDE_fr.pdf, 2005
Dimeglio, I., Janmaat, J.G. and Méhaut, P. (2012) ‘Social Cohesion and the Labour
Market: Societal Regimes of Civic Attitudes and Labour Market Regimes’ Social
Indicators Research, 111, 3, pp. 753–773. doi:10.1007/s11205-012-0032-x.
Dubet, F., Duru-Bellat, M. and Vérétout, A. (2010) Les sociétés et leur école, Seuil,
Paris.
Estevez-Abe, M., Iversen, T. and Soskice, D. (2001) ‘Social Protection and the For-
mation of Skills: A Reinterpretation of the Welfare State’ in Hall, P-A., Soskice,
D. (eds) Varieties of Capitalism: The Institutional Foundations of Comparative
Advantages. Oxford: Oxford University Press, pp. 145–158.
Feinstein, L., C. Hammond, L. Woods, J. Preston and J. Bynner (2003) The Contri-
bution of Adult Learning to Health and Social Capital. Wider Benefits of Learning
Research Report No. 8.
Field, S., Hoeckel, K., Kis, V. and A. Kuczera (2009) ‘Learning for Jobs’, Policy
Review of Vocational Education and Training. Initial report, OECD.
Green, Andy and Janmaat, Jan Germen (2011) Regimes of Social Cohesion: Soci-
eties and the Crisis of Globalization. Education, Economy and Society. Basingstoke:
Palgrave Macmillan.
Green, A., Preston, J. and Janmaat, J-G. (2006) Education, Equality and social
Cohesion: A Comparative Analysis. New-York: Palgrave MacMillan.
Janmaat, J.G. (2011). ‘Social Cohesion as a Real-Life Phenomenon: Assessing
the Explanatory Power of the Universalist and Particularist Perspectives’, Social
Indicators Research, 100, 1, pp. 61–83.
Jeannotte, M.S. (2000) Social Cohesion Around the World: An International Compar-
ison of Definitions and Issues. Paper SRA-309 (Strategic Research and Analysis
Directorate) Ottawa: Department of Canadian Heritage.
Jenson, J. (1998) Mapping Social Cohesion: The State of Canadian Research, Paper
SRA-321 (Strategic Research and Analysis Directorate) Ottawa: Department of
Canadian Heritage.
McCracken, M. (1998) Social Cohesion and Macroeconomic Performances, CSLS Con-
ference on the State of Living Standards and the Quality of Life in Canada,
Ottawa.
Mons, N. (2007) Les nouvelles politiques éducatives: la France fait-elle les bons choix?
Paris, PUF, coll. ‘Education et Société’.
Patrimoine Canadien (2004) Identité, culture et valeurs canadiennes: construire une
société cohésive. Ottawa.
Preston, J. and Feinstein, L. (2004) Adult Education and Attitude Change. London:
Center of Research on the Wider Benefits of Learning, Research Report No.11.
Ragin, C. (2008) ‘Qualitative Comparative Analysis Using Fuzzy Sets’ in Rihoux,
B. and Ragin, C. (eds) Configurational Comparative Analysis, Sage Publications,
Chapter 5.
Sabates, R. and Alii, (2010) Social Benefits of Vocational Education and Train-
ing for Individuals: Concepts, Contexts and Empirical Results. Thessaloniki:
Cedefop.
Shavit, Y. and Müller, W. (2000) ‘Vocational Secondary Education. Where

Diversion and Where Safety Net?’ European Societies, 2, 1, pp. 29–50.
Unevoc, UIS (2006) Participation in Formal Technical and Vocational Education and
Training Programmes Worldwide, an Initial Statistical Study. Bonn: Unevoc.
Verdier, E. (2009) L’éducation et la formation tout au long de la vie: une orien-
tation européenne, des régimes d’action publique et des modèles nationaux en
évolution. Sociologie et Sociétés, XL.1.
11
Education and Social Cohesion
in a Comparative Perspective
Marie Duru-Bellat, Antoine Vérétout and Francois Dubet
In current political discourses and policies at a national and European

levels, the development of education is generally assumed to be one of
the key determinants of what is now commonly referred to as social
cohesion: in other words, the longer a society educates its members,
the higher the level of social cohesion and the greater the likelihood
of fostering strong and trusting relationships (see for instance OECD-
CERI, 2006). Very few studies have provided empirical evidence for this
claim, and therefore the issue as to whether education determines social
cohesion to the extent that we assume it does remains largely open.
In order to shed light on this issue, we will begin by defining cohesion
based on a number of key indicators. The indicators used to define social
cohesion will be shown to be correlated with a number of more gen-
eral socioeconomic factors (such as employment rate, social inequalities,
etc.). Secondly, we will explore the links between cohesion and vari-
ous educational variables (in particular the percentage of adults with a
high level of education and educational inequalities) based on a sample
of countries. Figure 11.1 explains the datasets and methods of analy-
sis used for this purpose. Thirdly, we will investigate the underlying
mechanisms and will examine the hypothesis that the mean level and
organisation of education is less important for understanding the level
of cohesion than the way in which education is ‘used’ in society: the
impact of education on individuals’ life will be at the centre of the
analysis, focusing on the benefits brought by educational qualifications
in professional life. We make the hypothesis, contrary to the common
view, that when the impact of education, so defined, is significant, it
may increase educational inequalities and foster the reproduction of
social inequalities between generations: that is because students and
245
The data used are mostly derived from the OECD and international surveys
(World Values Survey and International Social Survey Programme). The data
were often aggregated to produce national indices (e.g. labour market index,
inequality index). Whenever possible, data covering several years were grouped
together. For example, most of the ‘objective’ indices relating to specific societies
cover the period 1995–2005, while the indicators used to measure attitudes and
values are based on the combined responses to 4 waves of surveys: 1981–1984;
1989–1993; 1994–1999; 1999–2004. The study is based on a sample of 27
countries (Australia, Austria, Belgium, Canada, Czech Republic, Denmark,
Finland, France, Germany, Greece, Hungary, Ireland, Italy, Japan, Korea,
Luxemburg, Netherlands, New Zealand, Norway, Poland, Portugal, Slovakia,
Spain, Sweden, Switzerland, United Kingdom and United States). In some cases,
the number of countries was lower because of missing data.
The approach adopted is based on ecological analysis, i.e. various indicators
relating to ‘ecological units’ (here, countries) are assessed and compared. These
indicators are designed to characterize countries and are constructed by
aggregating individual data (e.g. mean attitudes within a given society) or by
reflecting the distribution of individual characteristics in a given country (e.g.
income inequality index). One difficulty of this type of analysis is that there is a
temptation to draw unwarranted conclusions by moving from one level to another:
for instance, because the most highly educated individuals are also, on average,
the healthiest, highly educated countries may be assumed to have a higher
average level of health. The relations between different levels of analysis are
often far more complex than such inferences may suggest. In some cases,
individual effects may combine to produce compositional effects at a national
level that can be easily understood. Yet there may also be strictly contextual
effects, i.e. effects that only emerge at an aggregate level. All across the board
‘ecological paradoxes’ may emerge, i.e. discrepancies between the relations
observed at different levels of analysis. To ignore the possibility of such
discrepancies is to leave oneself open to the ‘ecological fallacy’, defined as
unwarranted inferences from one level of analysis to another.
In short, there is always the risk, in this kind of analysis, of inferring correlations
observed at an aggregate level from relations (causal relations) observed at the
level of individuals. This is particularly true when working on snapshot data since
it is impossible to infer temporal sequences between variables and to determine
causal trends.
Figure 11.1 Datasets and methods used
families have stronger interest in widening educational inequalities in

order to reap the attendant social and economic benefits. In this sense,
a strong impact of education may weaken social cohesion. An analysis
of the perception of social inequalities will be used to support this inter-
pretation. All in all, the assumption is that the various levels of social
cohesion cannot be understood without examining the perceived legiti-
macy of educational inequalities and the perceived justice of the impact
of education in specific countries.
Marie Duru-Bellat, Antoine Vérétout and Francois Dubet 247
Social cohesion and its correlates
The notion of social cohesion implies that societies ‘hold’ are ordered
and dynamic because their members have the qualities, attitudes and
dispositions required to live in harmony and to develop behaviours
that foster solidarity. The assumption is that individuals are actively
involved in networks of relationships and that they are integrated in
social groups that constitute significant resources. Social relationships
serve as the foundation of the density and unity of social life, while
self-confidence and trust in others and in institutions are deemed to be
cardinal virtues. It is not merely that confidence and trust increase rela-
tionships, civic-mindedness and community life, since they also foster
social capital, which is generally defined as one of the key factors of
economic development (see in particular Putnam, 1993).
The notion of social cohesion emerged when the beliefs and values
shared by individuals were deemed to play a major role in economic
development and social dynamism. In recent years, political decision-
makers, sociologists and experts in international organizations such as
the OECD, the IMF and the World Bank have increasingly used the
concept of social cohesion, defined as a set of values and attitudes fos-
tering collaboration, social capital, self-confidence and trust in others
(Coleman, 1990; Putnam, 2007). Cohesion can be defined as the set
of social values and virtues that serve as the foundation of solidarity
in democratic societies and that ensure economic development. Coun-
tries can be characterised according to their ‘regime of social cohesion’,
defined by the level of social cohesion and the nature of the social con-
tract binding a society together, whether it be the market, the state or
the community (Green and Janmaat, 2011).
Following the pioneering studies of Ingelhart (1977, 1993), many
studies have examined the values adhered to by individuals in modern
societies. As a result, a whole range of attitudes towards social relations,
family, inequalities, politics, religion and work (among others) have
been found to shape social cohesion. Since an exhaustive analysis of
the full range of attitudes is beyond the scope of this study, this chapter
will focus on two broad sets of attitudes: (1) Social capital includes the
density of social networks, community life and the relational resources
of individuals; (2) Trust refers to attitudes that are favourable to social
and political institutions and attitudes that are favourable to others.
Like all operational definitions, these definitions are to some extent arbi-
trary. However, the convergent nature of the indicators (the correlation
between social capital and trust, at the country level, is 0.491) and their
theoretical justification are deemed to provide a solid foundation for the
selected variable. The first group of indicators relates to social capital
(i.e. the density of social life and civil society) (see Figure 11.2). Follow-
ing Putnam, social capital will be defined as the intensity of horizontal
cooperation between actors, the importance of community life and the
strength of ties.
Trust and tolerance on the one hand and must be clearly distin-
guished from social capital on the other. While social capital measures
the intensity of associational life, trust and tolerance refer to a set of
beliefs and attitudes. All three can be understood as components of
social cohesion, however. Generalised trust, for instance, has been found
to be closely related to trust in strangers (Uslaner, 2002; Borgonovi,
2012), while tolerance has been associated with social cohesion by
scholars emphasising the importance of the management of pluralism
for attaining and maintaining a cohesive society (Beauvais and Jenson,
2002). Social capital at the grassroots level has been argued to perco-
late upwards and thus to yield more cohesion and civic engagement
at the societal level (Putnam, 1993, 2000). Accordingly we have devel-
oped a measure of social cohesion based on all three qualities. First
we created separate composite indicators for social capital, trust and
tolerance (see Figures 11.2 and 11.3). Next, because the indicators for
Social Composite index representing the national average of the following

capital seven indicators:
1. Percentage of people claiming active membership of a religious

association
2. Percentage of people claiming active membership of a sports or
cultural association
3. Percentage of people claiming active membership of a political or
union association
4. Percentage of people claiming active membership of another
type of association
5. Percentage of people claiming to spend time with other people in
various contexts ‘every week’ or ‘once or twice a month’
6. Average number of associations that people claim to belong to
7. Proportion of people involved in voluntary work in at least one
association
Source: World Values Survey 1999–2002; Inglehart R. et al (2004), ‘Human

Beliefs and Values: a cross-cultural sourcebook based on the 1999–2002
values surveys’, Siglo XXI Editores, Mexico.
Figure 11.2 Composite index of social capital

Trust Composite index representing the national average of the following

seven indicators:
1. Proportion of people stating that ‘you can trust most people’
(versus ‘you can never be too cautious when dealing with other
people’) Proportion of people placing ‘significant trust’ or ‘some
trust’ in the following institutions:
2. The army
3. The police
4. The justice system
5. Parliament
6. Trade unions
7. Administration
Tolerance Composite index representing the national average of the following
five indicators: Proportion of people who did not mention the
following groups as people they would rather not want to have as
neighbours:
1. People of a different race
2. Foreign workers/immigrants
3. People with aids
4. Homosexuals
5. Proportion of respondents citing tolerance in the 5 qualities
(from a list of 11 qualities in total) that parents must seek to
encourage in their children
Figure 11.3 Composite indexes for trust and tolerance
trust and tolerance proved highly correlated (0.33) at the national level,
a combined trust/tolerance measure was created by averaging the two
indexes for trust and tolerance.1 In the final step, this combined mea-
sure was merged with the social capital measure to create the measure
for social cohesion, which represents the average of the two. Social cohe-
sion in this study thus represents a national-level construct combining
social capital on the one hand and trust/tolerance on the other (for a
slightly different way of defining social cohesion, see Chapter 10 in this
book).
Before exploring the impact of education on cohesion, some prelim-
inary analyses of the relations between cohesion and a range of more
general socioeconomic were conducted.
The social correlates of social cohesion
The dynamics of the labour market prove strongly correlated with both
aspects of cohesion, that is social capital (the correlation is significant:
0.653) and trust/tolerance (the correlation is significant: 0.642).2 So, a
high level of cohesion is unlikely in the event of a depressed labour

market. However, while societies that privilege integration through work
tend to be more cohesive than other societies, the relation between the
two dimensions is not strictly mechanical.
Social cohesion tends to be lower when income inequalities are
greater, although the link is not as strong as the relation between the
dynamics of the labour market and cohesion. The correlation between
inequalities and trust is significant (–0.456), whereas it is not signifi-
cant with the social capital variable (–0.301). The combination of the
‘trust/tolerance’ variable and the ‘inequalities/poverty’ variable’ indi-
cates that a high level of cohesion is unlikely when inequalities are
pronounced, with the notable exception of the United States, a coun-
try that appears to tolerate high levels of inequality, as previously
demonstrated (e.g. Massey, 2007).
These results confirm some of the findings of the study by Green et al.
(2006), according to which the level of social cohesion is strongly cor-
related with the level of income inequality measured by the Gini index.
Taken together, these macro-social factors account to a significant degree
(67.1 per cent) for social cohesion (as shown by the model presented
in Table 11.1). We may conclude that these general socio-economic
correlates significantly impact the level of social cohesion of a given
country.
Educational factors of social cohesion
There are many ways of characterising educational systems. In the

specific context of this text – focusing on the creation of social
cohesion – we focus on two characteristics: educational integration and
educational inequality. We understand the former to include notions of
Table 11.1 Inequalities, GPD, dynamics of the labour market and social
cohesion.
Predictors Standardised T Sig.

coefficients beta
(Constant) 3.328 0.003

Dynamics of the labour market 0.448 3.355 0.003
GPD 1999–2003 0.408 3.101 0.005
Income’s inequalities −0.256 −2.149 0.043
Dependant variable: social
cohesion adjusted R2 = 0. 671;
sig. = 0.000 (N = 26).
coverage and achievement, and thus measured it as the combination of

the percentage of the population with upper-secondary education and
aggregate achievement levels at age 15. ‘Educational inequality’ repre-
sents the average of inequality of performance and the weight of pupils’
social origin on this performance (on the basis of PISA tests).3
In order to identify the key educational characteristics linked to social
cohesion, it is also important to examine students’ attitudes and values.
PISA surveys provide original data based on interviews exploring key
aspects of students’ educational experience at the age of 15. Examples of
the questions include: To what extent do you trust education? To what
extent is education useful for employment purposes? At school, do you
feel like a stranger or at home? PISA surveys also aim to explore students’
perceptions of teachers. Examples of questions in this area include: to
what extent do you trust your teachers? To what extent do you think
that teachers are there to help you? In this study, student attitudes
were categorized into broad groups based on different aspects of trust
in education: the perceived usefulness of school, trust in teachers, the
sense of belonging to a school, membership of a peer group and atti-
tudes towards group work. The five groups of attitudes were found to
be correlated,4 which is indicative of relatively different educational cli-
mates in the various countries included in the sample. A macro-variable
aggregating these five groups termed ‘educational cohesion’ was created.
Social cohesion and educational characteristics do not prove sig-
nificantly correlated. Among the variables characterising educational
systems, only educational integration is positively correlated with social
cohesion (0.454 correlation). Educational cohesion and inequalities
between students and inequalities between social groups were not found
to be positively correlated with social cohesion. When all three system
characteristics are entered into a multivariate model (see Table 11.2),
we find that both educational integration and educational cohesion
show a positive and significant link with social cohesion, while the
effect of educational inequality is not significant. These positive links,
moreover, become stronger, and the explained variance increases, when
educational inequality is omitted from the model (see Model 2 in the
table).
The lack of correlation between educational inequalities and social
cohesion may come as a surprise insofar as it might have been assumed
that educational inequalities would generate cultural distance between
pupils, making communication and trust more difficult. They might also
have been assumed to cause stress and anxiety in a context of increas-
ing competition in education. However, at this stage of the analysis,
Table 11.2 Educational predictors of social cohesion.
Model Predictors Standardised t Sig.

coefficients beta
(constant) −0.062 0.951

1 Educational cohesion 0.368 2.077 0.050
Educational inequalities −0.001 −0.004 0.997
Educational integration 0.510 2.904 0.008
(constant) −0.075 0.941
2 Educational cohesion 0.368 2.144 0.043
Dependant variable: social cohesion
Model 1: adjusted R2 = 0.248;
Sig = 0.026 (N = 26)
Model 2: adjusted R2 = 0.281;
Sig = 0.009 (N = 26)
social cohesion appears to be determined to a greater extent by gen-

eral socio-economic characteristics than by the specific characteristics
of educational systems. Through the level of integration and acquired
knowledge and the everyday climate in school, education seems to
have a more limited impact on social cohesion than general socioeco-
nomic factors such as the dynamics of the labour market and income
inequality.
Returns of education and social cohesion
It is impossible to conclude that education plays a marginal role based

solely on the previous finding. This may rather suggest that the relations
between education and social cohesion are mediated by the specific
impact of education on individual trajectories. It is possible to assume
that the link between societies and education is governed by the social
and economic value given to education and educational qualifications.
Therefore, it is important to examine the effect of both school hier-
archies and degrees on individuals’ integration and their impact on
social cohesion at a macro-level. The core issue is then the specific
ways through which educational systems impact on societies. Based
on Kerckhoff (2000), societies may be said to assign their members to
specific positions through two key processes: firstly, the assignment of
individuals to an educational track and career, secondly, the assignment
of individuals to specific social positions on the basis of their previous
educational career.
This line of inquiry appears to be a promising avenue. International

comparative studies of social mobility have shown that while educa-
tional qualifications are the main factors of professional and social
integration, the strength of the relation between educational credentials
and the social positions that young people get when leaving the edu-
cational system varies significantly across countries. The international
comparative studies conducted by Shavit and Muller (1998) showed that
countries tend to differ more in terms of the relation between educa-
tion and employment (i.e. when considering the benefits brought by
qualifications on the labour market) than in terms of the scale of social
inequalities at school (the key role of the various forms of transition
between school and the labour market is reassessed in Dumas, Méhaut
and Olympio in this volume). The key differentiating factor seems to be
the role played by educational inequalities in the social fate of individ-
uals, and not educational inequalities themselves. The ‘impact’ variable
was constructed to take into account the amount to which young peo-
ple draw a ‘return’ from their educational qualifications on the job
market.
The ‘impact’ variable is based on two aggregate variables: the impact
of education on employment measured by the employment rate of
higher education graduates compared to the employment rate of indi-
viduals with a qualification below upper secondary education; and the
impact of education on wage levels measured by the average wage of
individuals with a tertiary education compared to the average wage
of individuals with a qualification below upper secondary education.5
The ‘impact’ variable is indicative of the economic return of a qualifi-
cation, especially of a higher education degree or qualification, which
is the most discriminant educational level in the sample of rich coun-
tries under study here (the higher the variable, the greater the benefits
of a degree in terms of professional integration and/or wage level). The
‘market’ value of educational qualifications is deployed in a specific eco-
nomic environment, which does matter. Educational qualifications are
more likely to be a differentiating factor when wage levels are unequal
or when educational qualifications are rare. However, it is important
to note that these relations vary across countries. Graph 1 shows that
the impact of education, which varies significantly across countries,
is greater in the United States than in Japan – two countries where
higher education qualifications are as widespread but where the level
of wage inequality is significantly different. Similarly, in countries with
a less developed higher education system, the impact of education is far
greater in Central European countries than in Mediterranean countries.
10.00
Sweden
8.00 Norway
Netherlands
Denmark
Canada United States
Australia
Social cohesion
Switzerlands
6.00 Finland
Ireland
Luxemburg
United Kingdom
Belgium
Korea Austria
4.00 France Germany
Spain
Greece Czech Republic
Italy
Slovakia
2.00 Portugal
Japan Poland
Hungary
0.00
0.00 2.00 4.00 6.00 7.00 10.00

Impact of diploma on the job market
Figure 11.4 Impact of diploma and social cohesion
Figure 11.4 highlights an unexpected and important relation between

the impact of education (i.e. its returns in professional life) and social
cohesion: the greater the impact of education, the lower the level of
social cohesion (–0.547 correlation).
The meaning of this relation remains to be determined: Are quali-
fications used as a basis for hiring when there is little trust in others
or when there is a low level of social capital? On reverse, when edu-
cational qualifications are almighty, to what extent is it because their
impact undermines trust and social capital insofar as they provide key
benefits to those they protect, thereby depriving those with little or no
qualifications of any chance of success?
In any case, there appears to be a tension between the individuals’
benefits of a qualification on the labour market and social cohesion at
the country level, thus suggesting an ecological paradox: close relations
between education and work (i.e. large benefits of education for individ-
uals) are, at the level of societies, associated with lower levels of social
cohesion. There is a marked contrast between countries with a high level
of cohesion and a low impact of education (i.e. Northern European
countries, and also Australia) and less cohesive countries where the
impact of education is very high and the level of social cohesion very
low (especially Central European countries).6 Another group of coun-
tries that includes Korea, Spain and Japan is characterised by a relatively
low impact of education and a relatively low level of social cohesion.
The United States are a notable exception because of a relatively high
impact of education and a high level of social cohesion.
More generally, the negative correlation between the impact of edu-
cation and social cohesion (and, at a more global level of analysis,
the strength of the relation between the impact of credentials and
social reproduction, examined in Dubet et al., 2010a) suggests that the
analyses conducted by the American sociologist Randall Collins (1971)
on the notion of credentialism need to be called up. Collins defines
credentialism as the tendency of modern societies to base access to
social positions on educational attainment, and on increasingly high-
level qualifications and credentials, on the grounds that this is socially
fair and economically efficient. He maintains that increased levels of
education are not necessarily a functional response to the needs of
the economy but may represent a defensive strategy of social clo-
sure.7 Educational qualifications operate then as criteria of exclusion
since individuals with few or no qualifications are marked by a stigma
of professional incompetence and (in some cases) by social indignity.
Professional groups tend to use a defensive strategy that involves impos-
ing a strong belief in their educationally sanctioned skills and in the
legitimacy of the social benefits they draw from them.
The effect of credentialism is to rigidify and to restrict the allocation of
positions, since access to some jobs is limited to individuals with specific
educational qualifications. Selection operates from the outset, with very
little subsequent professional mobility, and credentialism serves to rein-
force social stratification and reproduction. Actually, if one confronts
indicators of degrees ‘emprise’ as defined previously, and indicators
of intergenerational transmission of income in a sample of countries
(defined as in d’Addio, 2007), what is observed then is a strong positive
correlation (0.745): the higher ‘emprise’, the stronger the reproduction
of income from fathers to sons.8 Degrees are then a conveyor belt of
social inequalities, not only because of the unequal assets families give
to their children to acquire them, but all the more so when degrees
impact strongly on positions on the labour market.
Moreover, the belief in educational credentials tends to increase the
sense of exclusion felt by the least qualified members of society, who
may conclude that they have missed their opportunity. Credentialism
also weighs on labour relations since educational credentials are often
allowed to prevail over merit and skills acquired on the job by assigning
positions to diploma holders and by imposing obstacles to career devel-
opment on individuals who are not necessarily less competent, despite
having lower qualifications. This may be further accentuated by the fact
that the qualities sanctioned by educational qualifications are some-
times unrelated to or remote from the qualities required by professional
activities. All in all, credentialism may jeopardise justice (Duru-Bellat,
2009; Dubet, 2010).
The notion of credentialism also suggests that it is important to
examine (as part of an analysis of the foundations of social cohesion)
perceptions of social reality. The perception of social inequalities may
be viewed as one of the key factors of social cohesion: a society in which
members believe that inequalities are acceptable and legitimate would
be more cohesive than a society that does not hold this belief.
Real inequalities, perceived inequalities

and social cohesion
The perception of inequalities can be assessed based on the answers

given by respondents to a question included in ISSP surveys9 : ‘are
inequalities too great in your country?’. The ISSP surveys also include
two questions relating to the perception of inequalities themselves: ‘to
what extent would you say that in your country, people are rewarded for
their efforts?’ and ‘to what extent would you say that in your country,
people are rewarded for their skills or abilities?’ Since the answers to the
two questions are heavily redundant (0.973 correlation), a single ‘merit’
variable was created.
It is important to note that a number of recent comparative stud-
ies (Noll and Roberts, 2003; Osberg and Smeeding, 2006; Duru-Bellat
and Tenret, 2012) showed that the perceptions of social inequalities
observed among members of different societies are not an accurate
reflection of the objective inequalities measured by various indicators.
The same result is found in the sample used here: the perception that
inequalities are excessive is not an accurate reflection of the objective
social inequalities measured by income inequalities (–0.172 correlation).
France and the United States provide contrasting examples of this dis-
junction. Whereas the scale of inequality is almost twice as great in the
United States compared to France, a far greater percentage of the French
population believe that inequalities are excessive.
These judgements do matter, since we observe (see Figure 11.5) that
the feeling that inequalities are too large is generally associated with
10.00
Sweden
8.00
Norway
United States
Social cohesion
Australia
6.00 Canada
Austria
United Kingdom
4.00 Germany France
Slovakia
Czech Republic Spain
2.00 Portugal
Japan Poland
Hungary
0.00
60.00 70.00 80.00 90.00 100.00

Inequalities too large
Figure 11.5 Perceptions of inequalities and social cohesion
a lower level of social cohesion or that a high level of social cohe-

sion is associated with a lower tendency to complain about inequalities
(in other words, the greater the tendency to complain about inequal-
ities, the lower the level of social cohesion). The correlation between
social cohesion and ‘real’ inequalities is –0.451. We may conclude that
while the level of social cohesion is associated with the real level of
income inequality, it is associated still more strongly with the perceived
level of income inequality since, based on answers to the question ‘are
inequalities too large?’, the correlation is –0.675. As a result of incor-
porating this variable in a model designed to explain social cohesion
statistically, it was found that for an equivalent level of income inequal-
ity, the mere fact of perceiving inequalities as excessive has a significant
and negative impact on social cohesion. Therefore, the perception of
inequalities may be said to be an important factor of social cohesion.
The level of tolerance of inequalities is indicative of the strength
of a society’s belief in merit (meaning that skills, efforts or abilities
are rewarded), insofar as degrees sanctioning merit are deemed to be
fair and open: in countries where society is supposed to reward merit,
inequalities are less likely to be perceived as being excessive (–0.723 cor-

relation). This result is so clear-cut that the belief in merit (conceived
as an indisputable principle of justice) may be said to constitute an
ideology promoting an acceptance of social inequalities.
In examining the relations between perceptions and social cohesion,
the correlations between social cohesion and the two variables of per-
ception and explanation prove very strong. A strong, negative and
significant correlation (–0.675) was found between a high level of per-
ceived inequalities and social cohesion. The same is observed, though
in a somewhat less pronounced manner, concerning the belief that
merit is rewarded and social cohesion (0.535 correlation). As expected,
both perceptions are inversely related: the perceived level of inequal-
ity will be lower in a society that has a strong belief in merit being
rewarded.
Society and education in the production of cohesion
Social cohesion is determined by a whole range of factors. Although

the identification of strict causal relations is beyond the scope of this
study, it is important to examine the specific relation of each factor with
the development of social cohesion, on the basis of multivariate mod-
els taking into account the various characteristics of societies and their
educational system.
The first group of models aims to assess the statistical impact of
global socioeconomic characteristics and the general characteristics of
educational systems on social cohesion. Socioeconomic factors (the
dynamics of the labour market, GDP and income inequality) account for
67 per cent of the explained variance of social cohesion (see Table 11.1).
Based on a model aimed at assessing the impact of the characteristics of
educational systems (integration, inequalities and cohesion), their con-
tribution to the explained variance of social cohesion was found to be
very limited (25 per cent; see Table 11.2). So, the general characteristics
of societies explain the level of social cohesion to a greater extent than
educational characteristics alone, suggesting that the specific influence
of educational systems is less important than the economic and social
framework in which their impact is embedded.
Then, since the impact of education (here the returns of educational
qualifications) is at the crossroads of education and society, we incor-
porated this variable in the model showing the effects of educational
characteristics on social cohesion (i.e. the model shown in Table 11.2).
As shown in Table 11.3, adding this variable significantly enhances
Table 11.3 Educational variables, impact of educational qualifications

and social cohesion.
Predictors Standardised T Sig.

coefficients beta
(Constant) 1.493 0.150

Educational cohesion 0.389 2.713 0.013
Impact of diploma −0.484 −3.325 0.003
Adjusted R2 = 0.500; Sig = 0.000 (N = 26)
the explanatory power of the model, while it does not undermine the
effects of the educational conditions. It is confirmed that the level of
educational integration and the level of educational cohesion are pos-
itively and significantly associated with social cohesion. Beyond these
variables, and for countries with similar levels of educational integra-
tion and cohesion, the level of the returns of educational qualifications
shows a strong negative link with social cohesion.
So, in a nutshell, there are two ways of explaining social cohesion
statistically – either by invoking ‘macro-social’ characteristics or by
using ‘macro-educational’ characteristics. While the explanatory power
of both families of models is high, the first one accounts to a greater
extent for the variety of levels of social cohesion observed in the
sample. It is confirmed that education, the level of educational integra-
tion and achievement, the style of educational relationships and the
impact of educational qualifications on subsequent job prospects are
all together significant factors of social cohesion, although they have
a more limited impact than general socioeconomic characteristics such
as economic wealth, the dynamics of the labour market and levels of
inequality.
To end, we incorporated the various social and educational charac-
teristics within the same model in order to examine their combined
impact. We only included the variables which showed a significant
link with social cohesion. The resulting model shows a high level
of explained variance (62.3 per cent, see Table 11.4).10 Although the
dynamics of the labour market still appear to be the most impor-
tant parameter and the negative impact of income inequalities is still
observed, the impact of these socioeconomic characteristics actually
combine with the impact of the general characteristics of educational
systems, that is educational cohesion and (still negatively) the economic
Table 11.4 Social and educational variables and social cohesion.
Predictors Beta coefficient t Sig.
(Constant) 2, 889 0.009

Dynamics of the labour market 0.493 3, 328 0.003
Income’s inequalities −0.255 −1, 994 0.059
Educational cohesion 0.256 2, 058 0.052
Impact of diploma −0.262 −1, 798 0.087
Adjusted R2 = 0. 623; sig. = 0.000 (N = 26).
impact of the level of education. This is indicative of the extent to

which the level of social cohesion within countries with similar social
structures may vary according to the educational climate, and in par-
ticular the perceived quality of human relationships (peer relations
and student–teacher relations). However, the high level of competitive-
ness fostered by high returns of education may counteract the positive
impact of a good climate.
Moreover, as a key factor of social cohesion, the influence of
perceptions is confirmed (see Table 11.5): based on a similar level of
income inequalities, social cohesion is likely to be lower when inequal-
ities are perceived to be significant. The level of income inequality is
still negatively associated with the level of social cohesion – that is
social cohesion is lower when income inequalities are large – but the
perceptions of such inequalities matter significantly (note once again
that the two variables are not correlated). In some non-egalitarian
countries (such as the United States and Australia), inequalities are
less likely to be perceived as being significant, and this perception of
Table 11.5 Educational cohesion, inequalities and perception of

inequalities, and social cohesion.
Predictors Beta coefficient T Sig.
(Constant) 7.601 0.000

Educational cohesion 0.497 3.918 0.002
Income’s inequalities −0.319 −2.574 0.024
Inequalities too large −0.841 −6.637 0.000
Dependent variable: social cohesion
Adjusted R2 = 0. 776; sig. = 0.000 (N = 16).
objective inequalities (itself based on a particular way of accounting for

inequalities, i.e. based on merit) may explain why these countries have
a high level of social cohesion. So, the key variables are the levels of
objective (i.e. measured) and subjective (i.e. perceived) inequality. How-
ever, educational cohesion remains consistent with social cohesion in all
models. More than three quarters (77.6 per cent) of the variance of social
cohesion is explained by the level of social inequality, the perceived level
of social inequality and educational cohesion.
The findings concerning the general relations between education and
social cohesion back the results of the study by Green et al. (2006)
based on the hypothesis that the impact of education on social cohe-
sion depends on the socialisation produced by education (measured
here in terms of educational cohesion) and its relations with employ-
ment (measured here in terms of the returns of education). This needs
to be nested within a wider context determined by the dynamics of the
labour market and the level of inequalities, but also by a whole range of
historical and political factors that were not explicitly incorporated into
the framework of this study.
Education proves to affect social cohesion, although not through the
channels that are generally privileged, that is the percentage of people
in full-time education and achievement levels. By contrast, educational
attitudes and the general educational climate (referred to here as educa-
tional cohesion) were found to play some role. Social cohesion appears
to be generally weakened by the impact of degrees on social positions
and income levels, even if the negative relation between the impact of
education and cohesion cannot be isolated from the broader historical,
social and economic context. It remains that general socioeconomic fac-
tors (particularly the level of economic wealth and vitality) are probably
more important determinants of social cohesion than the more specific
characteristics of educational systems.
We end by conducting a multilevel analysis of social cohesion (see
the textbox of Figure 11.6) to highlight contrasting effects of education
across micro- and macro-levels of analysis.
While computing multilevel models is part of the common method-
ology today, this practice can be debated here for two kinds of reasons.
One may argue that social cohesion is defined only at the macro-level
rather than as a sum of individuals’ attitudes. Beyond this theoretical
point, a second and more methodological one argue that using multi-
level models may be inaccurate when some of the units are not part
of a random sample, as it is the case here with the countries under
study.
262
A multilevel analysis of social cohesion

Since the different variables that define social cohesion can be observed (in the precise definition
retained in this study) both at the individual and at the aggregate levels, we computed multilevel
models, to try to disentangle the possible impact of education at those two levels. The models
confirm the previous results: individuals’ level of education proves associated with social cohesion
attitudes, while at the country level, the stronger the impact of diploma, the weaker social cohesion.
As the notion of ‘ecological fallacy’ suggests, while education (having a degree) is positive for
individuals’ integration (linked to indicators of trust, tolerance, social capital), at the country level,
giving degrees an important weight on social position runs counter to social cohesion. However,
this impact of diplomas is much weaker (even if it is always negative) when we add the country
economic dynamism in the model, which proves extremely important as far as social cohesion is
concerned.
Multilevel Regressions on the Social Cohesion
Parameters Model 1 Model 2 Model 3 Model 4 Model 5
Empty Only with With individual country-level variables
individual
variables
Fixed effects
Constant 3.484∗∗∗ 3.784∗∗∗ 2.771∗∗∗ 4.020∗∗∗ 3.033∗∗∗
Individual-level variables
education level (ref. high)
middle −0.329∗∗∗ −0.329∗∗∗ −0.329∗∗∗ −0.329∗∗∗
low −0.517∗∗∗ −0.517∗∗∗ −0.517∗∗∗ −0.517∗∗∗
Country-level variables
Dynamics of the labour market 0.189∗∗∗ 0.161∗∗∗
Income’s inequalities 0.016+
Educational cohesion 0.066+
Impact of diploma −0.183∗∗ −0.063+
Random effects
level 2: Between-country 0.629∗∗∗ 0.563∗∗∗ 0.275∗∗ 0.408∗∗∗ 0.263∗∗
variance
level 1 : Within-country 1.586∗∗∗ 1.552∗∗∗ 1.552∗∗∗ 1.552∗∗∗ 1.552∗∗∗
variance
Proportion of variance
attributable to each level
level 2: Between-country (%) 28.40 26.62 15.05 20.82 14.49
level 1 : Within-country (%) 71.60 73.38 84.95 79.18 85.51
Between-country variance 10.49 56.28 35.14 58.19
explained (%)
Within-country variance 2.14 2.14 2.14 2.14
explained (%)
(-2IogL) 144,382 143,43 143,419 143,428 143,417
N (Countries) 27 27 27 27 27
N (Individuals) 43,702 43,702 43,702 43,702 43,702
∗∗∗ P < 0.001
∗∗ P < 0.01
+ P < 0.10
While computing multilevel models is part of the common methodology today, this practice can be
debated here for two kinds of reasons. One may argue that social cohesion is defined only at the
macro-level rather than as a sum of individuals’ attitudes. Beyond this theoretical point, a second
and more methodological one argue that using multilevel models may be inaccurate when some
of the units are not part of a random sample, as it is the case here with the countries under study.
Figure 11.6 A multilevel analysis of social cohesion

Conclusion
To conclude this necessarily incomplete analysis of social cohesion

based on a sample of economically developed countries, it is important
to emphasise that the key variables found to be particularly influen-
tial are not the variables most frequently examined in the literature, so
that the present results may be judged heuristic, even if we cannot draw
definitive conclusions. What we observed is, first, that a strong impact
of education, assessed by the economic returns of educational qualifi-
cations, seems to be relatively deleterious to social cohesion. While a
strong correspondence between educational credentials and jobs has a
positive impact on the social and professional integration of individuals,
it also increases the exclusion of those without qualifications and inten-
sifies inter-individual competition. Social cohesion is thus weakened at
an aggregate level, providing an example of the contradictions that may
arise in observations at the level of individuals and observations at the
level of social groups.11
Similarly, the various societies under comparison were found to be
more cohesive when the belief that merit is rewarded is widespread
and when social inequalities are not perceived as being pronounced.
The strength of these perceptions is such that they cannot be viewed
as a simple reflection of real inequalities insofar as the perceived
legitimacy of inequalities is never an accurate reflection of objective
inequalities.
However, this research bears some important weaknesses. It confronts
data concerning either young people (their performances or attitudes
at 15) or adults (their mean level of education, their income, their
judgements on inequalities). That amounts to postulating that societies
are stable, so that, for example, adults did experience the same educa-
tional climate as young people do, and reciprocally that young people
will experience the same social or economic conditions as the adults
of today, notably as far as the ‘emprise’ of degrees is concerned. This
is a common limit of all cross-sectional study, so that, whatever the
methodological sophistication of the analysis, it is never possible to
strictly establish any causal relationships. Even if we used here words
such as ‘impact’ or ‘factors’, we cannot know precisely in which sense
the causality runs, which would require longitudinal data on successive
generations.
Moreover, our study involves sets of variables that cannot be truly
dissociated, as it is invariably the case in comparative sociology. For
example, the ‘emprise’ variable cannot be isolated from more gen-
eral socioeconomic factors such as the level of wealth or of income
inequalities, and the state of the labour market. Even if they use
relatively precise indicators, sociologists must be wary of benchmark-
ing and should not be tempted to construe indicators as embodiments
or reflections of reality. The nature of a society and of a social system
cannot be reduced to a small number of measurements, however reli-
able these may appear to be. It is also important not to forget that
national averages may conceal significant internal differences within
a given society and educational system. We may also emphasise that
nothing can replace more detailed qualitative comparisons – a task that
remains to be carried out. The only definite fact is that the benefits of
education are not merely dependent on what happens before school, that
is as a result of the mean level and social inequalities of attainment,
but are also dependent on what happens after school, that is in the
mechanisms governing the impact of the level of education on individ-
uals’ life and in the perceptions and legitimations of social inequalities
themselves.
Notes
1. Data from World Values Surveys; aggregate of responses to four waves of sur-
veys: 1981–1984; 1989–1993; 1994–1999; 1999–2004). World Values Surveys
use a range of questions aimed at measuring trust and tolerance. Only those
questions for which data were available for all of the countries in the sample
were selected. It was also important to ensure that the correlations between
the variables were positive. Questions such as trust in the educational system
were therefore excluded, as were questions concerning trust in the Church or
the press. The scales of trust and tolerance are a simple average; Cronbach’s
alpha was 0.798 for trust and 0.893 for tolerance.
2. A ‘macro’ variable expressing the dynamics of the labour market was created
by averaging the following six variables:
– Average employment rate between 1995 and 2005 of population aged

15–60
– Average economic activity rate (including the unemployed) between 1995
and 2005 of population aged 15–60
– Average employment rate of population aged 55–64 between 1995 and
2005
– Average female employment rate between 2000 and 2005
– Average unemployment rate between 1995 and 2005 Average long-term
unemployment rate (more than 12 months) between 1995 and 2005.
3. Both OECD data (available in Education at a Glance) and PISA surveys (2000.
2003, 2006) were used, providing comparative data on reading, numeracy
and science scores obtained by students aged 15. The average national data
from each of the three surveys were calculated to ensure greater stability.
4. Cronbach’s Alpha was 0.760.
5. The data are from 2002 (Education at a Glance, OECD, 2004). For Greece,
Japan and Slovakia, the ‘impact’ variable only measures the employment
rate differential.
6. This relation was ‘tested’ by altering the sample to determine whether some
atypical countries account for the trend: by removing Hungary and Slovakia
from the sample on the grounds that they are in an extreme position (since
they show a very strong impact of education and a low level of social cohe-
sion), the relation remains significant and negative though less pronounced
than for the sample taken as a whole.
7. A contemporary version of this thesis has been developed by the American
sociologists Grubb and Lazerson (2004), who refer, in discussing American
students’ educational choices, to ‘self-defense’ strategies (see also Duru-
Bellat, 2008).
8. This analysis has been achieved on only 12 countries, due to available data;
see Dubet et al. (2010b).
9. The International Social Survey Program achieve international comparative
surveys since 1985. The ISSP has conducted several surveys on the perception
of social inequalities, the most recent one being made in 1999. This piece of
information was available for only 16 countries of our sample.
10. This percentage of explained variance is lower than in the model that incor-
porates GDP since GDP is negatively correlated to the variable impact of
education (.501, Sig=0.008).
11. A contradiction of this kind was underlined by Boudon (1973), between
the economic value of a qualification at an individual level and the
devaluation of qualifications at a collective level, a finding later corrob-
orated by a number of studies (see in particular Duru-Bellat, 2006 and
2008).
References
Beauvais, C. and Jenson, J. (2002) ‘Social Cohesion: Updating the State of the
Research’, Ottawa, CPRN Discussion Paper.
Borgonovi, F. (2012) ‘The Relationship between Education and Levels of
Trust and Tolerance in Europe’ The British Journal of Sociology, 63, 1,
pp. 146–167.
Boudon, R. (1973) L’inégalité des chances, Paris: Colin (Education, Opportunity and
Social Inequality, New York: Wiley, 1974).
Coleman, J. S. (1990) Foundations of Social Theory, Cambridge MA: Harvard
University Press.
Collins, R. (1971) ‘Functional and Conflict Theories of Educational Stratification’
American Sociological Review, 36, 6, pp. 1002–1019.
Dubet, F. (2010) Les Places et les Chances, Paris: Seuil.
Dubet, F., Duru-Bellat M., and Vérétout A. (2010a) Les sociétés et leur école. Emprise
du diplôme et cohésion sociale, Paris: Seuil.
Dubet, F., Duru-Bellat M., Vérétout A. (2010b) Les inégalités scolaires entre
l’amont et l’aval, Sociologie, 1, 2, pp. 177–197.
Duru-Bellat, M. (2006) L’inflation scolaire. Les désillusions de la méritocratie, Paris:
Seuil.
Duru-Bellat, M.(2008) ‘Recent Trends in Social Reproduction in France: Should

the Political Promises of Education be Revisited?’ Journal of Education Policy, 23,
1, pp. 81–95.
Duru-Bellat, M. (2009) Le mérite contre la justice, Paris: Sciences Po, Les Presses.
Duru-Bellat, M. and Tenret, É. (2012) ‘Who is for Meritocracy’ Comparative
Education Review, 56, 2, pp. 223–247.
Green, A., Preston, J., and Janmaat, J. G. (2006) Education, Equality and Social
Cohesion, New York: Palgrave Macmillan.
Green, A. and Janmaat, J. G. (2011) Regimes of Social Cohesion: Societies and the
Crisis of Globalization, Basingstoke: Palgrave Macmillan.
Grubb, W. N. and Lazerson M. (2004) The Education Gospel, Harvard: Harvard
University Press.
Ingelhart, R. (1977) The Silent Revolution, Princeton: Princeton University Press.
Ingelhart, R, (1993) La transition culturelle dans les sociétés industrielles avancées,
Paris: Economica.
Kerckhoff, A. C. (2000) ‘Transition from School to Work in Comparative Perspec-
tive’ in Hallinan, M. T. (ed.), Handbook of the Sociology of Education, New York:
Springer, pp. 453–474.
Massey, D. S. (2007) Categorically Unequal. The American Stratification System,
New York: Russel Sage Foundation.
Noll, H-H. and Roberts, L.W. (2003) ‘The Legitimacy of Inequality on Both Sides
of the Atlantic. A Comparative Analysis of Attitudes in Canada and Germany’
The Tocqueville Review, 24, 2, pp. 153–189.
OECD-CERI (2006) Measuring the Effects of Education on Health and Civic Engage-
ment, Proceeding of the Copenhagen Symposium.
Osberg, L. and Smeeding, T. (2006) ‘ “Fair” Inequality? Attitudes toward Pay Dif-
ferentials: the United States in Comparative Perspective’ American Sociological
Review, 71, pp. 450–473.
Putnam, R. D. (1993) Making Democracy Work: Civic Traditions in Modern Italy.
Princeton: Princeton University Press.
Putnam, R. D. (2000) Bowling Alone: The Collapse and Revival of American Commu-
nity. New York: Simon & Schuster.
Putnam, R. D. (2007) ‘E Pluribus Unum: Diversity and Community in the Twenty-
first Century – The 2006 Johan Skytte Lecture’ Scandinavian Political Studies, 30,
2, pp. 137–174.
Shavit, Y. and Müller W. (1998) From School to Work. Oxford: Clarendon Press.
Uslaner, E.M. (2002) The Moral Foundations of Trust. New York: Cambridge
University Press.
12
The Mismatch between Level of
Education and Job Qualifications:
A Source of Mistrust and
Intolerance Depending
on National Context?
Vanessa di Paola and Stéphanie Moullet
This article is just one contribution to the wider investigation into what
education does to society.1 Its aim is to examine education systems
against the background of what can be described, very broadly, as social
cohesion (Green et al., 2003). There is no agreement on the definition
of this multidimensional notion, with its hazy boundaries. Thus it can
be measured in terms of equality and social mobility or of crime levels,
cultural conflict, gender equality, civility and democracy. It may also be
defined in terms of its basic components: (i) a feeling of trust, (ii) a feel-
ing of belonging, and (iii) the manifestation of these feelings in objective
behaviours, as Chan and Chan (2006) argue. Another difficulty in grasp-
ing the notion of social cohesion lies in the confusion of different levels
of analysis: social cohesion is a macro-level phenomenon and is not sim-
ply the sum of its individual components (Green and Janmaat, 2011).
When it comes to analysing the role of education in social cohesion, the
relationship between the micro- and macro-effects is not always straight-
forward: for example, an increase in an individual’s level of education
reduces his or her risk of involvement in criminal activity, whereas the
countries with the highest levels of education are not necessarily those
with the lowest crime levels. At the macro-level, the comparative anal-
yses carried out by Green et al. (2003, 2006) have already revealed the
correlations between education, in the broad sense, and indicators of
social cohesion (social unrest, crime, civil liberties, etc.). More broadly,
the impact of education differs from one country to another, depending
267
on the societal context. This makes it necessary to take into account

the conditions under which education is likely to have effects on social
cohesion.
One of the purposes of education is to equip individuals with norma-
tive frameworks, self-confidence and trust in others, in short everything
pertaining, at the level of the individual personality, to social cohesion.
Nevertheless, cohesion also depends on the way in which education and
the labour market are linked. The labour market acts as an interface
between the education system and social cohesion. Our microeconomic
analysis focuses on the role of this interface. More specifically, do the
impact of education on trust and tolerance, two of the individual
components of social cohesion, and that of individuals’ labour mar-
ket situation not interact with each other? Against the background
of intense international competition, individual countries have been
attempting to forge ever closer links between increasingly diversified
education programmes and jobs that are equally diverse; in doing so,
they are running the risk of turning education into a primarily instru-
mental good whose sole aim is to satisfy the needs of the labour market.
Thus education could have two contrary effects on social cohesion: not
only those effects that are usually identified as positive externalities but
also deleterious effects stemming from education to which only little
value, if any at all, is attributed in the labour market. Thus a mismatch
between an individual’s level of education and the level of skill required
for a particular job may give rise to dissatisfaction (Giret et al., 2006)
and hence to mistrust and intolerance (Peugny, 2009).
Furthermore, the effect of such a mismatch on individual trust or tol-
erance is likely to vary with national context, which therefore has to
be taken into account in any attempt to evaluate that effect. For exam-
ple, it might be proposed that individuals experiencing mismatch in
the labour market will be more tolerant and trusting if the unemploy-
ment rate in their country is high, encouraging them to put the lower
return to their educational investment into context. This proposition is
based on the notion of ‘adaptive preferences’ (Sen, 1979; Elster, 1982),
which holds that individuals can put things into perspective by com-
paring themselves to people close to them in similar situations (family
members, friends, and colleagues). Similarly the strength of a country’s
meritocratic ideology may also be postulated to influence the effect of
the mismatch on individuals’ trust and tolerance. Key to this ideology
is the belief that educational qualifications should and also do deter-
mine labour market position. If individuals notice that this situation
does not apply in their own case because they are unable to acquire the
jobs linked to their qualifications, they lose belief in and support for
Vanessa di Paola and Stéphanie Moullet 269
the meritocratic system (Duru-Bellat and Tenret 2009), which is likely

to lead to alienation, distrust and intolerance. This sense of alienation
can be expected to be particularly severe in countries with a strong
meritocratic ideology.
This notion of mismatch takes us back to the more familiar one of
downgrading (or downward mobility), that is a situation in which an
individual’s level of education exceeds that ‘normally’ required for the
job held. Since our analysis concerns some 20 countries, the notion of
mismatch is more appropriate than that of downgrading: the hierar-
chy of educational levels and, more widely, the links between education
and employment differ from one country to another, which leads us
to prefer a notion based on correspondence rather than on ‘grading’ or
classification. Furthermore, the use of databases compiled on a compar-
ative basis places at our disposal education and employment variables
that make it possible, through the standard international classifications,
to apply the indicator of mismatch to all the countries under investiga-
tion. Thus we adopted a statistical indicator constructed in such a way
that, within each country, the situation of those individuals whose level
of education does not equate to that of the majority of people holding
the same type of job as them is characterised by mismatch. We seek to
assess the impact, for controlled levels of education, of individual situ-
ations characterised by objective mismatch on personal feelings of trust
and tolerance. In the following section, the data used are presented. This
is followed by a presentation of the descriptive and econometric results.
Data
The data used are drawn from the fifth wave of the World Value Sur-
vey (WVS) (2005–2008). These data relate to values and opinions and
provide information on the values held in such major spheres as the
family, work, social relations, the environment, religion, economics and
politics as well as on the socio-demographic characteristics of the indi-
viduals surveyed. The national samples are constructed on the basis of
the national population aged 18 and over. As far as the selection of
countries to be investigated is concerned, the analysis relates to a sam-
ple of 20 countries2 that have certain similarities in terms of economic
development and differences in the organisation of their education sys-
tems. Within this sample, only the economically active population is
considered.
Individual opinions, which are the dependent variables in the models
to be estimated, are captured by two indicators, which were adopted
because they are micro-level translations of the macro-level indicators
of social cohesion3 and also because their response rates are satisfactory.
The first of them is interpersonal trust, which has the value 0 when an
individual declares he has no trust in others and 1 otherwise.
The second indicator reflects the level of tolerance as evinced by the
answers to four questions concerning the importance of transmitting
tolerance to one’s children, acceptance of neighbours of a different race,
having neighbours who are HIV-positive and having homosexual neigh-
bours. The indicator represents the sum of the answers to these four
questions [–1 (intolerance); 0 (neutral); 1 (tolerance)].
National context is accounted for by means of various macro-
level indicators. On the one hand, indicators reflecting each country’s
macroeconomic context are taken into account; they include the unem-
ployment rate, the Gini index of income inequalities,4 GNP per capita,
the female employment rate and education expenditure as a share of
GDP. On the other hand, the empirical content of the notion of social
cohesion is captured by indicators of attitudes and opinions that aggre-
gate, at the individual country level, the responses given by individuals,
namely the level of trust in others, attitude towards earnings inequalities
and the level of tolerance. When all is said and done, these indicators
reflect the degree of social cohesiveness in each country. Taking these
macro-level indicators into account is all the more important since they
also help to define the cultural and political contexts that obviously
affect each country’s social cohesion.
Finally, over and above these variables, taking account of national
context when assessing the degree of mismatch between educational
levels and job qualifications makes it possible to put into perspective
the impact of individual properties: the frequency of mismatch at the
country level can be expected to lessen the effect of mismatch at the
individual level.
The data also include individual characteristics, such as gender, age,
marital situation (cohabiting or not), the highest qualification obtained
and labour market situation on the survey date (with a distinction being
made between full-timers and part-timers, non-wage earners and the
unemployed).
The key variable in the analysis, the indicator of mismatch in the
labour market, has the value one when the individual holds a job not
equating to his or her level of education and zero if not. This reflects
the notion that some individuals’ labour market situation does not
accord with the relationship between education level and level of job
that prevails in each country. The mismatch is defined on the basis of
a statistical criterion: any individual whose education level is greater
than that normally required to hold the job in question is in a mis-

match situation. This ‘normalness’ is based on a statistical definition of
the correspondence between education and job. After all, a statistical
measurement of this kind is based on contingency tables that com-
bine education levels and job categories. This correspondence table is
constructed for the whole of the economically active population and
for each of the countries separately. Since each country has its own
employment and qualification structure, the mismatch indicator can-
not be constructed at a general level. With reference to the international
classifications, the following five education levels were adopted: pri-
mary or lower, incomplete secondary, completed secondary (technical
or vocational), completed secondary (academic) and incomplete tertiary
secondary and, finally, tertiary-level qualifications. The five job cate-
gories adopted are as follows: company directors (including farming),
managerial, intermediate occupations, white-collar/clerical and manual.
Finally, the WVS enables us to take account of the unemployed pop-
ulation, since it provides information on the employment situation in
which individuals mainly found themselves in the past. Thus through
their past employment history the unemployed become part of the
definition of mismatch for each country. Since the unemployed pop-
ulation is included in the analysis, the labour market situation on the
survey date has to be taken into account in the econometric estimations.
It should be noted that the mismatch indicator is sensitive to the level of
detail adopted for the education levels and job categories in the contin-
gency table; however, the WVS does not provide a greater level of detail.
Aggregating all tertiary-level qualifications into a single category limits
any evaluation of the mismatch situations affecting graduates, whereas
they are in fact the population most ‘at risk’.
In the following section, the results obtained using the multilevel
approach are presented. This methodological approach enables us not
only to evaluate the relationship between individual mismatch and the
behaviours more or less likely to encourage social cohesion by neu-
tralising the effects of the context within which that relationship can
be observed but also to highlight the extent to which the character-
istics of these contexts impact directly or indirectly on the individual
link between labour market situation and individual feelings of trust or
tolerance.
Education and social cohesion: What is the labour market’s role?

Before attempting to take contextual effects into account, we will iden-
tify the main difference between the countries in terms of indicators
of social cohesion. With regard to the first criterion of social cohesion

adopted, namely interpersonal trust, dispersion among the countries is
high, with the value for the indicator ranging from 4.8 per cent for
Turkey to 67.9 per cent in Sweden (see Table 12.1 and Figure 12.1).
In Chile, Morocco, Mexico, Germany, France and Poland, the propor-
tion of trusting individuals ranges from 13 to 19 per cent. Conversely,
only in New Zealand (51.1 per cent), Finland (58.9 per cent) and Sweden
(67.9 per cent) do more than half of the population say they trust other
people. The degrees of tolerance vary less between the countries and lev-
els also tend to be higher. The degree of tolerance dips below 50 per cent
only in Korea and Turkey, while Great Britain and Sweden have the
highest levels, at 87.8 per cent and 95.9 per cent respectively.
The average level of mismatch varies considerably between countries,
from less than 4 per cent for Morocco, Finland and Russia to almost
20 per cent and more for Argentina, the Netherlands and New Zealand.
It should be noted that, since the construction of the indicator is
dependent on the distribution of education and employment among
individuals within each country, a low value for the indicator does not
mean that that country’s situation is ‘better’ than that of a country
with a high value. After all, the education system’s role in determining
access to a given social position is not the same in all countries (Dubet
et al., 2010). In France, for example, where the dominant model is that
of success through education, there is a strong link between level of
initial education and employment. In the British model, on the other
hand, the link between initial education and employment is looser.
To a greater extent than in France, access to social positions is obtained
through on-the-job training. Consequently, less well-qualified individu-
als can gain access to skilled jobs, which in turn means that more highly
qualified individuals in the same jobs suffer from mismatch. As a result,
the mismatch rates in the two countries cannot be directly compared
without carrying out a comparative analysis of the relationship between
education and employment. Nevertheless, this indicator remains abso-
lutely relevant when it comes to comparing individual situations within
the same country.
Furthermore, the indicators of social cohesion are not all correlated in
the same way with the macroeconomic variables. As might be expected,
the better a country’s macroeconomic situation is, the higher the level
of interpersonal trust is. After all, greater trust goes hand in hand with
high GNP per capita and a high female employment rate, as well as
with low-income inequalities (see Table 12.2). On the other hand, the
level of tolerance is correlated only with GNP per capita (positively),
Table 12.1 Mismatch and indicators of social and macroeconomic cohesion by country (in %).
Country Economically Mismatch Trust in Tolerance Unemployment Gini index of GNP per Female Education
active rate others rate∗∗ 2005 earnings capita employment expenditure
population (total inequalities∗∗ (US $)∗ rate∗∗ 2004 as % of
WVS sample) 2004 2004 GDP∗∗ 2005
Germany 1020(/2064) 6.3 36.8 84.3 10.6 0.3 33 323 59.7 5.1
Argentina 617 (/1002) 19.6 17.5 86.3 14.8 0.5 3 991 3.8
Australia 899(/1421) 6.9 46.1 86.8 5.1 0.3 32 757 63 5.5
Chile 554(/1241) 11 12.5 78.3 8 0.55 5 931 37.8 5.7
Korea 702(/1200) 13.1 28.2 34.9 3.7 0.31 15 242 52.2 7.2
Spain 1141(/1209) 7.4 20 85.3 9.2 0.32 24 580 49 4.6
US 673(/1249) 10.8 39.3 83.5 5.1 0.38 38 793 65.4 7.1
Finland 504(/1014) 3.6 58.9 83.5 8.4 0.27 36 139 65.5 6
France 570(/1001) 6 18.8 75.9 9.3 0.28 33 012 57.7 6
Great Britain 601(/1041) 14.5 30.5 87.8 4.8 0.34 36 662 66.6 6.2
Italy 538(/1012) 10.4 29.2 76.4 7.7 0.35 29 642 45.2 4.7
Japan 699(/1096) 12.4 39 nd◦ 4.4 0.32 36 158 57.4 4.9
Morocco 1022(/1200) 1.2 13 56.2 12.1 0.41 1 889 6
Mexico 826 (/1560) 14.8 15.6 79.8 3.6 0.47 7 273 40.9 6.5
New Zealand 865 (/954) 25 51.1 90.9 3.7 0.34 24 459 66.1 6.7
Netherlands 528(/1050) 20.6 45 85.2 4.7 0.27 37 552 64.1 5
Poland 414(/1000) 5.3 19 70.8 17.8 0.37 6 610 46.4 5.8
Russia 1257(/2033) 2.8 26.2 56.9 7.2 0.37 4 113 62.2 3.8
Sweden 681(/1003) 13.5 67.9 95.9 7.3 0.23 39 607 71.8 6.4
Turkey 588(/1346) 4.1 4.8 43.5 10.2 0.43 5 582 22.3 4
Total 14699
◦ nd:no data.
∗
United Nations, Department of Economic and Social Affairs, Population Division, World Population Prospects: The 2008 Revision, New York, 2009.
273
∗∗ OECD Factbook 2009: Economic, Environmental and Social Statistics – ISBN 92–64–05604–1 – © OECD 2009.
Sweden
Finland
New Zealand
Australia
Netherlands
United States
Japan
Germany
UK
Italy
Korea
Russia
Spain
Poland
France
Argentina
Mexico
Morocco
Chile
Turkey
0 50 100 150 200 250
Trust in others Attitude to income inequalities Tolerance
Figure 12.1 Indicators of social cohesion by country
while opinions on income inequalities are not correlated with any

macroeconomic indicator. The indicators of social cohesion are not
related with the per cent of mismatch in the country either.
The aim of the multilevel analyses carried out at individual level was
to assess the effect of individual experiences of mismatch on the indi-
cators of trust and tolerance and to do so by including a number of
individual explanatory variables, in particular level of qualification and
labour market situation on the survey date, as well as gender, age and
marital status (see Appendix 12.1 and Appendix 12.2 for the full results
of these analyses).
As far as the multilevel modelling is concerned, only those indicators
for which the contextual dimension proves to be relevant during the
first stage (empty model) need to be retained. For this reason, when the
results of the estimations were obtained, the only ones retained were
interpersonal trust and the degree of tolerance. In both cases, the differ-
ences between individuals within the same country proved to be greater
than those between countries: the share of the unexplained variance
attributable to context is 27 per cent and 33 per cent respectively.
While the effect of context on the various indicators of social cohe-
sion does seem to exist, it is difficult to capture in the estimations,
despite the numerous combinations of macro-level variables that we
tested5 (cf. Table 12.3). Thus GNP per capital and the share of education
Table 12.2 Correlation matrix for macro-level variables.
Labour Trust in Attitude to Tolerance Unemployment Gini index GNP Female Share of
market others income rate of income per employment education
mismatch inequalities inequalities capita rate expenditure
in GDP
Labour market 1.0000

mismatch
Trust in others 0.3461 1.0000
Attitude to income −0.1581 −0.0606 1.0000
inequalities
Tolerance 0.3613 0.5320 0.2375 1.0000
Unemployment rate −0.6128 −0.2886 0.2386 −0.0636 1.0000
Gini index of −0.0057 −0.6788 0.0573 −0.2641 −0.0442 1.0000
income
inequalities
GNP per capita 0.2308 0.6872 0.2016 0.6600 −0.1348 −0.7138 1.0000
Female employment 0.2217 0.8195 −0.1286 0.4685 −0.2993 −0.6730 0.6309 1.0000
rate
Share of education 0.5944 0.3619 −0.2115 0.2392 −0.4394 −0.0605 0.3408 0.2729 1.0000
expenditure in
GDP
275
Table 12.3 The significance of the level 2 variables.
Individual dependent
variables of social
cohesion
Level 2 macro-level variables Trust Tolerance

Mismatch in the labour market ns −∗∗∗
Indicators of social cohesion:
Trust −∗∗∗ +∗
Attitude towards earnings inequalities ns Abs
Tolerance +∗∗ +∗∗∗
Macroeconomic indicators
Unemployment rate ns Ns
Gini index of income inequalities −∗∗ ns
GNP per capita ns Abs
Female employment rate +∗∗∗ Ns
Education expenditure (% of GDP) ns Abs
∗ P < .05; ∗∗ P < .01; ∗∗∗ P < .001
‘abs’ indicates that the variable was not retained in the modelling following
the probability ratio test;
‘ns’ indicates that the variable was retained following the probability ratio
test but that it is not significant.
expenditure in GDP prove to be non-significant when it comes to

explaining both interpersonal trust and tolerance. The macroeconomic
variables characterising the labour market (unemployment and female
employment rates) seldom influence individual attitudes: the unem-
ployment rate has no effect, while the female employment rate has a
favourable influence only on trust. It would seem that the feminisation
of the labour market tends to foster trust without having any effect on
the level of individual tolerance.
The existence of income inequalities has the effect of reducing
interpersonal trust but does not influence tolerance.
As far as the indicators of social cohesion are concerned, attitudes to
income inequalities have no effect on individuals’ attitudes, whereas
this same indicator is negatively correlated with social cohesion consid-
ered from a macro-point of view (Dubet et al., 2010). On the other hand,
a country’s climate of trust helps to increase tolerance but, surprisingly,
has a (weakly) negative effect on interpersonal trust. Moreover, the aver-
age level of tolerance in a country (among the whole population, not
just the economically active) has a positive influence on interpersonal
trust and the degree of tolerance.
A number of individual characteristics were taken into consideration
and capture a significant proportion of the variability of the indicators
of trust and tolerance. Thus, while men and women claim to be equally
trusting of others, women claim to be more tolerant (all things being
equal). Age also has no effect on trust, whereas older people (age 65 and
over) are less tolerant than those in the 35–44 age group and young
people (25–24-year-olds) turn out to be the most tolerant. Living with a
partner fosters trust in others but reduces the degree of tolerance.
Individuals’ labour market situation also has different effects, depend-
ing on the nature of the feeling expressed. The fact of being unemployed
on the survey date, rather than being in full-time employment, reduces
trust in others but without affecting tolerance. On the other hand, self-
employed workers claim to be less tolerant but are neither more nor less
trusting.
Finally, mismatch in the labour market is taken into account in two
forms: at the individual level and at the aggregated level of the country.
This method captures both an intra-country effect (produced by the fact
of being – or not being – in a state of labour market mismatch within
the country) and a context effect (produced by the fact of being in a
country with a more or less high level of mismatch, for a given level of
individual mismatch).
The more frequent the gap is between employment situations and
the statistical norm for the education/employment link (i.e. the aggre-
gate level of mismatch), the more individuals express a high level of
intolerance (all other things being equal). The aggregate level of mis-
match does not influence interpersonal trust, however. We can only
speculate about the reason why there is a relationship with tolerance
but not with trust. Possibly people in countries where qualifications
matter less and where there is thus more mismatch feel more vulner-
able in terms of facing competition from immigrants. Perhaps that is
the reason why they are less tolerant (towards immigrants) but not
necessarily less trusting. The more frequent the gap is between employ-
ment situations and the statistical norm for the education/employment
link, the more individuals express a high level of intolerance (all other
things being equal). The link between the micro- and macro-effects
of the mismatch variable may be attributable to what is known as an
‘ecological paradox’: the negative effects of mismatch on the individ-
ual level are not necessarily replicated at the population level. After
all, individuals suffering from mismatch in the labour market are less
trusting and less confident; nevertheless, living in a country in which
the level of mismatch is high does not cause individuals to be less
trusting.
It is shown that, all things being equal and with national contextual
effects taken into account, individual mismatch in the labour market
reduces not only interpersonal trust but also the degree of tolerance.
In other words individuals are both less trusting and less tolerant when
they are suffering from mismatch in their own country.
Alongside these mismatch effects, the education variables also play a
major role, as described in studies on the positive externalities of edu-
cation (Wolfe and Haveman, 2001). More specifically, and controlling
for age, the higher the level of education is, the stronger interpersonal
trust is and the higher the level of tolerance is. Thus over and above any
direct effects that might be expected, the labour market also impacts
indirectly on the link between education and social cohesion through
the education–employment relationship. Individuals in jobs of a lower
level than those held by employees with the same level of education
tend, it is argued, to be less tolerant and less trusting. Thus a shortfall
in the return to an individual’s investment in education is said to foster
mistrust and intolerance. Introducing a variable that captures educa-
tional levels alongside the individual mismatch variable is all the more
important since mismatch is all the more likely to happen the higher an
individual’s level of education is. The least well qualified are at lowest
risk of mismatch.
Finally, taking into account possible indirect effects of national con-
texts mediated through labour market mismatch proves to be irrelevant
to the indicators of interpersonal trust and of tolerance (coefficients
not shown in the two tables6 ). This result suggests that national con-
texts, and in particular mismatch aggregated to the national level, do
not influence the effect of individual mismatch on expressions of trust
or tolerance. To put it another way, while contextual effects directly
influence individual expressions of trust and tolerance, the effect of mis-
match on these same indicators does not vary according to national
context. This does not support our earlier proposition that the effect
of individual mismatch will be smaller in contexts where people suffer-
ing from mismatch find themselves surrounded by people in a similar
situation.
Conclusion
Regardless of their national context, individuals’ experience of mis-
match in the labour market serves to increase levels of mistrust and
reduce levels of tolerance. The national context also has direct effects
on these two indicators, but no indirect effects were identified in our
analysis. However, the value ascribed to education in the labour mar-
ket depends, among other things, on the state of that market; it might
have been assumed, therefore, that the effect of mismatch would differ
depending on the state of the labour market and hence by country. Nev-
ertheless, while the environment in which individuals find themselves
influences what they feel, their own characteristics exert a much greater
influence.
Moreover, whatever the country, and the individual characteristics
being given, mismatch, which is a way of analysing the role of educa-
tion through the prism of the labour market, offsets the ‘positive’ direct
effects of education on individual patterns of behaviour with regard to
trust and tolerance. The effect on individual attitudes of a reduced return
to education in the labour market does not seem to depend on the over-
all level of mismatch in the country as a whole: in countries with a low
level of mismatch, individuals suffering from mismatch in the labour
market do not display greater mistrust or intolerance towards others
than similar individuals in countries with a high level of mismatch.
Finally, the role of individual characteristics in explaining trust or
tolerance differs quite considerably. Thus all things being equal, men
declare themselves to be less tolerant than women but are equally trust-
ing (or mistrusting). The effect of mismatch and of educational levels
is the only one to be the same on both trust and tolerance, which
makes these dimensions key factors in understanding the individual
mechanisms influencing social cohesion.
Notes
1. This paper is part of the EDESCO research project, funded by the ANR and
coordinated by LEST.
2. Argentina, Australia, Chile, Germany, Spain, Finland, France, Great Britain,
Italy, Japan, Korea, Mexico, Morocco, Netherlands, New Zealand, Poland,
Russia, Sweden, Turkey and the United States.
3. To regard the micro-level indicators as reflecting the macro-level concept of
social cohesion may be debatable. We are referring here more to the social
capital dimension. On this concept, see the studies by Ponthieux (2004) or
Putnam (1995).
4. The Gini index measures the degree of inequality in the earnings distribution,
ranging from zero (same income for all) to one (one single individual has all
the income).
5. In the end, in order to satisfy the criterion of log probability minimisation, we
adopted vectors of specific macro-level variables for each indicator. Thus some
variables were retained in the modelling, even though they are not significant.
6. The model with these indirect effects did not significantly improve the
explained variance of both outcomes by comparison to the direct effects
model (i.e. Model 2) and was therefore omitted from the two tables.
References
Algan, Y. and P. Cahuc (2010) Peut-on construire une société de confiance?, Post-face
à la Société de défiance, Albin Michel (editeur P. Askenazy et D. Cohen).
Chan, J., H-P. To, and E. Chan (2006) ‘Reconsidering Social Cohesion: Developing
a Definition and Analytical Framework for Empirical’ Social Indicators Research,
75, 2, Janvier, pp. 273–302.
Dubet, F. and M. Duru-Bellat, A. Vérétout (2010) Les sociétés et leur école, Emprise
du diplôme et cohésion sociale, Seuil, p. 212.
Duru-Bellat, M, and É. Tenret (2009) ‘L’emprise de la méritocratie scolaire: quelle
légitimité?’ Revue Française de Sociologie, 50, 2, pp. 229–258.
Elster, J. (1982) ‘Sour Grapes – Utilitarianism and the Genesis of Wants’ in
A. Sen and B. Williams (eds.) Utilitarianism and beyond, Cambridge: Cambridge
University Press, pp. 219–238.
Galland, O. (2008) ‘La confiance dans les autres’, dans «Les Valeurs des Français»,
Pierre Bréchon (ed.), Presses Universitaires de France.
Giret, J.F., E. Nauze-Fichet, and M. Tomasini (2006) ‘Le déclassement des
jeunes sur le marché du travail’ Données sociales, La société française, Insee,
pp. 307–314.
Green, A. and Janmaat, J. G. (2011) Regimes of Social Cohesion: Societies and the
Crisis of Globalization, Basingstoke: Palgrave Macmillan.
Green, A., J. Preston, and J.G. Janmaat (2006) Education, Equality and Social
Cohesion: A Comparative Analysis, New-York: Palgrave MacMillan.
Green, A., J. Preston, R. Sabates (2003) ‘Education, Equality and Social Cohesion:
a distributional approach’ Compare, 33, 4, pp. 451–468.
Peugny, C. (2009) Le déclassement, Grasset Ed., p. 173.
Ponthieux, S. (2004), ‘Le Concept de Capital Social, Analyse critique’ Contribution
au 10ème colloque de l’ACN, Paris, pp. 21–23 Janvier.
Putnam, R.D. (1995) ‘Bowling Alone: America’s Declining Social Capital’ Journal
of Democracy, 6, 1, pp. 65–78.
Sen, A. (1979) ‘Personal Utilities and Public Judgements: Or What’s Wrong with
Welfare Economics?’ The Economic Journal, 89, September, pp. 537–558.
Wolfe, B. and R. Haveman (2001), ‘Accounting for the Social and Non-Market
Benefits of Education’, in Helliwell J. (eds.), The Contribution of Human and
Social Capital to Sustained Economic Growth and Well-Being (Proceedings of an
OECD/HRDC conference, Quebec, March 19–21, 2000).
13
A Common Civic Culture in
Europe: Has the Process of
European Integration Been
Followed by the Convergence
of Citizenship Values?
Bryony Hoskins and Rebecca Ridley
Theories of Europeanisation (Lawn and Lingard, 2002) and Globalisation

(Ramirez et al., 2006, 2009) have proposed that national education poli-
cies and practices are converging towards a uniform model. The premise
is that national policies are becoming increasingly alike due to the expe-
rience of coping with similar problems and as a result of processes of
competition between countries on the world stage, policy borrowing
and pressure from international organisations (Green, 1999). In terms of
Europeanisation, the European Union (EU) institutions and the Council
of Europe have been identified as key actors in the process of draw-
ing national policies of European countries closer together (Novoa and
Lawn, 2002). In the field of active citizenship the EU has developed
broad common objectives, funded education activities and training,
shared good practices and monitored policy outcomes through indica-
tors (Hoskins et al., 2006, 2008, 2012a; Hoskins and Mascherini, 2009).
The Council of Europe has also been active on developing tools and
sharing good practice on Education for Democratic Citizenship within
its member states for over 15 years and has recently adopted a charter
on Education for Democratic Citizenship and Human Rights Education
(Council of Europe, 2010) that provides a common framework for action
across Europe. The general aim of all these soft policy initiatives has
been to enhance participation in a broad range of activities based on
the values of equality, tolerance and human rights. At the same time
global actors, including the United Nations, have been identified to
281
be pulling the policy discourse towards cosmopolitan values (Ramirez,

2003; Ramirez et al., 2009). Nevertheless, theories of policy conver-
gence rarely discuss the outcomes of the policy process with one notable
exception being the study of Wiseman et al.’s (2011) on citizenship
knowledge and attitudes that we will explore later in this chapter. How-
ever, most studies that explore trends on civic attitudes come from the
political science field and utilise theories of modernisation (Inglehart
and Welzel, 2005) rather than specific policies or policy convergence
to explain these trends. This chapter combines theories of policy con-
vergence and modernisation to test the hypothesis that young people’s
civic attitudes and values in Europe are becoming more similar and that
these attitudes and values are moving towards a common cosmopolitan
ideal.
This chapter will test this hypothesis by identifying trends in civic
values of young people. It will compare two datasets from the IEA, the
Civic Education study (CIVED) from 1999 and the International civic
and citizenship data (ICCS) from 2009. Thus it will identify trends of
citizenship values across ten years. The trends will be examined for the
values of gender equality, tolerance towards immigrants, concepts of
national civic duty and more global forms of civic engagement.
European Union policy convergence
Two levels of policy convergence have been described by Green (1999).

The first level is when the ‘discourse and objectives’ in a set of coun-
tries become more similar. The second, labelled ‘structural convergence’
by Green (1999, p. 56), occurs when ‘actual structures, processes and
outcomes’ become increasingly alike cross-nationally (1999, p. 56).
Although contested, evidence for the first type of convergence has been
described in the field of education in Europe (Lawn and Lingard, 2002).
However, it is the second type of convergence in which less evidence
has been found. The convergence of outcomes of education civic values
is the focus for this chapter.
The chapter is based on the assumption that some convergence at the
first level of objectives on active citizenship across the 27 member states
of the EU had occurred between1999 and 2009. Despite mass education
and more specifically citizenship education having traditionally been
the prerogative of national governments mindful of the needs of the
nation state (Benavot and Truong, 2006), we argue that recent trends
suggest a move towards less nationalistic citizenship education agenda
in Europe that focuses more on rights (Keating, 2009). A multitude of
Bryony Hoskins and Rebecca Ridley 283
political events can be marshalled as evidence for this position. In 2004

ten former communist countries became members of the EU and as part
of this process the EU provided support to these countries through the
European Structural Funds to help economic and democratic develop-
ment. The rights of European Citizenship have been given to all citizens
across the EU just before 1999 and started to be experienced by citi-
zens through opportunities to study and work in other EU countries.
The charter of fundamental rights had been adopted in the Amsterdam
treaty in 1999 and put into European Union Primary law in the Lisbon
treaty 2009. In the field of education between 1999 and 2009, EU coun-
tries through the policy tool of the Open Method of Coordination, for
the first time adopted common objectives in the field of Education and
Training and more specifically on the topic of active citizenship (Edu-
cation Council, 2001). Later EU countries agreed a common definition
of civic competence as one of the eight key competencies required from
education and training for ‘personal fulfilment, active citizenship, social
cohesion and employability in a knowledge society’ (Education Council
and European Parliament, 2006). During the ten-year period a variety of
conferences, networks, forums and publications for the sharing of good
practice have been promoted by the Council of Europe and European
Commission (Hoskins, 2012a). In addition, indicators on active citizen-
ship and civic competence have been developed (Hoskins et al., 2006,
2008, 2012a). Finally, European countries predominantly had right-
wing governments that emphasised liberal values between 1999 and
2009 (Hoskins et al., 2012a).
From these policy initiatives we believe there is enough evidence to
suspect that general policy objectives on citizenship across the EU have
reduced in their differences. Nevertheless, despite all these initiatives,
it is necessary to acknowledge that differences between national poli-
cies remain and that these can be traced back to more liberal or civic
republican traditions, a greater sense of nationalism in more recently
formed countries and historical distinctions regarding more ethnic or
more civic conceptions of citizenship (Brubaker, 1992; Favell, 1998;
Baumann, 2004; Green and Janmaat, 2011; Hoskins, 2012a). However,
this chapter is based on the evidence that differences in terms of their
overall policy objectives have become less distinct during this specific
period. The focus for this investigation is then to establish the extent to
which the outcomes in terms of attitudes and values are also becoming
more similar.
Taking the premise that EU countries’ general objectives on Active
Citizenship have decreased in their differences, the aspect which this
chapter seeks to test is the extent to which we can see a convergence

of citizenship outcomes. The next question to address is then around
which set of values we would suspect that this convergence is taking
place. At a European level the values which are included with the defi-
nition of active citizenship and participatory citizenship have combined
liberal conceptions of citizenship (respect for democratic process, toler-
ance and volunteering), civic republican forms of citizenship (in terms
of the importance of participation and in particular within the politi-
cal processes) and critical and cosmopolitan citizenship values (values
of human rights and social justice) (Hoskins et al., 2012a). If the EU
is pulling countries towards a similar perspective one would also sus-
pect that the values in the policies and outcomes would also reflect this.
Thus we should see a common direction towards an increase in interest
in participation and a move towards the values associated with critical
and cosmopolitan citizenship.
Globalisation and cosmopolitan values
Theories of globalisation and world cultural theory support the hypoth-

esis that citizenship policy objectives are converging towards cosmopoli-
tan citizenship values (Ramirez, 2003; Ramirez et al., 2006, 2009;
Nuhoğlu Soysal and Suk-Ying Wong, 2007). These scholars have identi-
fied a gradual transition within national policies and curriculum towards
a more global understanding of citizenship that includes human rights
and multiculturalism (Ramirez et al., 2009). These theorists argue that
the discourse of a global humanity with individual human rights
and a recognition of diversity developed since the Second World War
(in which nationalism was considered to be one of the key contributors
to the conflict). They argue that this is a result of the influence of inter-
national organisations, international networks and conferences and a
country’s need to adjust and compete on the global stage. In particular
they focus on the role of global international organisations such as the
United Nations that validate and support these values (Ramirez et al.,
2009). World Culture theorists’ claim that the process of valorisation of
social justice values has been accompanied by the process of globalisa-
tion that enhances a cosmopolitan rather than nationalistic outlook, as
for instance demonstrated by (1) the effects of global warming (which
are conceptualised as a global rather than national policy issue), (2)
the development of international networks that enhance cooperation
and (3) the deeply interconnected nature of national economies. They
have used evidence from textbooks of various countries to identify the
rise of the discourse of human rights, international organisations and

wider global identities and the diminishing prevalence of nationalizing
narratives.
While cultural theorists have focused on convergence of policy they
have rarely discussed the effects or outcomes of these policies. However,
if education has any effect on values one might suspect that a greater
policy and education focus on cosmopolitan values may well lead to
higher levels of beliefs in human rights, tolerance and less nationalistic
concepts of citizenship. Thus one consequence of policy convergence
around cosmopolitan objectives could be that teaching from more
cosmopolitan texts books would facilitate a convergence towards the
adoption of the universal values of gender equality and tolerance and
towards a reduced emphasis on national conceptions of citizenship
and national forms of political participation. Wiseman et al.’s (2011)
study on political socialisation built on theories of global convergence
and attempted to demonstrate that through political socialisation pro-
cesses in school that young people’s civic attitudes and knowledge and
skills were becoming similar. However, Wiseman et al.’s study used the
IEA CIVED cross-sectional survey data for one point in time (1999) and it
is necessary to use time series data in order to test convergence. In this
chapter we compare the IEA 2009 International Civic and Citizenship
education Study data (ICCS) with the original CIVED study to identify
convergence.
To explore in more detail the literature concerning trends of civic out-
comes we need to look towards political science. Political scientists using
the World Values Survey have demonstrated that the overall trends in
values over the last 30 years indicate a moving away from conserva-
tive traditional values and nationally orientated perspectives towards a
post-materialistic set of values that favours the values of human rights
and gender equality. These post-materialistic values are said to have the
dual effect of enhancing qualities that include individualistic and ad
hoc protest activities that ‘challenge the elite’, while at the same time
increasing critical thinking regarding national authorities that may well
lower forms of engagement that are supportive towards the elite such as
voting in elections or party membership. These post-materialistic out-
comes are quite similar to the values of cosmopolitanism and there is
coherence between the input theories of world culture and the outcome
orientated trends regarding modernisation. However, the distinction
is that modernisation theorists (Inglehart and Welzel, 2005) cite eco-
nomic modernisation as the key contributing factor to value change as
opposed to policy convergence. They argue that as countries become
wealthier the attitudes of that nation will become less traditional and
more post-materialistic. The implication is that these theorists do not
necessarily see countries becoming more similar with respect to the
adoption of cosmopolitan (or post-materialistic) values. Thus, they agree
with the World Culture advocates on the direction of the value change
but disagree on the issue of convergence. They argue that as countries
become wealthier the attitudes of that nation will become less tradi-
tional and more post-materialistic. Nevertheless it may well be that
economic development and globalisation-driven policy convergence are
two sides of the same coin and that the main factor enhancing both pol-
icy convergence towards cosmopolitanism and value change is universal
economic growth up until the economic crisis of 2008.
Drawing on the convergence of European policy objectives and World
Culture and Modernization theories, the hypotheses that this chapter
will test are fourfold; First, we suspect that values of young people across
the EU are following a similar trajectory and are converging; second, the
direction of this convergence is towards cosmopolitan values; third, par-
ticipation levels will be increasing along with the greater emphasis on
participation at the supranational level; finally, the more global forms of
participation are more likely to be on the increase while more national
and conventional forms of political participation such as voting and
party membership may well increase less or decrease during this ten-year
period.
Data source, country selection, outcomes of interest

and methods of analysis
Data source
The data analysis consists of analysing trends between two sets of
data collected by the IEA, CIVED 1999 and International Civic and
Citizenship education Study (ICCS) 2009. Both studies aimed towards
understanding ‘how young people are prepared to undertake their role
as citizens’ (Torney-Purta et al., 2001). ICCS is the most recent IEA study
on civic and citizenship education among school pupils and was con-
ducted in Europe in 2009. Data was collected from over 140,000 Grade
8 students, 62,000 teachers and 5,300 school principals, from 38 coun-
tries. The ICCS student population comprised students in Grade 8
(pupils approximately 14 years of age, although some are above and
below this age). The second set of data that we use CIVED was also
administered by the IEA in 1999 in 28 countries with approximately
90,000 students participating. For both studies one classroom in the
target grade per school was selected in most countries, and IEA recom-
mended a sample of around 150 schools per country although there
were some exceptions.
Both studies contained a number of instruments including a student
questionnaire which had some common scales. The student question-
naire consisted of items measuring student background variables and
students’ attitudes and behaviours. The affective-behavioural aspects
included questions on value beliefs, attitudes, behavioural intentions and
behaviour. In most cases the response categories were a set of Likert-type
items with four categories (e.g., ‘strongly agree,’ ‘agree,’ ‘disagree,’ and
‘strongly disagree’). In the IEA CIVED study the additional category of
‘don’t know’ was also available.
It must be acknowledged that there were some differences between
the two studies. Firstly, there were some differences between the theo-
retical and assessment frameworks. In CIVED, for example, ‘democracy’
was included under the domain of Democracy and Citizenship, but in
ICCS it was referred to as a ‘key concept’ under the domain of Civic Soci-
ety and Systems. Likewise, concepts relating to civic participation were
captured in CIVED under the domain of Democracy and Citizenship
and under the sub-domain of ‘rights and duties of citizenship’. However,
in ICCS those concepts were included under a new fourth domain: Civic
Participation. The emphasis between the two studies therefore shifted.
In ICCS a broader number of questions were tapping a range of differ-
ent civic institutions and much more importance was afforded to civic
participation, of which became a domain in itself (Torney-Purta and
Barber, 2012). Though somewhat different, the theoretical and assess-
ment frameworks did not impinge on the methodology and findings for
this analysis.
Secondly, there were differences between the studies in terms of how
specific items and question were worded. This was particularly impor-
tant for this analysis when it came rescaling the attitudinal scales
(Torney-Purta and Barber, 2012). If the items were very different then
they were removed from each respective scale to strengthen the com-
parability of the dimensions across the two studies. The studies also
differed in terms of the instructions heading each section. In the
sections that explored young peoples’ understandings of conventional
citizenship and social movement–related citizenship, Torney-Purta and
Barber (2012) note for example that ‘the CIVED study began with the
stem “A good citizen . . . ,” with the implication that the items that fol-
lowed completed the sentence (e.g., “A good citizen obeys the law”)’.
In the ICCS study this was changed to ‘How important are the following
behaviours for a good citizen?’ (p.54). Despite this most of the word-
ing of the items across the two surveys was identical and therefore was
worthy of a comparison over time.
Country selection
In this chapter we have selected all EU countries that participated
in both the ICCS 2009 and the 1999 CIVED study which enables us
to make a comparison across time. Thus in total we will examine
16 European Countries: Bulgaria, Cyprus, Czech Republic, Denmark,
England, Estonia, Finland, Greece, Italy, Latvia, Lithuania, Poland,
Slovak Republic, Slovenia and Sweden.
Outcomes of interest
The concept of cosmopolitanism is about moving beyond national cit-
izenship to a global concept of humanity which included recognition
of human rights and the valorisation of diversity (Held, 2010). In terms
of education two curricular changes are said to demonstrate the shift
towards cosmopolitan teaching. First, a focus on global issues and inter-
national organisations and second a greater focus on sub-groups of the
population such as women, ethnic groups and migrants (Ramirez et al.,
2009). The values which are associated with the concept of cosmopoli-
tanism we would suggest to be the values of human rights and tolerance
and, in addition, a much less strong concept of national and con-
ventional citizenship. To measure the concept of human rights and
increasing tolerance of sub-groups we adopt the indicators of gender
equality and tolerance of immigrants. To measure concepts of national
citizenship and participation we adopt measures of conventional citi-
zenship values and social movement related citizenship values. The first
of which we expect to be becoming weaker if cosmopolitanism is on
the rise and the second of which we may decide to be less effected
by the rise in cosmopolitanism as it is less orientated towards national
governments.
We tapped the value of ethnic tolerance with a scale denoting positive
attitudes towards immigrants. This scale sought to measure young peo-
ple’s attitudes towards immigrants and their families Shulz and Sibberns,
2005; Shulz et al., 2011 and included five items relating to the rights and
opportunities: ‘immigrants should have the opportunity to keep their
own language’, ‘Immigrant children should have the same opportuni-
ties for education that other children in the country have’, ‘Immigrants
who live in a country for several years should have the opportunity to
vote in elections’, ‘Immigrants should have the opportunity to continue
their own customs and lifestyles’, and ‘Immigrants should have the
same rights that everyone else in a country has’. The items were identi-
cal for both ICCS and CIVED and therefore no items required removal.
The pooled Cronbach alpha for this scale based on the common set of
countries used in this analysis for CIVED 1999 was 0.82 and 0.79 for
ICCS 2009.
The human rights dimension was tapped with a scale on gender equal-
ity (GENEQU). This scale sought to measure young people’s attitudes
towards gender equality and included five items: ‘Men and women
should have the same rights in every way’, ‘Women should stay out of
politics’, ‘When there are not many jobs [jobs are scarce (CIVED, 1999)]
available, men should have more right to a job than women’, ‘Men and
women should get equal pay when they are doing the same jobs’, Men
are better qualified to be political leaders than women’. For this dimen-
sion the items ‘women should run for public office and take part in the
government just as men do’ (CIVED, 1999) and ‘men and women should
have equal opportunities to take part in government’ (ICCS, 2009) were
removed from each of the respective scales in order to assure compara-
bility between the two data points. The pooled Cronbach alpha for this
scale, based on the common set of countries in this analysis, was 0.74
for CIVED 1999 and 0.78 for ICCS 2009.
The concept of citizenship is split into two dimensions. The first
dimension is conventional citizenship. This scale set out to measure
young people’s attitudes of what constitutes a good adult citizen (Shulz
and Sibberns, 2005; Shulz et al., 2011) and included six items in response
to ‘a good citizen . . . ’ (CIVED, 1999) and ‘how important are the follow-
ing behaviours for a good citizen . . . (ICCS, 2009). These were: ‘Votes in
every [national1 ] election’, ‘Joins a political party’, ‘Knows [learns] about
the country’s history’, follows political issues on in the newspaper, on
the radio, or on TV [or the internet]’, ‘Shows respect for government rep-
resentatives’, ‘Engages in political discussion’. No items were removed
from each respective scales as there had been no changes to the word-
ing and the pooled Cronbach alpha for this scale based on the common
set of countries used in this survey for CIVED 1999 was 0.701 and 0.714
for ICCS 2009.
The second dimension of citizenship pertains to social movement–
related citizenship. Analogous to conventional citizenship the items on
social movement citizenship were in response to the same question (i.e.
a good citizen . . .’ (CIVED, 1999) and ‘how important are the follow-
ing behaviours for a good citizen . . . (ICCS, 2009). The items included:
‘Would participate in a peaceful protest against a law believed to be
unjust’, ‘Participates in activities to benefit people in the community’,

‘Takes part in activities promoting human rights’, ‘Takes part in activ-
ities to protect the environment’. Again, no items required removal as
they were identical for both the studies. The pooled Cronbach alpha for
this scale for CIVED 1999 was 0.64 and 0.74 for ICCS 2009.
All the scales created were summative indexes with values rang-
ing between 0 and 10, and the value 5 as mid-point. Values below 5
reflect disagreement or less likelihood to varying degrees (i.e. negative
responses). Values above 5 represent different degrees of agreement or
more likelihood (i.e. positive responses). These summative indexes have
the advantage of not only allowing for easy cross-national comparisons,
but also of assessing absolute scores. Absolute levels cannot be examined
with scales reflecting factor scores or other standardised values.
Method
In order to compare the citizenship scales across time it was necessary
to make some modifications to both the CIVED and ICCS data. The first
step involved the reversal of the values on the ICCS survey items so that
they matched with the CIVED items. As Torney-Purta and Barber (2012)
noted, ‘in CIVED a response of 1 indicated the lowest possible endorse-
ment of an item, while 4 indicated the highest . . . [i]n ICCS the reverse
was the case’ (p. 57). This step also involved re-coding the CIVED ‘don’t
know’ option as ‘missing’ for the relevant variables. This was necessary
because in ICCS a ‘don’t know’ option was not included on any of the
Likert scales.
The last step involved merging the relevant data from CIVED and
ICCS into one dataset. The data was merged via the variable ‘coun-
try numerical code’; it was therefore necessary to recode some of the
countries prior to merging so as to ensure that they would match-up.
To measure convergence we calculated the overall mean from the 16
countries for each of the scales for both years. We then calculated each
country’s departure from this mean by subtracting this mean from each
country mean and squaring the resulting difference. We then summed
all these differences for each of the two years to establish the variance for
each year for each of the scales. If the variance has reduced from 1999
to 2009 then the particular attitudes were considered to have converged
across Europe. The direction of the change was decided by comparing
the overall means of the two years. To test whether the changes in
the overall means and the variances were significant we conducted an
independent samples t-test.
Results
Positive attitudes towards immigrants

In all 16 countries in both 1999 and 2009 young people are more posi-
tive than negative towards rights for immigrants in absolute terms (see
Table 13.1). The results show that there is a convergence during the
ten-year period on attitudes towards immigrants across these European
countries, but the t-test did not show this process to be significant
(see Table 13.2). The two countries which created most of the varia-
tion in 1999 by having considerably more positive attitudes towards
immigrants were Sweden and Cyprus (see Table 13.3). These two coun-
tries became more similar to the rest of the group through a decline in
their young people’s positive attitudes and at the same time an increase
in more positive attitudes from the remaining countries. Overall, the
direction for the convergence is towards a more positive attitude of
young people with just over half of the countries becoming increas-
ingly supportive towards this group. The difference, however, between
the two overall means over the ten-year period was not significant (see
Table 13.2).
Table 13.1 Mean scores for the four scales in 1999 and 2009 by country.
Mean score Mean score Mean score Mean score Social

Immigrants Gender Conventional Movement
1999 2009 1999 2009 1999 2009 1999 2009
BGR 6,44 7.50 6.72 6.80 6.20 5.04 6.95 7.67

CYP 7.79 6,96 7.90 7.16 7.42 6.06 7.92 7.33
CZE 7.00 6.83 7,90 7,37 5.67 4.46 6.68 6.49
DNK 6.29 6.90 8,44 8,53 5.34 5.27 6.52 5.62
ENG 6.58 6.35 8,48 8,33 5.51 5.49 6.26 6.41
EST 6.69 6.72 7.23 7.55 5.44 4.87 6.26 6.51
FIN 6.74 6.76 8.23 8.33 5.39 4.78 5.75 6.08
GRC 7.50 7.32 7.63 7.59 7.27 5.29 8.29 7.52
ITA 6.75 6.87 7.85 8.07 6.43 6.25 7.25 7.35
LTU 6.49 7.43 7.35 7.27 6.76 5.54 7.54 6.74
LVA 6.56 6.31 6.74 6.96 6.01 5.71 6.52 6.82
NOR 7.16 7.20 8.46 8.37 5.54 6.02 7.32 7.23
POL 7.36 7.24 7.64 7.25 7.13 5.51 7.17 6.74
SVK 6.69 7.19 7.37 7.36 6.45 4.70 7.35 6.57
SVN 6.29 7.20 7.64 8.03 5.77 4.96 6.52 6.59
SWE 7.78 7.46 8.12 8.61 5.76 5.06 6.82 6.48
Table 13.2 Overall mean score and variation for the 16 European countries for
1999 and 2009 with p-values given from T-tests on mean differences.
Overall Overall Difference Overall Overall Difference

mean mean (p-value) variation variation (p-value)
1999 2009 1999 2009
Immigrants 6.88 7.02 .391 3.67 2.02 0.82

Gender 7.73 7.72 .958 4.60 5.29 0.99
Conventional 6.13 5.31 .001 7.47 4.06 0.34
Social 6.95 6.76 .397 6.55 4.47 0.83
movement
Table 13.3 The 16 European countries contributions to the overall variation for
the four scales.
Variation Variation Variation Variation Social

Immigrants Gender Conventional Movement
1999 2009 1999 2009 1999 2009 1999 2009
BGR 0.19 0.24 1.01 0.85 0.00 0.07 0.25 0.83

CYP 0.82 0.00 0.03 0.32 1.67 0.55 0.97 0.32
CZE 0.01 0.03 0.03 0.13 0.21 0.73 0.07 0.07
DNK 0.35 0.01 0.50 0.65 0.62 0.00 0.18 1.29
ENG 0.09 0.45 0.56 0.36 0.38 0.03 0.47 0.13
EST 0.03 0.09 0.24 0.03 0.48 0.20 0.46 0.06
FIN 0.02 0.06 0.25 0.37 0.56 0.28 1.40 0.46
GRC 0.38 0.09 0.01 0.02 1.30 0.00 1.80 0.58
ITA 0.02 0.02 0.01 0.12 0.09 0.87 0.09 0.35
LTU 0.15 0.17 0.15 0.20 0.40 0.05 0.35 0.00
LVA 0.10 0.50 0.99 0.57 0.01 0.16 0.18 0.00
NOR 0.07 0.03 0.52 0.41 0.35 0.50 0.14 0.22
POL 0.23 0.05 0.01 0.23 1.00 0.04 0.05 0.00
SVK 0.04 0.03 0.13 0.13 0.10 0.37 0.16 0.04
SVN 0.35 0.04 0.01 0.09 0.13 0.12 0.19 0.03
SWE 0.81 0.20 0.15 0.80 0.14 0.06 0.06 0.08
Gender equality
For all the countries studied in Europe young people’s attitudes are
in absolute terms more positive than negative towards gender equal-
ity at both points of time (see Table 13.1). In contrast to the results
on attitudes towards immigrants, the results of young people’s atti-
tudes towards gender equality show that the total variance has slightly
increased across Europe showing a small degree of divergence (see
Table 13.2). The 16 countries’ mean remained stable across the ten-
year period with no significant changes. Bulgaria and Latvia contributed
most to the variation in 1999 with their lower levels of positive atti-
tudes towards gender equality (see Table 13.3). In contrast Sweden and
Denmark contributed considerably to the variation in 2009 with very
high levels of positive attitudes towards gender equality while Bulgaria
also contributed with a much lower levels of support for gender equal-
ity. Neither the divergence nor the (minimal) change in absolute levels
turned out to be significant.
Conventional citizenship
In 1999 all European countries selected were in absolute terms more pos-
itive than negative towards conventional forms of participation being
considered manifestations of good citizenship (see Table 13.1). How-
ever, in 2009 five countries (Czech Republic, Estonia, Finland, Slovakia
and Slovenia) became more negative than positive towards norms
such as voting in national elections. According to the results of our
analysis there has been a (non-significant) convergence of values (see
Table 13.2). This convergence has been in the direction towards less
conventional participatory norms. In contrast to the convergence this
drop in support for conventional participation is significant. Cyprus,
Greece and Poland contributed most to the variation in 1999 with
their very positive attitudes towards conventional forms of participa-
tion (see Table 13.3). Levels of positive attitudes in these three countries
significantly decreased during the ten-year period.
Social movement–related citizenship

All countries selected for this study remained more positive than neg-
ative across the ten-year period in their attitudes towards social move-
ment norms of good citizenship that include protesting and volunteer-
ing (see Table 13.1). The results of our analysis also show that there has
been convergence regarding these more global and cosmopolitan forms
of citizenship norms (see Table 13.2). However, the direction of this
convergence is also in a downward direction towards less participatory
norms. Neither the process of drawing closer together nor the down-
ward trend is significant. There were two main outliers in 1999 that
created most of the variation (see Table 13.3). In 1999 Greece con-
tributed most to the variation with considerably higher norms than the
other European countries while Finland contributed the second most
with considerably lower participatory norms regarding social move-
ment related citizenship. In 2009 it was Denmark which contributed
the most to the variation with the least positive attitudes towards social
movement participatory norms.
Discussion
Our first hypothesis, that attitudes across European youngsters are

becoming similar, receives some support as converge emerged on three
of the four attitudes while divergence was only observed for gender
equality. It must be noted, however, that these trends were not sig-
nificant. The evidence is more mixed in relationship to our second
hypothesis, which tests world cultural theory and states that the direc-
tion of convergence should be towards cosmopolitan values. The atti-
tudes that we have selected regarding values on human rights (attitudes
towards gender equality) and diversity (attitudes towards immigrants)
give conflicting results, with attitudes regarding gender equality show-
ing divergence and those regarding immigrants showing convergence.
Moreover, while the combined direction of change for attitudes towards
immigrants followed the pattern proposed by World Culture theory
towards more positive attitudes (the change not being significant, how-
ever), for gender equality a change in the same direction could not
be observed. This evidence alone does not provide convincing sup-
port to fully accept this hypothesis. The significant overall downward
trend in support for conventional national forms of civic participation
is consistent with World Culture’s proposition that participation should
become less nationalistic and have a greater orientation towards global
issues of human rights and the environment and the activities are more
likely to take the forms of protest. However, the slight downward trend
in support for social movement citizenship, which measures precisely
these aspects of participation, is certainly not in accordance with this
proposition.
The third hypothesis that European policy is pulling young people
towards greater norms of participation is not supported by the evi-
dence as there is a clear trend towards less participatory norms for
both social movement and (especially) conventional forms of partici-
pation. Nevertheless it should be taken into account that in absolute
terms young people are still more positive than negative to both forms
of participation in the majority of the 16 European countries studied.
In addition, we should take into account that this data is cross-sectional
and that it is therefore not possible to evaluate precisely the effects of
Europeanisation of objectives on citizenship education as we cannot
control for individual students prior values. In fact, is it possible that
what we are seeing is a generational difference and that actually the

downward trends on civic attitudes could have been considerably more
extreme without any policy intervention on participatory attitudes.
These results are puzzling and deserve some further consideration.
We could speculate that the results may be due to the fact that European
policy on active citizenship is part of an agenda on lifelong learning and
that young person of averaging about 14 years old have yet to experience
a great deal of the measures and programmes that have been devel-
oped. A second reason could be that there may be other larger global
incidents and trends occurring that provide a more powerful influence
on young people’s attitudes and values than European soft policies on
participatory citizenship. The main global political event that was first
felt in 2008 was the global economic crisis and this was associated with
a significant reduction in the levels of adult trust in institutions in
European countries in particular in those countries most affected by the
crisis (Hoskins and Kerr, 2012b). In addition, a more long-term inter-
generational decline in trust in political institutions has been identified
as has the decline in social relationship within and between communi-
ties (Putnam, 2001). It is also the case that during the ten-year period
between 1999 and 2009 right of centre parties in Europe countries were
elected leading to wide spread implementation of market-orientated
policies (which was also part of global trend towards the implemen-
tation of policies on competitiveness) at the expense of the European
social model (Hoskins et al., 2012b). These policies have further empha-
sised individualisation and personal responsibility leaving young people
the challenge of having to create and choose their own careers, identities
and pathways in life (Giddens, 1991; Beck, 2000). The focus on the indi-
vidual is unlikely to enhance a sense of common duty that could build
participatory values. However, with the data available it is difficult to
test which global events and trends, if any, have influenced citizenship
attitudes and behaviour.
Conclusions
Drawing strict conclusions from this analysis is not possible due to the
limitations of the data. We have used cross-sectional data and thus we
have not been able to compare an individuals’ change in attitudes and
subsequently there is little certainty concerning regarding the relation-
ship between the implementation of a policy and a change in attitudes.
The difficulty is that there is very limited comparative longitudinal data
available on citizenship with which to test these ideas.
However, what we would propose is that global incidents and trends

including the economic crisis appear from this study to have some asso-
ciation with convergence and trends on lower levels of engagement
in particular on a national level for young people aged about 14 in
Europe. We find the evidence regarding supporting World Culture the-
ory to be limited. As they claim, there are overall reductions in national
forms of participation but at the same time we also find reductions on
social movement participation (if smaller). We would argue that this is
hardly a basis for suggesting that European Union youngsters are becom-
ing more cosmopolitan when considering the additional evidence that
young people have diverging attitudes on gender equality, even if there
is convergence in a positive but not significant direction on positive
attitudes towards immigrants.
An alternative hypothesis could be that both the limitations of
national and international government interventions have been high-
lighted by recent global events. In the case of the economic crisis it has
been global international markets that have failed to be tamed by either
national or international organisations such as the EU. Drawing from
this we could speculate that the individuals may well question both the
trustworthiness and influence of governance at any level and this could
form part of the reason why young people are wishing to participate
less. Therefore we could suggest that global events and trends have a
greater influence on civic attitudes than relatively small national and
international citizenship programmes with modest budgets (compared
to market orientated spheres such as trade and competition). Viewed
from this perspective the expectation that young people in Europe
would have become more cosmopolitan and participatory between 1999
and 2010 appears to be rather more unlikely than World Culture theory
would suggest.
Note
1. For this scale the word in brackets was added to the item in the ICCS 2009
study.
References
Baumann, G. (2004) ‘Introduction: Nation-State, School and Civil Enculturation’
in Schiffauer, W., G. Baumann, R. Kastoryano, and S. Vertovec (eds.) Civil
Enculturation: Nation-State, School and Ethnic Difference in the Netherlands, Britain,
Germany and France. Oxford: Berghahn, pp. 1–18.
Beck, U. (2000) ‘Risk Society Revisited: Theory, Politics and Research Pro-
grammes’. The Risk Society and Beyond: Critical Issues for Social Theory,
pp. 211–229.
Benavot, Aaron, and Nhung Truong. (2006) ‘Introduction’ pp. 1–14 in Aaron
Benavot and Cecilia Braslavsky (eds.), School Knowledge in Comparative and
Historical Perspective: Changing Curricula in Primary and Secondary Education.
Amsterdam: Springer.
Brubaker, R. (1992). Citizenship and Nationhood in France and Germany. Cambridge,
MA: Harvard UP.
Council of Europe (2010). Charter on Education for Democratic Citizenship and
Human Rights Education: Recommendation CM/Rec(2010)7 and explanatory
memorandum. Strasbourg: Council of Europe.
Education Council (2006) Recommendation of the European Parliament and the
Council of 18 December 2006 on Key Competencies for Lifelong Learning. Brussels:
Official Journal of the European Union, 30.12.2006.
Education Council (2001) Report from the Education Council to the European Coun-
cil. The Concrete Future Objectives of Education and Training Systems. Brussels,
14 February 2001. Brussels: European Council.
Favell, A. (1998) Philosophies of Integration: Immigration and the Idea of Citizenship
in France and Britain. Basingstoke: Palgrave Macmillan.
Giddens, A. (1991) Modernity and Self-identity: Self and Society in the Late Modern
Age. Stanford, CA: Stanford University Press.
Green and Janmaat (2011) Regimes of Social Cohesion: Societies and the Crisis of
Globalization. Houndmills, Basingstoke: Palgrave Macmillan.
Green, A. (1999) ‘Education and Globalization in Europe and East Asia: Conver-
gent and Divergent Trends’ Journal of Education Policy, 14: 1, pp. 55–71.
Held, D. (2010) Cosmopolitanism: Ideals and Realities. Cambridge: Polity press.
Hoskins, B., Villalba, C. and Saisana, M. (2012a) The 2011 Civic Competence Com-
posite Indicator (CCCI-2): Measuring Young People’s Civic Competence across Europe
based on the IEA International Citizenship and Civic Education Study. Ispra, Italy,
European Commission (doi: 10.2788/67938).
Hoskins, B. and Kerr, D. (2012b) Final Study Summary and Policy Recommenda-
tions report: Participatory Citizenship in the European Union, Institute of Education
report for EU. Southampton: Southampton University, ISBN 9780854329403.
Hoskins, B. and Mascherini, M. (2009) Measuring Active Citizenship through
the Development of a Composite Indicator. Social Indicator Research, 90,
pp. 459–488. http://dx.doi.org/10.1007/s11205-008-9271-2http://dx.doi.org/
10.1007/s11205-008-9271-2.
Hoskins, B., Villalba, E., Van Nijlen, D., and Barber, C. (2008) Measuring Civic Com-
petence in Europe: A composite Indicator based on IEA Civic Education Study 1999
for 14 years old in School. CRELL research paper, EUR 23210, Ispra: European
Commission.
Hoskins, B. et al. (2006) Measuring active Citizenship in Europe. CRELL research
paper 4, EUR 22530en, European Commission: Italy.
Inglehart, Ronald F. and Christian Welzel (2005) Modernization, Cultural Change,
and Democracy: The Human Development Sequence. New York: Cambridge Uni-
versity Press.
Keating, A. (2009) ‘Educating Europe’s Citizens: Moving from National to Post-
national Models of Educating for European Citizenship’ Citizenship Studies, 13:
2, pp. 135–15.
Lawn, M. and Lingard, B. (2002) ‘Constructing a European Policy Space in
educational Governance: The Role of Transnational Policy Actors’ European
Educational Research Journal. 1: 2, pp. 290–307.
Novoa, A. and Lawn, M. (2002) Fabricating Europe: The Formation of an Education

Space. New York: Kluwer Academic Publishers.
Nuhoğlu Soysal, Y. and Wong, S. Y. (2007). ‘Educating Future Citizens in
Europe and Asia’ in Benavot, A., and Braslavsky C. (eds) School Knowledge in
Comparative and Historical Perspective, Dordrecht: Springer 73–88.
Putnam, R. D. (2001). Bowling alone: The Collapse and Revival of American
community. New York: Simon & Schuster.
Ramirez, F. (2003) ‘The Global model and National Legacies’ in Anderson-Levitt,
K. (ed.) Local Meanings, Global Schooling: Anthropology and World Culture Theory.
New York: Palgrave Macmillan.
Ramirez, F., Bromley, P. and Russell, S. (2009). ‘The Valorization of Humanity and
Diversity’ Multicultural Education Review, 1, 1, pp. 29–54.
Ramirez, F., Suarez, D. and Meyer, J. (2006) ‘The Worldwide Rise of Human Rights
Education’ in Benavot, A., and Braslavsky, C. (eds) School, Knowledge in Com-
parative and Historical Perspective. Hong Kong. China: Comparative Education
Research Centre, pp. 35–52.
Shulz, W., Ainely, J. and Fraillon, J. (2011) ICCS 2009 Technical Report. IEA.
Amsterdam, The Netherlands.
Shulz, W. and Sibberns, H. (2005). IEA Civic Education Study Technical Report. IEA.
Amsterdam, The Netherlands.
Torney-Purta, J. and Barber, C. (2012) ‘Comparing Attitudes in the 1999 and 2009
IEA CIVC and Citizenship Education Studies: Opportunities and Limitations
Illustrated in Five Countries’. Journal of Social Science Education, pp. 47–63.
Torney-Purta, J., Lehmann, R., Oswald, H. and Schulz, W. (2001) Citizenship
and Education in Twenty-eight Countries: Civic Knowledge and Engagement at Age
Fourteen. Amsterdam: IEA.
Wiseman, A.W., Astiz, M.F., Fabrega, R. and Baker, D.P. (2011) ‘Making Citizens
of the World: The Political Socialization of Youth in Formal Mass Education
Systems’ Compare, 41, 5, pp. 561–77.
Conclusion
All the contributions in the second part of this book comparatively

analysed national education systems to explore the links between edu-
cation and social cohesion. These contributions proceeded from the
assumption that such systems show sufficient variation to make a com-
parative analysis worthwhile. In view of the influential idea that such an
analysis is increasingly irrelevant due to education systems becoming
increasingly alike, we thought it necessary to address this assumption
before examining the relations between education and social cohesion.
Hence the first part of the book includes chapters exploring differ-
ences between education systems, the direction of changes in system
characteristics and patterns of convergence and divergence. However,
the chapters in the first part were not only written in function of
the second part. They addressed distinct omissions in the literature
on convergence and divergence by examining trends in a wide range
of indicators, by engaging in a series of in-depth case studies and by
assessing cross-national variation in hitherto poorly researched areas.
Part I: Enduring differences between educational systems
The key finding of the contributions to Part I was that national edu-
cation systems continue to show conspicuous differences despite some
evidence of convergence or movement in the same direction. Broadly,
they found little empirical support for the aforementioned perspective
of increasing similarity. As explained in the second and sixth chapter,
this perspective did not claim random convergence but convergence
towards specific outcomes. In a nutshell, it proposed that countries
increasingly adopt decentralised structures, school choice policies and
central modes of assessment in response to an increasingly powerful
299
300 Conclusion
transnational discourse of (neo-) liberalism advocating quasi-markets,

new public management techniques, detached central rule and educa-
tional expansion in order to attain high stocks of human capital (e.g.
Astiz et al., 2002; Baker and LeTendre, 2005; Krucken and Meier, 2006;
Ball, 2012). This same discourse is said to have led to a cross-national
convergence towards a set of policies and curricula emphasising individ-
ualism, human rights, democracy and active citizenship (Hafner-Burton
and Tsutsui, 2005; Suarez, 2007).
The empirical analyses of Chapter 1 partly endorsed this perspec-
tive as it found more instances of convergence (17) than divergence
(6) in its assessment of 24 indicators of system properties and out-
comes over a period of two decades. Yet, the convergence found was
often small as it was statistically insignificant in 14 of the 24 indica-
tors. Moreover, it happened primarily on outcome measures, such as
average scores in PISA tests, the distribution of these scores and propor-
tions graduating from upper secondary and higher education. Although
the convergence perspective includes expectations on outcomes (con-
vergence towards ever higher graduation rates of higher education, for
instance), it in first instance relates to policies on the organization,
structure and curriculum content of education systems. On the latter
at least as many instances of continuing differences or divergence were
found (e.g. on student–teacher ratios, the share of private spending on
tertiary education, and selection to schools by ability). What is more,
the convergence that was found on a structural property in some cases
did not go in the expected direction. For instance, although countries
appeared to become more similar on decision-making structure, the
trend was towards more central control instead of more school auton-
omy. In similar vein, Chapter 13 showed that despite active citizenship
programmes supported by the European Commission and the Council
of Europe, the first decade of the new millennium has not seen any
convergence towards more civic outcomes among adolescents from 16
European countries.
In addition to these two chapters examining trends, two other
chapters have analysed the differences between countries in models
of post-compulsory education (Chapter 2) and in the structure and
curriculum of citizenship education (Chapter 5). As these chapters anal-
ysed cross-sectional data, they could not address the dynamic aspect of
the convergence perspective. Yet the pronounced variation they found
across countries is in itself difficult to reconcile with the homogeneity of
structural characteristics and outcomes anticipated by this perspective.
Thus, the chapter on post-compulsory education finds wide variation
in rates of enrolment to upper secondary and higher education across

26 European countries. This suggests that the educational expansion
predicted by the convergence perspective has not happened to the
extent that laggards have been able to catch up with frontrunners.
In addition the chapter finds considerable cross-national differences
regarding enrolment in vocational tracks in upper secondary educa-
tion, the percentage of students taking apprenticeships and enrolment
in adult vocational training courses, indicating that systems of post-
compulsory education do not only differ in degree but also in kind. The
chapter on citizenship education similarly finds considerable variation
across a selection of Western and East-Asian states on the dimensions of
state control and collective values. While English-speaking states have
low scores on both dimensions (reflecting decentralised structures and
an emphasis on moral autonomy, individualism and procedural values),
East Asian states show the opposite pattern with high degrees of cen-
tralised control and a strong emphasis on national homogeneity and
substantive values. The high absolute values of the latter are not antici-
pated by the convergence perspective as it expects convergence towards
school autonomy and curricula fostering cosmopolitanism and human
rights.
Finally, the two chapters in the first part that have carried out in-depth
case studies of a number of European countries have demonstrated
pronounced variation in education systems reflecting choices between
different political and moral principles (cf Chapters 3 and 4). They, for
instance, find that in Sweden and Denmark a clear choice has been made
for the principles for universalism and inclusion judging by the low
rates of non-completion in upper secondary, the high uptake of adult
education and the efforts to compensate for initial inequalities through
universal provision of pre-primary education and comprehensive edu-
cation in lower secondary. In France and Italy the principle of merit
based on academic achievement reigns supreme as indicated by strict
systems of assessment and selection (including grade repeating), diplo-
mas reflecting different standards, relatively high non-completion rates
and a large difference in status between academic and vocational tracks.
The education system in Germany is found to be inspired by the notion
of a calling, as demonstrated by early selection in lower secondary edu-
cation (in preparation for a specific profession), the prominent place and
relatively high social esteem of vocational education and the institution-
alisation of on the job training through apprenticeships. The United
Kingdom, finally, represents a country where the logic of the market
has influenced educational policies, resulting in a system characterised
302 Conclusion
by school choice, a diverse offer of schools and courses, pronounced

inequalities of skills and qualifications, and output-based central assess-
ments (the results of which are made public to enable individuals to
make an informed choice for a particular school or education). The case
studies thus reveal that the education systems of West-European coun-
tries are inspired by very different political ideals and considerations,
which contradicts the claim of convergence theorists that national pol-
icymakers are increasingly in agreement on the main objectives of
education systems due to the growing influence of a global discourse
of liberal democracy and the transnational agencies (OECD, European
Union, Council of Europe) advocating it.
Together, all these findings suggest that wholesale convergence has
not taken place and that countries remain distinctive on key properties
of their education systems. They thus underline the continuing rele-
vance of cross-national comparative research as a means to investigate
the social consequences of education systems.
Why do our findings provide such limited support for the conver-
gence perspective? Aside from the reasons offered by the alternative
perspectives (i.e. those of parallel development, regimes, and cultural
traditions; see the introductory chapter), the above discussion of the
two case-study chapters suggests that there are in fact competing dis-
courses at both the national and transnational level pushing countries
in different directions. Indeed, Carney et al. (2012) have argued that
the assumption of widespread consensus characterising the thinking
of convergence advocates (or so-called World Culture theorists) has
prevented these scholars from seeing and acknowledging examples of
disagreement, power struggles and coercion. Even at the transnational
level consensus is anything but a given. OECD (2010, p. 14) research,
for instance, has observed that “countries creating a more compet-
itive environment in which many schools compete for students do
not systematically produce better results”, which directly contradicts
the assumption of the neoliberal discourse that competition enhances
overall achievement. Other research has similarly found that the top
performance of Finland in the PISA league table has nothing to do with
the policy prescriptions of neoliberal theory (Sahlberg, 2011; (Teach-
ers’ College Columbia University: New York). Finland, for instance, has
no inspection system, little school evaluation, limited testing of stu-
dents, no school league tables, few private schools, no selection and
little streaming and setting.
Nonetheless, two qualifications need to be made regarding our anal-
yses of patterns and trends in system properties. Firstly, although
convergence theorists typically view the world from a very high van-
tage point and accordingly make statements about policy convergence
that stretch across centuries, include all countries of the world and apply
as a rule to key legislative acts only, our analyses are based on data from
a much more limited set of countries (Europe and OECD) which cover
a much smaller period (the 1990s and 2000s) and concern many more
features of education systems than policies alone. In other words, while
width characterises the research of convergence theorists, our analyses
go into depth. Our approach could therefore be criticised for not adopt-
ing the historical and geographical scope that is needed for a proper
test of the convergence perspective. We acknowledge this drawback, but
maintain at the same time that our in-depth approach has been pro-
ductive as it has revealed the limited value of this perspective. Firstly, as
we found limited evidence for convergence even among rich countries
(OECD and Europe), it is likely that there would be still less convergence
across the whole world. Secondly, if, as our analyses suggest, converg-
ing trends are matched by divergence and continuing differences at the
level below formal legislation, it can seriously be questioned what the
social relevance of the convergence perspective is. We may, after all,
assume that it is the policies as implemented, the structures on the
ground, actual school practices and the curriculum as taught which are
much more important in shaping the lives of youngsters – and thus in
influencing social cohesion – than formal policy statements.
Secondly, the limited support for the convergence perspective does
not mean that any of the alternative perspectives on system variation
and change are validated automatically. Thus, while the comparative
political economy (CPE) perspective figures prominently in all the
contributions to Part I, only Chapter 1 finds countries clustering in
ways that were anticipated by this perspective (with the exception of
Germany, which joins the liberal cluster composed primarily of English-
speaking countries). Chapter 2 provides mixed support for the CPE
perspective. On the one hand, it does find evidence for a clearly demar-
cated corporatist cluster characterised by high rates of enrolment in
vocational education and strong apprenticeship systems (consisting of
the German-speaking countries, Denmark, Netherlands, and the Czech
and Slovak Republics), but on the other hand the countries with weak
apprenticeship systems and those with weak VET systems altogether
constitute very diverse groups not corresponding to any of the other
regimes proposed by the CPE perspective. More importantly, the clusters
identified on the basis of VET enrolment and apprenticeship systems
break down when other aspects of upper secondary or other stages of
304 Conclusion
education are examined. The country groups identified on the basis of

enrolment in higher education and enrolment in vocational tracks in
higher education for instance are not in agreement with any of the pro-
posed regimes. What is more, the authors of the two case-study chapters
have argued that despite national education systems still being informed
by distinct political principles all countries show signs of compromise
and hybridity. Sweden, for instance, appears to have made concessions
to the logics of the market and the Beruf (the calling) by introducing
school choice (through the creation of free schools and a voucher sys-
tem) and by setting up an apprenticeship training programme. In similar
vein Italy compromises between the principles of universalism and
merit by adding an upgraded system of vocational education (giving
access to university) to an otherwise selective upper secondary sys-
tem centred on academic achievement. Because of these findings, the
authors contend that the regimes and varieties of capitalism proposed
by the CPE perspective should be understood not as empirical realities
but as ideal types representing a set of unambiguous choices on a range
of issues. Their value then lies in highlighting the political and norma-
tive principles that are at stake in educational reforms and in locating
the countries in the conceptual space between these ideal types. Finally,
Chapter 5 has shown the limitations of the parallel development per-
spective by finding that the two dimensions of variation on citizenship
education are not related to stage of socio-economic development.
In sum the findings of Part I show that national education systems
continue to be distinctive in many ways and that no existing theoretical
framework can adequately explain these differences. Although it is not
one of the stated objectives of this book to develop a framework that
can do this, we believe we can at least draw the provisional contours of
such a framework based on the findings of the aforementioned chapters.
In our view, it is essential that any new attempt to construct a theory
that aims to explain and predict developments in education systems
should incorporate interest groups and the varying sources of power and
ideological principles they draw upon.
Part II: Tenuous links between education system

properties and social cohesion
Having established that there still is enough variation in education

systems to make cross-national comparative analysis worthwhile what
can we say about the macro-social consequences of education systems?
In other words, what picture emerges from the contributions to Part
II? One of the chapters with the most unequivocal findings is the one
on pre-school education and care (cf Chapter 7). This chapter identified
PSEC as a clear ‘win-win’ policy as it found that attending PSEC not
only improves educational performance for all social groups but also
that expanding enrolment in PSEC reduces inequalities in performance.
In fact a simulation analysis based on data on PSEC participation for
the mid-1990s (and test scores later achieved at 15 years of age) showed
that these inequalities would be smallest if participation were univer-
salised among the bottom 70 per cent of children (ranked by social
background) while remaining the same among the top three deciles.
On top of that it found that universalising PSEC had a favourable effect
on female employment, which is consistent with the proposition that
PSEC frees up mothers of young children to undertake paid work. The
chapter thus makes a valuable contribution to the achievement litera-
ture, which has demonstrated strong effects on achievement (in terms
of both average levels and disparities) for many other system proper-
ties, such as overall inputs (Hanushek, 2006; Hanushek et al., 2008),
grouping by ability (Duru-Bellat and Mingat, 1989; Lucas, 2001; Green
et al., 2006) and school autonomy (Mons, 2007; Horn, 2009; Van de
Werfhorst and Mijs, 2010). Particularly the research on grouping by abil-
ity is interesting as some studies show similar ‘win-win’ effects in terms
of comprehensive schooling both increasing overall performance and
diminishing inequalities (e.g. Gamoran, 1996; Duru-Bellat and Kieffer,
2000).
But are non-cognitive outcomes such as trust, participation and tol-
erance equally responsive to system properties? Five chapters examined
the links between aspects of education and these outcomes, which are
widely considered to be key components of social cohesion. Drawing
on Basil Bernsteins’s work on ‘educational knowledge codes’, Chapter 8
identifies three curriculum models in the compulsory education phase
across Western societies, labelled the whole education model, the pro-
ducer model and the academic model, respectively, and examines the
links between these models and a number of student attitudes towards
schooling. The whole education model prevails in the Scandinavian and
English-speaking countries and is characterised by an all-encompassing
common core curriculum, child-centred education, an informal climate,
democratic procedures within schools, and inclusion of minority reli-
gions and languages. In contrast, the other two models are characterised
by strict discipline and by a hierarchy between subjects and tracks. The
producer model is inspired by the link between school and the job mar-
ket and includes distinct vocational and academic tracks. This model
306 Conclusion
can be found in countries with strong traditions of guilds, craftsmanship

and industry (Germany; Belgium; Austria; Netherlands). The second
model can be referred to as ‘academic’ as it focuses on more traditional
and encyclopaedic contents and presents itself as remote and elevated
from daily life. France and the other Southern European countries are its
typical representatives. The models appeared to be differentially related
to student attitudes. The whole education model showed a strong pos-
itive connection with interpersonal exchanges (expressed as support
for collaboration and competition within schools). No significant link,
however, appeared between educational models and instrumental moti-
vation towards schooling and school belonging, indicating that the
hypothesis about the whole education model enabling closer interac-
tions among students and staff and stronger identification with the
school is supported only partially.
An equally qualified set of results emerges from the other contribu-
tion on compulsory education (Chapter 9). This chapter has explored
whether two dimensions of system differentiation in compulsory edu-
cation, namely grouping by ability and school autonomy, are linked
to inequalities of civic engagement among 14-year-olds. Proceeding
from the proposition that both forms of differentiation should enhance
these inequalities because of distinct peer effects and curriculum dif-
ferences across tracks and schools, it finds that grouping by ability is
indeed linked with larger gaps in civic engagement across classrooms
(which particularly applies to civic knowledge and intention to vote but
less for the other four indicators of civic engagement). School auton-
omy, however, is not associated with larger gaps. Moreover, neither of
the two forms of system differentiation appears to mediate the effect
of social background on civic engagement. In other words, on aver-
age social background did not matter more as a determinant of civic
engagement in countries with high degrees of grouping by ability and
school autonomy. The chapter offers four possible reasons for these non-
relationships, one of which holds that youngsters may only become
receptive to curriculum content relating to social and political issues
in late adolescence. If true, it would imply that citizenship programmes
aimed at enhancing levels and reducing inequalities of civic engagement
should be implemented in upper secondary.
That the link between system characteristics and non-cognitive out-
comes is not as straightforward as that between system properties and
achievement is further underlined by Chapter 10. This chapter repre-
sents one of the very few studies that have explored the links between
aspects of post-compulsory education and social cohesion. It proceeds
from the country groupings developed by Chapter 2 and relates these

to attitudes on trust, tolerance and participation among adults. These
country groups represent different categories and values on four areas of
interest: the degree and mode of VET in upper secondary; enrolment and
completion rates of upper secondary; enrolment in (vocational tracks in)
higher education; and adult vocational education. The results do not
show any consistent pattern of relationships of these country group-
ings with the three social cohesion indicators across all four areas of
interest. There are interesting findings for some individual areas, how-
ever. Thus, the countries characterised by general and school-based VET
systems, which roughly equate to the academic group of countries of
Chapter 8, perform poorly on social cohesion by comparison to other
country groups. In addition, systems ensuring a high access to adult
vocational training score notably higher on social cohesion. This rela-
tion could be due to the strong second chance character of such training,
which, by compensating for initial inequalities, is likely to foster feel-
ings of inclusion and engagement among the (disadvantaged) groups
that make use of it.
Chapter 13 is not demonstrating strong links between education and
social cohesion attitudes either. It starts by highlighting a plethora of
initiatives by the European Union to promote active citizenship and
liberal-democratic values, and continues by arguing that these initia-
tives have indeed impacted on the member states in terms of making
them adopt increasingly similar objectives on citizenship education.
Proceeding from the hypothesis developed by some convergence the-
orists that convergence in citizenship education policies has led to
convergence in active citizenship outcomes (Wiseman et al., 2011), the
chapter subsequently assesses trends in attitudes among 14-year-olds on
gender equality, immigrants, conventional political participation and
social movement participation across two points in time (1999 and
2009). It does not find a statistically significant trend of convergence
on any of the four attitude scales. The trends moreover do not go in the
expected direction: youngsters across Europe appear not to have become
more tolerant towards immigrants, more supportive of gender equal-
ity and more supportive of alternative modes of participation. They
even became less supportive of conventional modes of participation.
This suggests that, if the aforementioned policies have had any effect
at all in sustaining active citizenship, it must have been completely
overwhelmed by that of other processes (notwithstanding the possibil-
ity – noted by the authors – that the downward trends could have been
considerably worse without these policies). A likely candidate of these
308 Conclusion
other influences is the economic crisis of 2008. Previous research has

found that levels of post-materialist values (which are broadly synony-
mous with active citizenship) decline in all age groups in times of crisis
(Inglehart, 1990; Inglehart and Welzel, 2005). A recent report of the
European Commission, moreover, has shown that trust in the European
and in the national parliament has decreased most in countries most
seriously affected by the economic crisis (Hoskins et al., 2012).
In view of the at best partial evidence for a direct link between system
characteristics and social cohesion, some contributors to this book have
argued that it is primarily through the interaction with the labour mar-
ket that education shows its impact on non-cognitive outcomes. Two
chapters have addressed this possibility. Chapter 11 has investigated the
relation at the aggregate level between rates of return to educational
qualifications and a composite measure of social cohesion combining
trust, tolerance and social capital. It finds that controlling for aggregate
achievement level, enrolment in upper secondary, educational inequal-
ity and attitudes on schooling, rates of return are negatively linked to
social cohesion. In other words, the more qualifications matter for access
to prestigious, well-paid jobs in a given society, the lower that society’s
level of social cohesion. The authors explain this counter-intuitive find-
ing by noting that high rates of return do not indicate a high level of
meritocracy and social mobility (as one might be inclined to think at
face value), but, in contrast, point to the ability of privileged groups to
use the education system to secure valued positions for their offspring
and to deny disadvantaged groups access to such positions. Indeed, they
refer to research showing that high rates of return go hand in hand
with low rates of intergenerational mobility. They also find that educa-
tional cohesion, a composite variable based on student attitudes towards
teachers, schools and education in general, shows a strong positive rela-
tionship with social cohesion. In addition, they find that the impact of
aggregate achievement levels on social cohesion is rather weak, confirm-
ing other research showing that the non-cognitive benefits of education
for individuals cannot be generalized to societies (in other words, while
at the individual level education is positively linked with trust, toler-
ance and participation, at the aggregate level it is unrelated to these
social cohesion outcomes). However, when labour market dynamism,
income inequality and subjective appraisals of income inequality are
entered into the (multilevel) model, the negative relation between rates
of return and social cohesion is strongly reduced (though remaining
significant) while the effect of educational cohesion becomes insignifi-
cant. The authors therefore conclude that macro-economic conditions
and cultural discourses on deservedness are likely to have a greater

influence on social cohesion than educational factors. The finding on
rates of return also mirrors the observation of Green et al. (2006)
that the effects of education on adult social cohesion are mostly indi-
rect in that they run through the distribution of skills and incomes.
After all, rates of return could be understood as a proxy for income
inequality.
Chapter 12 has investigated the effect of over-qualification at both the
individual and societal level on trust and tolerance. Over-qualification is
understood as the gap between the qualification obtained and the qual-
ification usually required for the job. The lower the latter in relation to
the former, the more an individual is over-educated. The authors expect
over-qualification to have significant negative effects on social cohesion
as it is likely to contribute to feelings of disappointment, unmet expec-
tations and exclusion. The results show that being over-qualified as an
individual diminishes one’s trust and tolerance controlling for many
other individual and societal characteristics. With regards to tolerance,
the negative effect of over-qualification is stronger than the positive
effect of educational attainment. This suggests that raising the general
level of education may have dangerous unintended side effects if this
is accompanied with rising numbers of over-qualified people. Indeed,
the chapter also found over-qualification to have a strong negative link
with tolerance at the aggregate level (i.e. societies with high rates of
over-qualification are also the more intolerant societies). In addition,
the effect of personal over-qualification on both trust and tolerance was
not found to depend on the aggregate level of over-qualification or on
other socio-economic characteristics of the country, suggesting that the
negative effects of this phenomenon are universal (at least in the world’s
richest countries).
Reviewing all the findings of Part II we can conclude that the relation
between system characteristics and social cohesion is not as straightfor-
ward as that between system properties and achievement. The evidence
of direct links between system properties such as the curriculum, early
selection, school autonomy and VET on the one hand and social cohe-
sion outcomes on the other was found to be patchy. With the possible
exception of PSEC and adult vocational education, no clear win-win
options were identified. Yet the findings were quite telling on the effect
of educational attainment. They indicate that raising the general level of
education in a society will not by itself enhance social cohesion. To the
contrary, it may lead to higher numbers of over-qualified individuals
who, as Chapter 10 revealed, have markedly lower levels of trust and
310 Conclusion
tolerance. These findings are in agreement with those of other studies

showing that more educated societies do not have higher levels of tol-
erance (Borgonovi, 2012), political participation (Tenn, 2007; Burden,
2009) and democratic citizenship (Persson and Oscarsson, 2010). They
also broadly fit the argument that not aggregate levels but distribu-
tions of skills and qualifications are related to social cohesion (Green
et al., 2006; Schuller and Desjardin, 2010; Janmaat and Green 2013).
According to this argument, a wide distribution of skills in a society
(i.e. high educational inequality) impacts negatively on social cohe-
sion through increasing competitive anxiety, stress and social distance
between groups. The findings of Part II further suggest that it is through
the links with other social domains, and particularly with the labour
market, that education influences social cohesion, as both high rates
of return and over-qualification were found to have negative effects
controlling for many other micro and macro-level conditions.
Implications for research and policy
What can these findings tell us in terms of recommendations for future

research and policy interventions? To begin with research, we have to
observe that there are still many black boxes that existing research has
so far been unable to open. For instance, research on the socio-political
outcomes of education system characteristics other than educational
attainment is still in its infancy. Although several contributions to this
volume have not been able to demonstrate consistent links between a
number of system properties and non-cognitive outcomes, it would be
premature to conclude that public education systems, in terms of their
structures, organisation and curriculum, therefore have no role to play
in fostering social cohesion. We should not forget that educational pro-
fessionals, particularly in Western democratic countries, are reluctant to
socialise youngsters in distinct values and behaviours for fear of being
accused of indoctrination. Indeed, this can partly explain why schools
generally feel more comfortable teaching practical skills than moral and
behavioural issues. The curriculum in Western countries reflects this
as it prioritises the learning of literacy and numeracy skills without
exception. As a rule the subjects and activities serving the socialisation
function are treated as secondary, as evidenced by the limited number
of hours devoted to them, their status as optional subjects and their
casual modes of assessment. Small wonder, then, that performance is
very susceptible to systemic influences while attitudinal and behavioural
outcomes on the face of it appear not to be. However, would the latter
still be so unresponsive if schools took the socialisation task much more

seriously? Anecdotal evidence from geographical and historical contexts
where political socialisation is (or was) not a taboo (usually countries
with authoritarian regimes) shows that public education can be very
effective in inculcating certain values and identities (e.g. Hametz, 2012),
though admittedly these are not the values that Western democratic
states seek to foster. Moreover, as noted in Chapter 9, there is research
showing that inequalities of socially responsible behaviour are smaller
in schools where civic education has a prominent place in the curricu-
lum. In other words, the fact that the contributions in this book have
not found strong links between system characteristics and non-cognitive
outcomes may simply be due to the neglect of the socialisation function
across the board. Future research needs to find contexts where citi-
zenship education figures prominently in the curriculum and compare
these to contexts where it does not, in order to assess the non-cognitive
effects of learning.
More work can also be done on determining the durability of the
socio-political effects of early selection. With the exception of a study
by van de Werfhorst (2007), studies so far have tended to assess the
links between early selection and social cohesion outcomes measured at
age 15 as from this age onwards all systems in Europe become tracked.
But much research has found that socio-political attitudes are not stable
at that age and can still be influenced by educational and other condi-
tions. Of particular interest here is the role played by VET in the upper
secondary phase in the states with early selection systems. As observed
by several studies, VET in these states has considerable social status,
prepares youngsters well for the labour market through the appren-
ticeship system (Bol and van de Werfhorst, forthcoming) and is likely
to play an important role in mitigating educational inequalities across
social groups (Green et al., 2006). In contrast, VET schooling in many
states with comprehensive systems has low prestige, is associated with
school failure and is likely to only exacerbate inequality of opportu-
nity (Hoskins et al., 2012). It might be postulated that these different
socialisation experiences of youngsters in VET systems across Europe
radically alter the civic engagement patterns visible at age 15, producing
a completely different cross-national picture of civic engagement among
adults. One comparative study has indeed found very different cross-
national patterns for adolescent and early adult civic engagement levels
across Europe (Amnå and Zetterberg, 2010). Future research thus needs
to track socio-political attitudes across the life course of individuals
starting from primary education until adulthood.
312 Conclusion
What the previous paragraph is essentially calling for is research capa-

ble of assessing the net effect of different phases of education on social
cohesion outcomes across a wide selection of countries. Such research
can only be carried out if international longitudinal survey data is avail-
able tapping socio-political attitudes. Currently, no such data sources
exist, which means that for many years to come educational research
will not be able to adequately assess the effects of many system proper-
ties. Research using international longitudinal data would not only be
able to assess the net effect of distinct phases of education in each coun-
try but also to explore whether the socio-political context shapes the
effect of a particular system component (e.g. VET) on social cohesion
outcomes.
Finally, what advice can we give to policymakers? We can be brief.
Our findings do not suggest that a sole emphasis on raising education
levels will lead to more cohesive societies. Instead, more promising are
policies targeting the opposite ends of the lifelong learning continuum:
pre-school education and care (PSEC) and adult education. Our findings
indicate that universalising PSEC is a win-win policy yielding significant
social and economic benefits. They also suggest that expanding adult
education programmes with a pronounced second chance character
is particularly beneficial for social cohesion. We would further rec-
ommend policies mitigating credentialism by recognising non-formal
education and on the job training as valid educational achievements
and stimulating employers to look beyond diplomas in recruiting new
personnel. Additionally, transition possibilities across tracks and levels
should be expanded to ensure that no track represents a dead end keep-
ing youngsters and adults locked in disadvantaged positions. Together
these policies would significantly reduce educational inequality, both in
general terms and more specifically across social groups. They can fur-
ther be expected to enhance social cohesion given the evidence from
existing research that educational inequality undermines social cohe-
sion. Last but not least, resources should be made available to finance
a comprehensive international longitudinal study in Europe tracking
youngsters from early childhood into adulthood.
References
Amnå, E. and Zetterberg, P. (2010). ‘A Political Science Perspective on Socializa-
tion Research: Young Nordic Citizens in a Comparative Light’ in L. R. Sherrod,
J. Torney-Purta and C.A. Flanagan (eds), Handbook of Research on Civic Engage-
ment in Youth, Hoboken: Wiley NJ, pp. 43–66.
Astiz, M.F., Wiseman, A.W., and Baker, D.P. (2002) ‘Slouching towards Decen-
tralisation: Consequences of Globalisation for Curricular Control in National
Education Systems’ Comparative Education Review, 46, 1, 66–88.
Baker, D. P. and LeTendre, G. (2005) National Differences, Global Similarities: World
Culture and the Future of Schooling. Stanford, CA: Stanford University Press.
Ball. S. (2012) Global Education Inc: New Policy Networks and the Neo-Liberal
Imaginary. London: Routledge.
Borgonovi, F. (2012) ‘The Relationship between Education and Levels of Trust and
Tolerance in Europe’ The British Journal of Sociology, 63, 1, pp. 146–67.
Burden, B. C. (2009) ‘The Dynamic Effects of Education on Voter Turnout’
Electoral Studies, 28, 4, pp. 540–49.
Culture Theory and Comparative Education’ Comparative Education Review, 56,
3, pp. 366–93.
Duru-Bellat, M. and Kieffer, A. (2000) ‘Inequalities in Educational Opportuni-
ties in France: Educational Expansion, Democratization or Shifting Barriers?’
Journal of Education Policy, 22, pp. 541–77.
Duru-Bellat, M. and Mingat, A. (1989) ‘How Do French Junior Secondary Schools
Operate? Academic Achievement, Grading and Streaming of Students’ European
Sociological Review, 5, 1, pp. 47–64.
Gamoran, A. (1996) ‘Curriculum Standardization and Equality of Opportunity in
Scottish Secondary Education: 1984–90’ Sociology of Education, 69, pp. 1–21.
Green, A., Preston, J. and Janmaat, J.G. (2006) Education, Equality and Social
Hafner-Burton, E. M. and Tsutsui, K. (2005) ‘Human Rights Practices in a Global-
izing World: The Paradox of Empty Promises’ American Journal of Sociology, 110,
5, pp. 1373–411.
Hametz, M. (2012) In the Name of Italy: Nation, Family, and Patriotism in a Fascist
Court. New York: Fordham University Press.
Hanushek, E. A. (2006) ‘Alternative School Policies and the Benefits of General
Cognitive Skills’ Economic Education Review, 25, pp. 447–62.
Hanushek, E. A., Jamison, D. T., Jamison, E. A., and Wossman, L. (2008) ‘Edu-
cation and Economic Growth: It’s Not Just Going to School, but Learning
Something While There That Matters’ Education Next, 8, pp. 62–70.
Horn, D. (2009) ‘Age of Selection Counts: A Cross-Country Analysis of Educa-
tional Institutions’ Educational Research Evaluation, 15, pp. 343–66.
Hoskins, B., Janmaat, J. G., Han, C. and Muijs, D. (2012) ‘The De-Politicisation
of Future Workers: The Political Engagement of Young People in General and
Vocational School Tracks in Germany, Denmark and England’ Paper Presented
at the European Conference on Educational Research, Cadiz, 18–21 September
2012.
Hoskins, B., Kerr, D., Abs, H. J., Janmaat, J. G., Morrison, J., Ridley, R.,
and Sizmur, J. (2012). Participatory Citizenship in the European Union: Ana-
lytic Report. Available at: http://ec.europa.eu/citizenship/pdf/report2_analytic_
report.pdf
Inglehart, R. (1990). Culture Shift in Advanced Industrial Society. Princeton, NJ:
Princeton UP.
Inglehart, R. and Welzel, C. (2005). Modernization, Cultural Change, and Democ-
racy: The Human Development Sequence. Cambridge: Cambridge UP.
314 Conclusion
Janmaat, J.G. and Green, A (2013). ‘Skills Inequality, Adult Learning and Social
Cohesion in the United Kingdom’ British Journal of Educational Studies, 61, 1,
pp. 7–24.
Krücken, G. and Meier, F. (2006) ‘Turning the University into an Organizational
Actor’ in G. S. Drori, J. W. Meyer, and H. Hwang (eds), Globalization and Orga-
nization: World Society and Organizational Change, Oxford: Oxford University
Press, pp. 241–57.
Lucas, S. R. (2001) ‘Effectively Maintained Inequality: Education Transitions,
Track Mobility, and Social Background Effects’ American Journal of Sociology,
106, pp. 1642–90.
Mons, N. (2007). Les Nouvelles Politiques Educatives, Paris: PUF.
OECD (2010) PISA 2009 Results: What Makes a School Successful? – Resources,
Policies and Practices (Volume IV) http://dx.doi.org/10.1787/9789264091559-en.
Persson, M. and Oscarsson, H. (2010) ‘Did the Egalitarian Reforms of the Swedish
Educational System Equalize Levels of Democratic Citizenship?’ Scandinavian
Political Studies, 33, 2, pp. 135–63.
Sahlberg, P. (2011) Finnish Lessons: What Can the World Learn from Educational
Change in Finland. Teachers’ New York: College, Columbia University.
Schuller, T. and Desjardin, R. (2010) ‘Wider Benefits of Adult Education’ in
K. Rubenson (ed), Adult Learning and Education, Oxford: Elsevier, pp. 294–98.
Suárez, D. (2007) ‘Education Professionals and the Construction of Human Rights
Education’ Comparative Education Review, 5, 11, pp. 48–70.
Tenn, S. (2007) ‘The Effect of Education on Voter Turnout’ Political Analysis, 15,
4, pp. 446–64.
Van de Werfhorst, H. (2007) ‘Vocational Education and Active Citizenship
Behavior in Cross-National Perspective’ AIAS Working Paper Number 2007/62,
1 February 2012.
Van de Werfhorst, H. and J.B. Mijs (2010) ‘Achievement Inequality and the Insti-
tutional Structure of Educational Systems: A Comparative Perspective’ Annual
Review of Sociology, XXXVI, pp. 407–28.
World: The Political Socialization of Youth in Formal Mass Education Systems’
Compare, 41, 5, pp. 561–77.
Appendices
Appendix 2.1 Country code
Country code Country
AUS Austria
BEL Belgium
CZE Czech Republic
DEN Denmark
EST Estonia
FIN Finland
FRA France
GER Germany (including former GDR from
1991 onwards)
GRE Greece
HUN Hungary
IRL Ireland
ITA Italy
LAT Latvia
LIT Lithuania
LUX Luxembourg (Grand Duchy)
NL Netherlands
NOR Norway
POL Poland
POR Portugal
ROM Romania
SLK Slovakia
SLN Slovenia
SPA Spain
SWE Sweden
SWI Switzerland
UK United Kingdom
315
316 Appendices
Appendix 2.2 Indicators and statistics
Variable meaning Variable name Minimum Maximum Average
Level of vocational provision upper secondary education

Level of vocational provision PROFSS 23.60 77.30 51.98
in secondary education
Importance of apprenticeship APRSS 0 58.29 16.80
within secondary education
Success rates in upper secondary education
Rate of access to upper ESPSS 50.10 95.90 81.76
secondary education
Difference in success rates DTSS 0.29 3.69 1.49
between vocational and
general education
Higher education
Rate of access to higher ESPSUP 20 46.10 30.42
education
Level of vocational provision SUPC 1.06 57.22 20.82
in higher education
Continuing vocational training
Rate of access to a continuing FC 1.30 32.40 11.27
vocational training course
Rate of access to a continuing FCD 0.51 15.84 5.64
vocational training course
leading to a qualification
Appendix 8.1 The ‘Mons’ dataset on education systems

The original dataset (Mons et al, 2008), on which this chapter draws, has been
compiled using the following sources:
– National reports on educational systems belonging to the dataset World Data

on Education de l’UNESCO (1999, 2003),
– National education system descriptions included in international surveys con-
ducted by international organizations (TIMSS, 1995, IEA, 1997 and TIMSS,
1999, IEA, 2000),
– National and thematic reports provided by the UK organization INCA (Inter-
national Review of Curriculum and Assessment Frameword Archive).
This dataset documents educational policies related to curriculum development

implemented in the second half of the 1990 decade at primary and lower
Appendices 317
secondary levels. The specific choice of the 1990 decade time period allows us
to investigate correlations between national curriculum features and students
attitudes indicators provided by the PISA 2000 survey. In addition to these vari-
ables dealing with curricula, our investigation uses two variables drawn from a
previously designed dataset (Mons, 2004), related to the context governing the
implementation of curricula (in terms of ability-grouping and the provision of
individualised education).
Variable Categories Description l
Regional very Regional languages are of high importance in

languages important education in these countries with large
populations speaking minority languages and
specific schooling organisation for learning
minority languages (specialised schools,
minority languages compulsory)
important Regional languages are important in education
but countries with low levels of population
speaking regional languages
weak Little attention is paid to regional language
teaching
no No attention to regional languages because
very low percentage of regional minorities
Religious high Religious studies are of high importance in
education education (number of hours dedicated to this
subject is more than 7% of total teaching hour
volume during compulsory education)
medium Religious studies are offered and number of
hours dedicated to this subject varies between
4% and 6,9% of global teaching hour volume
weak Religious studies are taught at school but
number of hours is very limited (less than 4%)
no Religious studies are not included in curriculum
Civic high Pupils are highly encouraged to participate
participation in school democratic activities (official
recommendations mention all forms of pupil
participation (election of class representatives,
election of a pupil council at school level,
existence of pupil unions/associations))
weak School organisation does not provide an
encouraging framework for student civic
participation within school
318
(Continued)
Variable Categories Description l
Civic high High attention paid to civic education (high

education number of teaching hours, civic education as
a separate subject in curriculum)
medium Medium attention paid to civic education in
terms of number of hours dedicated to this
subject and subject integrated in a larger
curriculum field (e.g. history) or optional
weak Little attention paid to civic education in
curricula
Life skills high High attention paid to life skill teaching
(high number of teaching hours,
compulsory subject in primary and lower
secondary levels)
medium Medium attention paid to life skill teaching
(eg only compulsory at primary level)
weak No or very limited life skill teaching
Pre-vocational high High attention paid to career guidance
counselling (designed as a separated subject integrated in
curricula and compulsory for all students)
medium Medium attention paid to career guidance
(career guidance is provided as a separate
optional subject or as an extra-curriculum
help provided by class teacher and/or
professional advisors mainly to student
dedicated to vocational tracks)
weak Very limited attention paid to career
guidance
Pre-vocational compulsory Vocational education compulsory for all
education students at lower secondary level
optional Option vocational education at lower
secondary level
within Vocational education provided only within
track pre-vocational tracks or pre-vocational
schools at lower secondary level
no No vocational education provided at lower
secondary level
Quality of 45%–50% Average % of students who « strongly agree »
teacher– 50%–60% or « agree » with the following items in
student 60%–70% PISA 2000 and 2003 survey:
relations 70%–75% Teachers show interest in each student’s learning
If I need extra help, I will receive it from my
teachers
Teachers continue to explain until students
understand
Appendices 319
Indiscipline 15%–30% Average % of students who report

30%–35% that the following events happen
35%–40% regularly or frequently in PISA 2000
and 2003 surveys:
Noise and disorder in the class
Teacher has to wait for a long time for
students to quieten down
In the class, it takes time before the
teacher effectively can start teaching
Student tracks Tracks, frequently integrated in

grouping ability schools with differential status
(at lower grouping setting for all subjects, within
secondary) comprehensive schools
(Mons 2004)
no Mixed-ability grouping or flexible
grouping ability-grouping (for short periods of
time or for some subjects) within
comprehensive schools
Individualised widespread Widespread individualised teaching
education (small-group teaching or
(at lower individualised tutoring, representing
secondary) more than ¼ of the whole teaching
(Mons 2004) hour volume)
no or Absence of individualised teaching
remedial or individualised teaching dedicated
teaching exclusively to low-performer
students
elite tracks individualised teaching dedicated
exclusively to high-performer
students
Appendix 9.1 Composition of scale variables.

Civic Efficacy (‘Citizenship Self-Efficacy’)
How well do you think you would do the following activities?’
(1) ‘Discuss a newspaper article about a conflict between countries’

(2) ‘Argue your point of view about a controversial political or social issue’
(3) ‘Stand as a candidate in a <school election>’
(4) ‘Organise a group of students in order to achieve changes at school’
(5) ‘Follow a television debate about a controversial issue’
(6) ‘Write a letter to a newspaper giving your view on a current issue’
(not at all – not very well – fairly well – very well)
Institutional trust (‘trust in government-related institutions’)

320 Appendices
How much of the time can you trust each of the following institutions?
(1) ‘national government’

(2) ‘local government’
(3) ‘courts’
(4) ‘the police’
(5) ‘political parties’
(6) ‘national parliament’
(never – only some of the time – most of the time – always)
Gender Equality (‘attitudes towards women’s political and economic rights’)
(1) ‘Women should run for public office and take part in the government just as
men do’
(2) ‘Women should have the same rights as men in every way’
(3) ‘Women should stay out of politics’ (negative)
(4) ‘When jobs are scarce, men have more right to a job than women’ (negative)
(5) ‘Men and women should get equal pay when they are in the same jobs’
(6) ‘Men are better qualified to be political leaders than women’ (negative)
(strongly disagree – disagree – agree – strongly agree)
Ethnic tolerance (‘positive attitudes towards immigrants’)
(1) ‘Immigrants should have the opportunity to keep their own language’
(2) ‘Immigrants’ children should have the same opportunities for education that
other children in the country have’
(3) ‘Immigrants who live in a country for several years should have the oppor-
tunity to vote in elections’
(4) ‘Immigrants should have the opportunity to keep their own customs and
lifestyle’
(5) ‘Immigrants should have all the same rights that everyone else in a coun-
try has’
(strongly disagree – disagree – agree – strongly agree)
Expected political participation (‘expected participation in political activities’)

‘When you are an adult, what do you expect that you will do?’
(1) ‘Join a political party’

(2) ‘Write letters to a newspaper about social or political concerns’
(3) ‘Be a candidate for a local or city office’
(I will certainly not do this – I will probably not do this – I will probably do
this – I will certainly do this)
Appendices 321
Appendix 12.1 Multilevel model for Interpersonal trust
Parameters Empty model Model 1 Model 2
Coeff. Standard Coeff. Standard Coeff. Standard

deviation deviation deviation
Constant −0.50 0.24∗∗ −0.84 0.25∗∗∗ −3.44 1.40∗∗∗

Man − 0.03 0.04 0.05 0.04
Statistical mismatch − −0.15 0.07∗∗∗ −0.15 0.07∗∗
Age 15–24 − 0.04 0.07 0.01 0.08
Age 25–34 − −0.02 0.06 −0.02 0.06
Age 35–44 Réf
Age 45–54 − 0.043 0.05 0.02 0.06
Age 55–64 − 0.06 0.06 0.02 0.07
Age 65 and over − 0.28 0.12∗∗∗ 0.19 0.13
Situation on survey date:
Full-time waged employee Réf
Part-time − −0.03 0.06 −0.02 0.07
Self-employed − −0.04 0.06 −0.06 0.07
Unemployed − −0.48 0.08∗∗∗ −0.44 0.09∗∗∗
No data − 0.08 0.32 0.47 0.34
Cohabiting − 0.11 0.04∗∗∗ 0.09 0.05∗∗
Level of education
Primary level or lower Ref
Incomplete secondary − 0.21 0.07∗∗∗ 0.26 0.08∗∗∗
level
Technical or vocational − 0.15 0.07∗∗ 0.20 0.08∗∗∗
secondary
Completed academic − 0.45 0.07∗∗∗ 0.49 0.08∗∗∗
secondary education or
incomplete tertiary
Tertiary level − 0.47 0.07∗∗∗ 0.50 0.08∗∗∗
Level 2 variables:
Statistical mismatch 2.42 2.36
Trust in others − − − −0.03 0.01∗∗∗
Attitude to income 0.01 0.01
inequalities
Tolerance − − − 0.02 0.01∗∗
Unemployment rate − − − −0.06 0.04
Gini index of income − − − −5.34 2.61∗∗
inequalities
GNP per capita −0.00002 0.00002
Female employment rate 0.08 0.01∗∗∗∗
Education expenditure −0.03 0.11
(% of GDP)
Random effect
Level 2 (country: variance 1.22649 0.3927∗∗∗ 1.1559 0.3731∗∗∗ 0.11 0.04∗∗∗
of constants
Inter-country variance 27.2 26 3.1
(%): rho
322 Appendices
(Continued)

Log-probability −8175.75 −8118.8 −6999.3

AIC 16355.5 16273.7 14052.5
BIC 16370.6 16409.8 14252.0
Number of individuals 14 208 14 208 11 943
Number of countries 20 20 17
Appendix 12.2 Multilevel model for tolerance

Constant 2.17 0.27∗∗∗ 1.96 0.28∗∗∗ −3.46 0.72∗∗∗

Man −0.32 0.03∗∗∗ −0.31 0.03∗∗∗
Statistical mismatch −0.09 0.05∗∗ −0.12 0.05∗∗
Aged 15–24 0.02 0.06 0.03 0.06
Aged 25–34 0.08 0.04∗∗ 0.09 0.04∗∗
Aged 35–44 Ref
Aged 45–54 0.01 0.04∗∗ 0.02 0.04
Aged 55–64 −0.09 0.05∗∗ −0.08 0.05
Aged 65 and over −0.41 0.09∗∗∗ −0.44 0.09∗∗∗
Situation on survey date:
Ful-time waged employee Ref
Part-time −0.003 0.05 0.02 0.05
Self-employed −0.16 0.04∗∗∗ −0.18 0.05∗∗∗
Unemployed −0.01 0.06 0.003 0.06
No data 0.35 0.33 −0.35 0.22
Cohabiting −0.09 0.03∗∗∗ −0.10 0.03∗∗∗
Education level
Primary or lower Ref
Incomplete secondary 0.35 0.06∗∗∗ 0.36 0.06∗∗∗
Technical or vocational 0.38 0.05∗∗∗ 0.38 0.06∗∗∗
secondary
Completed academic 0.63 0.05∗∗∗ 0.65 0.05∗∗∗
secondary education or
incomplete tertiary
Tertiary level 0.80 0.06∗∗∗ 0.82 0.06∗∗∗
Level 2 variables:
Statistical mismatch −3.73 1.32∗∗∗
Trust in others 0.01 0.006∗
Tolerance 0.08 0.004∗∗∗
323
Unemployment rate −0.006 0.02

Gini index of income 1.15 1.09
inequalities
Female employment rate −0.01 0.008
Random effect
Level 2 (country): 1.35 0.46 1.40 0.48 0.04 0.02
variance of constants
Level 1: inter-individual 2.75 0.03 2.64 0.03 2.68 0.03
variance
Inter-country variance 33.0 34.7 1.4
(%): rho
Log-probability −24610. 6 −24382. 7 −23280. 9
AIC 49227.3 48803.4 46611.8
BIC 49249.7 48945.0 46796.9
Number of individuals 12 12 12
759 759 142
Number of countries 18 18 18
∗ P < .05; ∗∗ P < .01; ∗∗∗ P < .001

Author Index
Abrahamsson, K., 80 Buechtemann, C., 73

Alaluf, M., 225 Burden, B.C., 310
Alexander, R., 183 Busemeyer, M., 19
Algan, Y., 223
Almond, G., 155, 203 Cahuc, P., 223
Amadeo, J., 207 Caillods, F., 4
Archer, S., 218 Campbell, D.E., 2
Ashton, D., 19 Carney, S., 4, 119, 302
Astiz, M.F., 28, 118, 300 Cedefop, 100, 101, 102, 241
Aventur, F., 73 Chan, E., 3
Chan, H.C., 134, 267
Baker, D.P., 4, 300 Chan, J., 3, 223, 225, 230, 267
Ball, S., 37, 41, 42, 300 Charest, J., 49, 56, 57, 95
Ballatore, M., 101, 105, 106 Chaumont, J.M., 153
Barber, C., 287, 290 Checci, D., 97
Bartels, L.M., 202 Chevalier, A., 163
Bartlett, W., 78 Chiapello, E., 145
Baudelot, C., 223 Chubb, J.E., 78
Baumann, G., 5, 283 Clegg, J., 168
Beauvais, C., 224, 248 Cognard, E., 50
Beck, U., 4, 144, 295 Coleman, J.S., 142, 154, 155, 247
Becker, G., 78 Collins, R., 255
Bellah, R., 203 Connelly, R., 163
Benavot, A., 117, 282 Conseil de l’Europe, 224
Berger, P., 156 Cort, P., 84, 102
Berger-Schmitt, R., 224 Couppié, T., 47
Bernstein, B., 183, 185, 186, 189–90, Crick, B., 123, 125
198, 199, 305 Crossley, M.W., 16
Bevort, A., 153 Crouch, C., 19, 38, 39
Boli, J., 17, 117 Crul, M., 205
Boltanski, L., 73, 145 Culpepper, P., 71
Borgonovi, F., 2, 248, 310
Bosch, G., 49, 56, 57, 82, 83, 95 Dalton, R.J., 120, 121
Bouchard, L., 225 Danish Ministry of Education, 100,
Boudon, R., 265n11 102, 108
Bougroum, M., 78 Desjardin, R., 310
Bourdieu, P., 86, 143, 154, 182 Di Paola, V., 9, 223, 240, 242, 267
Bovens, M., 219 Dimeglio, I., 3, 223, 224, 226, 231,
Brockmann, M., 228 232, 236, 242
Brown, P., 19, 38, 39 Donzelot, J., 143
Brubaker, R., 5, 283 Dore, R., 16, 19
Brunello, G., 205 Doyle, A., 5
324
Author Index 325
Dubet, F., 2, 50, 141, 145, 151, 223, Greinert, W.D., 19, 48, 51, 56
224, 228, 230, 238, 240, 241, 245, Griffiths, T., 16, 17
255, 256, 276 Grimaldi, E., 97
Dumas, A., 9, 46, 253 Grimshaw, D., 49, 50
Duran, P., 147
Duru-Bellat, M., 9, 74, 85, 134, 181, Habermas, J., 145, 146
245, 235, 256, 269, 305 Hafner-Burton, E.M., 300
Hall, P., 16, 19, 70
Elster, J., 268 Hallinan, M.T., 205
Epstein, E., 16 Hametz, M., 311
Esping-Andersen, G., 5, 16, 19, 50, 72, Han, C., 116
74, 89n1, 122, 143–4, 150, 160, Hank, K., 163
164, 168, 169–71, 177 Hanushek, E.A., 305
Estevez-Abe, M., 50, 56, 70, 227, Harrison, L.E., 121
240 Haveman, R., 278
Eurostat, 99, 104, 105, 109 Heckman, J., 163
Eurydice, 98, 117, 135–6n2, 213, Held, D., 4, 16, 288
217 Hilbert, J., 81
Evans, M., 163, 167, 168 Honneth, A., 146–7
Eyraud, F., 76 Horn, D., 305
Hoskins, B., 203, 219, 281, 283, 284,
Favell, A., 283 295, 308, 311
Feinstein, L., 239 Huntington, S., 16, 121, 134
Field, S., 52, 228, 241 Hutton, W., 16, 19, 164
Flabby, L., 97 Hyland, N.E., 205, 206
Florenzano, F., 97, 101, 109, 110
Frank, D.J., 116 Ibourk, O., 78
Fukuyama, F., 16, 120 Inglehart, R., 120, 121, 130, 203, 282,
285, 308
Galston, W., 203 Iribarne, P. de, 151
Gamoran, A., 305 Iversen, T., 5, 50, 164
Garonna, P., 81
Gauchet, M., 147 Jacques, M., 19
Géhin, J-P., 51, 82 Jakobi, A.P., 119, 132
Geroldi, G., 109 Jallarde, J., 19
Giddens, A., 79, 144, 295 Janmaat, J.G., 19, 116, 122, 134, 169,
Giraud, O., 70 202, 203, 205, 206, 207, 223,
Giret, J.F., 268 225–6, 229, 231–2, 236, 247, 267,
Goossens, L., 218 283, 310
Granovetter, M., 155 Jarvis, P., 4
Gray, J., 15 Jeannotte, M.S., 223, 224
Green, A., 2, 3, 4, 15, 18, 19, 20, 36, Jennings, M.K., 217
38, 48, 52, 66, 73, 119, 120, 122, Jenson, J., 156, 224–5, 248
124, 151, 160, 164, 165, 169, 170,
205, 223, 224, 225, 226, 227, 228, Kasahara, S., 133
229, 241, 247, 250, 261, 267, 281, Keating, A., 282
282, 283, 305, 309, 310, 311 Kelly, J., 163, 167–8
Green, F., 19 Kerckhoff, A.C., 252
Greenfeld, L., 15 Kerr, C., 16
326 Author Index
Kerr, D., 122–3, 295 Méhaut, P., 46, 51, 74, 82, 94, 223,
Kieffer, A., 85, 305 226, 231–2, 236, 253
Kimmel, J., 163 Meilland, C., 84
Knezevic, L., 16, 17 Mengucci, R., 100
Koch, R., 81 Merle, V., 76
Kok, W., 160 Meuret, D., 86
Koopmans, R., 121 Meyer, J.W., 4, 16–18, 41, 118
Kreyenfeld, M., 163 Mijs, J.B., 203, 205, 207, 214,
Krücken, G., 300 305
Kuhlthau, K., 163 Ministry of Education, Culture
Kumar, K., 15 and Science (Netherlands),
Kuzio, T., 121, 130–1 108
Kvist, J., 108 Möbus, M., 73
Moe, T.M., 78
Lahire, B., 145 Mons, N., 19, 47, 49, 51, 59, 65–6, 80,
Lallement, M., 153 82, 134, 149, 181, 189, 205,
Langton, K.P., 217 206–7, 230, 305, 316–19ap
Lascoumes, P., 148 Morris, P., 116, 119, 122, 132
Lawn, M., 281, 282 Moscati, R., 107
Layard, R., 165, 178n2 Mostafa, T., 15, 20, 119, 160, 165,
Leclercq, F., 223 171
Le Galès, P., 148 Moullet, S., 9, 240, 242, 267
Legrand, J., 78 Mrok, J., 98
LeTendre, G., 4, 300 Muller, P., 147
Levinson, M., 202 Müller, W., 47, 240, 253
Lichterman, P., 203
Lin, N., 155 Netjes, J.E., 217
Lingard, B., 41, 281, 282 Nickell, S., 165
Lipset, M.S., 16 Nie, N.H., 2
List, F., 15 Noll, H.H., 256
Lorrain, D., 153 Novoa, A., 281
Loveless, T., 205
Lucas, S.R., 305 OECD, 28, 33, 41, 81, 84, 85, 87, 94,
100, 169, 176, 208
Macedo, S., 206 Ohmae, K., 4, 16
Maggi-Germain, N., 76 Olympio, N., 46, 51, 223, 253
Mahnig, H., 120 Osberg, L., 256
Mannitz, S., 5 Osborn, M., 190
Mansuy, M., 47 Oscarsson, H., 310
Maroy, C., 41, 42, 73
Marsh, C., 122 Paci, M., 109
Mason, K.O., 163 Palier, B., 72
Massey, D., 250 Pedersen, L., 108
Maurice, M., 67, 71, 82 Persson, M., 310
May, T., 116 Peugny, C., 268
McCracken, M., 225 Polanyi, K., 142
McEneaney, E., 118 Ponthieux, S., 279n3
McLean, M., 19 Porter, A., 123, 125
McMahon, W.W., 2 Powell, L., 163
Author Index 327
Preston, J., 239 Ten Dam, G.T.M., 204

Putnam, R.D., 142, 155, 203, 247–8, Tenn, S., 310
279n3, 295 Tessaring, M., 87
Thelen, K., 16, 19, 71
Ragin, C., 6, 54, 232 Thévenot, L., 73
Ramirez, F.O., 4, 16, 118, 281–2, 284, Tickly, L.P., 16
288 Tletemann, J., 119, 132
Rasmussen, M., 98 Torney-Purta, J., 219, 286, 287, 290
Rauner, M.H., 118 Trampusch, C., 19
Ribar, D., 162
Richini, P., 163 Unevoc, 52, 241
Ridley, R., 281 Uslaner, E.M., 248
Roberts, L.W., 256
Robertson, R., 16 Vaniscotte, F., 96, 98
Romano, R., 100 Velde, C. van de, 84, 103
Rosanvallon, P., 219 Verba, S., 155, 203
Rostow, W.W., 16 Verdier, E., 8, 19, 51, 56, 59, 65, 66,
Rubery, J., 49, 50 70, 71, 73, 90n3, 95–6, 98–9, 111,
Rubin, B.C., 205, 206 112, 164, 229, 230
Rui, S., 147
Vérétout, A., 2, 245, 254, 283
Ryan, P., 81, 86
Vermeulen, H., 205
Viitanen, T., 163
Sabates, R., 223
Vinokur, A., 74
Sahlberg, P., 302
Volman, M., 204
Salatin, A., 97
Salzbrunn, M., 82
Waldfogel, J., 168
Sauvageot, C., 53
Wallerstein, I., 16, 17
Savina, Y., 134, 181
Schissler, H., 116 Wannan, J., 87
Schuller, T., 310 Waterman, A., 218
Schulz, W.J., 207–8 Weber, M., 73, 74, 148
Sen, A., 51, 268 Welzel, C., 299, 303, 327
Serpieri, R., 97 Werfhorst, H. van de, 204, 205, 206,
Shavit, Y., 47, 240, 253 207, 214, 305, 311
Sibberns, H., 288, 289 Westergaard-Nielsen, N., 109
Simonsen, M., 163 Westerhuis, A., 57, 103, 104, 105
Singly, F. de, 145 Whitty, G., 42, 79
Sklair, L., 4 Wiborg, S., 41, 76, 84, 90n4, 97–8,
Smeeding, T., 256 102
Soskice, D., 16, 19, 70 Wilkinson, R.G., 169
Soysal, Y.N., 116, 284 Wimmer, A., 120
Steedman, H., 87, 89n2 Wiseman, A.W., 4, 116, 118, 282, 285,
Steiner-Khamsi, G., 4, 17–18 307
Stenberg, A., 80 Wolf, P.J., 206
Stephens, J.D., 50, 50, 96, 164 Wolfe, B., 278
Stevens, F., 205, 219
Streeck, W., 16, 19, 71 Young, M.F.D., 182
Stubager, R., 2
Suárez, D., 300 Zimmerman, B., 148
Subject Index
Tables and figures are referred to by page numbers in italics. Chapter endnotes
are referred to by page numbers followed by an ‘n’ and endnote number. Content
referred to in appendices is indexed by page numbers followed by ‘ap’.
Cited authors have been indexed in a separate author index.
ability grouping Belgium, 20, 57, 62

and civic engagement, 204–6 curriculum typology, 190
convergence in, 36–9, 42 LLL education, 64
in curriculum patterns, 189–98, 194, and PSEC, 164, 167, 167,
195, 196, 199, 316–19ap 168
on educational regimes, 20, 25–8, Beruf, 49, 67, 73, 229, 304
26, 27, 103 Berufsakademie, 81
and inequality reduction, 202, Bologna process, 46, 105, 107
204–9, 306 Boolean logic, 232–3
academic regimes, 66, 77, 77, 84–7,
102, 111–13, 230 capabilities theory, 51
see also decommodification centralisation
access rates and inequality, 206–7, 208, 213,
higher education, 60–3, 61, 85, 105, 214–16
236–8, 237, 316ap see also decentralisation
training, 86, 238–9, 238, 316ap citizenship education, 116–17
upper secondary education, 58–60, and cosmopolitanism, 286, 288,
58, 234–6, 235, 316ap 293, 301
vocational education, 63–5, 64, cross-national studies, 122,
316ap 124, 147, 149,
adult qualifications, 31–2, 31, 32 291–4
Anglo-Saxon countries and globalisation, 284–6
models of education, 48, 86–7 indicators of, 124–9, 126, 127,
see also UK 286–90
apprenticeships and inequality, 204, 209, 213,
and access to higher education, 216–19
61–5 theories of, 117–23
cross-national studies, 81–7 value convergence, 281, 282–4,
and models of education, 48, 53, 291–4
56–9, 88, 94, 316ap CITO test, 98, 99
and social cohesion, 228–9 CIVED (Civic Education Study), 3,
and upper secondary education, 207, 213, 282, 285–6,
102–4 287–90
and vocationalisation, 234 Upper Secondary Study, 214
Austria, 18, 48, 56, 61, 229 civic engagement, 202–4
curriculum typology, 190 and ability grouping, 204–6
LLL education, 63, 109, 239 cross-national studies, 209–14, 210,
and PSEC, 164, 167, 167 215
328
Subject Index 329
indicators of, 319–20ap cosmopolitanism

and inequality, 203–20, 210, 212, and citizenship education, 286, 288,
215, 216, 306 293, 301
civic participation Council of Europe
and social cohesion, 231, 235, 236, and national policies, 281, 283,
237, 237 302
value convergence, 294–6 social cohesion, definition of, 235
Clash of Civilisations (Huntington), 16 CPE, 5, 8, 303–4
cluster analysis, 39–40, 40 Cramér’s V, 186–7, 187, 200n1
Coefficient of Variation, see CV credentialism
computation and social cohesion, 255–6, 312
cohesion, see social cohesion cross-national studies, 83–9
collection model, 183, 186, 190–3, 199 citizenship education, 122, 124,
see also producer model; total 147, 149, 291–4
education model civic engagement, 209–14, 210,
Commissariat Général au Plan 215
social cohesion, definition of, 225 compulsory education, 96–9,
comparative political economy, see 113n2–3, 124–6
CPE convergence trends, 36–9, 299–304
comprehensive education, 20, 47–8, curriculum models, 188–93, 188,
52, 59, 65–6, 96–9, 112 191
and equality, 204–7, 209–14, 216 decision-making levels, 28
and social cohesion, 227–30, 240, education vs. employment, 269–8,
305 273, 275
see also compulsory education educational spending, 21–3, 22
comprehensivisation higher education, 105–8
and school autonomy, 211–14, 212, PSEC and female employment,
216 164–8, 167
compulsory education, 46–7, 305–6 school organisation, 23–8
cross-national studies, 96–9, skills outcomes, 29–36, 29, 30, 31,
113n2–3, 124–6 32
curriculum patterns, 186–93 upper secondary education, 56–60,
and social cohesion, 226–30 99–105
convergence of systems, 4, 6, 9 vocational education, 65–7
in ability grouping, 36–9, 42 curricula
in citizenship education, 281, and ability grouping, 189–98, 194,
282–4, 291–4 195, 196, 199, 316–19ap
cross-national trends, 36–9, 299–304 and compulsory education, 186–93
debates on, 15–20 cross-national studies, 188–93, 188,
findings, 21–36 191
in hybridisation models, 94–6, 105, and decentralisation, 117–18, 124,
111–13 130, 131–2
and national curricula, 118–23 definition of, 182
corporatist regimes, 57, 59, 74–6, 75 and democracy, 124–5, 131–5
cross-national, 81–6, 88 in Mons dataset, 316–19ap
hybridisation models, 104, 106, student attitudes, 181; Cramér’s V,
109, 111, 112 186–7, 187; national models,
see also decommodification 183, 189–93, 188, 191, 198–200
330 Subject Index
CV computation, 20–1, 42n1 Education at a Glance, 3, 7, 19, 53–4

on decision-making, 28 Education for Democratic Citizenship,
on educational spending, 21–3 281
on school organisation, 23–7 ‘education knowledge codes’, 183
on skills outcomes, 29–35 education systems, 3, 15–42, 149
convergence in, 94–6
datasets, 3, 7, 20, 28 national systems, 56–64
on civic engagement, 213–15 perspectives, 119–23, 131–5
Mons dataset, 316–19ap see also models; regimes; World
on social cohesion, 245, 246 Culture theory
see also PISA studies educational spending, 21–3, 22, 24,
day-care centres, 150 38–9, 83, 109, 300
decentralisation, 28, 299, 301 employment, 51, 73, 84, 87–8, 240
and civic engagement, 206 equalities in, 151–2
and curricula, 117–18, 124, 130, female (PSEC), 160–1, 162–70,
131–2 177–8
and social cohesion, 148 and social cohesion, 253–4, 254
see also centralisation see also labour
decision-making level, 28 Enlightenment theories, 15, 84, 142
decommodification, 74–6, 75, equality
89n1 and comprehensive education,
see also academic regimes; 204–7, 209–14, 216
corporatist regimes; universal and gender, 80, 102, 204, 282, 285,
regimes 288–9, 292–3, 294, 320ap
democracy, 84, 155, 202, 287 and social cohesion, 151–3
and curricula, 124–5, 131–5
see also inequality
see also participatory democracy
ESCS
Denmark
and PSEC, 171–7, 173, 174, 175
compulsory education, 97–9
and school variance, 34–5, 35, 37
folkeoplysning, 84
ethnicity
higher education, 106
and inequality, 202, 204, 205–7,
LLL education, 83–4, 87–8, 108–9
209, 211–13, 214–19
and PSEC, 163
and tolerance, 288, 291, 320ap
upper secondary education, 102–3
European Union
vocational education, 63
models of education in, 18, 46, 52,
differentiation, see inequality
105
divergence, see convergence of
systems policy convergence, 281, 282–4,
296
East Asia, 8, 19, 32, 301 Eurostat, 52, 53–4
and curricula, 117, 122–3, 130,
132–5 France, 8, 49, 58, 60, 67, 301, 306
and social cohesion, 226 citizenship education, 125, 130,
see also Japan 147, 149
East Europe, 57, 58, 164 hybridisation models, 71, 82, 84–6,
see also Hungary; Romania 88
ecological analysis, 246 and PSEC, 164, 167
Economic, Social and Cultural Status, and social cohesion, 256, 272
see ESCS fuzzy-set analysis, 6
Subject Index 331
GDP indicators
and educational spending, 21–3, 22 on citizenship education, 124–9,
gender equality, 80, 102, 204, 320ap 126, 127, 286–90
and citizenship values, 282, 285, on civic engagement, 319–20ap
288–9, 292–3, 294 on social cohesion, 230–2, 247–50,
Germany, 15, 19, 41, 51, 57, 61, 65, 271–8, 273, 274, 275, 276
301 on vocational education, 53–4,
Berufsakademie, 81 316ap
LLL regimes, 71, 81–3, 88 individualism vs. social cohesion,
and PSEC, 163, 164, 167, 169 144–7, 154
and social cohesion, 227, 229, 241
inequality, 42, 47–50, 75, 79–81, 87,
Gini index, 36, 250, 270, 273, 275,
89, 99, 109, 311–12
276, 279n4
and centralisation, 206–7, 208, 213,
globalisation, 4, 16, 18, 19, 41, 118,
214–16
223
and citizenship education, 284–6 and citizenship education, 204, 209,
Government of Canada’s Policy 213, 216–19
Research Sub-Committee and civic engagement, 203–20, 210,
social cohesion, definition of, 224 212, 215, 216, 306
graduation rates and ethnicity, 202, 204, 205–7, 209,
adult qualifications, 31–2, 31, 32 211–13, 214–19
upper secondary education, 29–30, income inequalities, 270, 272, 273,
30, 31 274, 275, 276, 276, 308
Greece, 5, 20, 48, 60, 62, 64, 293 and PSEC, 162, 168–70, 171–2,
Gross Domestic Product, see GDP 176–8, 305
and social cohesion, 142, 151–3,
heterogeneity, 48, 62, 66, 96, 165, 230 227, 249–52, 250, 252, 256–8,
higher education 257, 260–4, 260
access rates, 60–3, 61, 85, 105, see also equality
236–8, 237, 316ap integrated models, 48, 49, 65, 105,
see also tertiary education; 229, 236, 240
vocational education integrated vocationalism, 59, 84–6,
human capital, 49–51, 70, 157, 300 88
human rights integration
and citizenship education, 118, 123,
and social cohesion, 146, 148, 157,
132, 285, 288, 289, 294, 301
250–2, 252
Hungary, 58, 60, 62, 64, 66, 124, 190
International Adult Literacy Survey,
hybridisation models
227
convergence in, 94–6, 105, 111–13
LLL education, 70–2 International Civics and Citizenship
Education Study, see ICCS
IALS, 227 Intraclass Correlation Coefficients,
ICCC, 209 209
ICCS Investments, see spending,
civic values, 282, 285, 286–90 educational
school autonomy, 207–9 ISSP (International Social Survey
IEA studies, 282, 285, 286–7 Programme), 265n9
income inequalities, 270, 272, 273, female employment, 163–4
274, 275, 276, 276, 308 inequalities, 246, 256
332 Subject Index
Italy models
compulsory education, 96–8 of education, 46, 47–9, 52, 65–7,
higher education, 62, 106–7 79–87
LLL education, 110–11 of societal organisation, 17–18
social capital, 155 see also collection model; curricula;
upper secondary education, 99–105 producer model; total
education model
Japan, 20, 24, 33, 35, 36, 38, 49 modernisation theory, 15,
curriculum patterns, 122, 124, 131, 119–21, 123, 133–4, 282,
133–5 285
and social cohesion, 253, 255, 273 Mons dataset, 316–19ap
multilevel analysis, see MLA
Knowledge and Control (Young), 182
‘Knowledge Lift’ programme, 80 national systems, 56
higher education, 60–3, 61
labour, 2, 9, 30, 49–50, 151, 227, LLL education, 79–87
264n2 upper secondary education, 56–60,
and LLL education, 70–9 58
and social cohesion, 226, 228, 242, vocational education, 63–5, 64
249–50, 250, 268, 271–2, 276–8,
308, 310
NEET (not in education, employment
see also employment
or training), 87, 88
lifelong learning, see LLL education
Netherlands, The
Lisbon strategy, 72, 109, 148
ability grouping, 204
literacy tests, 29, 29
access to education, 57, 59, 62, 63,
LLAKES (Learning and Life Chances in
65
Knowledge Economies and
compulsory education, 96, 98,
Societies), 124, 128, 135
99–100, 103
LLL education, 8, 9
models, 20, 46, 70–2, 79–87, higher education, 105–8
89–90n3 inequalities, 217
public policy, 72–4; decommodified, LLL education, 110–12
74–6, 75; market regimes, 77–9, New Deal for Young People, 86–7
77 Nordic countries
and unemployment, 108–11 access to education, 57, 59, 63, 65,
and vocational education, 63–5, 63 236
Luxembourg, 41, 57, 58, 64, 236 inequality mitigation, 160–1
models of education, 19, 41, 48
Mediterranean countries, see Southern and PSEC, 169
Europe see also Denmark; Sweden
meritocracy, 58, 60, 66–7, 85, 105,
110, 301 OECD, 3, 7, 8, 224, 241, 302
and social cohesion, 228, 230, regimes of social cohesion, 23–5,
235–6, 249, 257–8, 263, 268 226
MLA, 6–7, 261, 262, 274 regulations, 28, 41
for tolerance, 322–3ap see also Education at a Glance; PISA
for trust, 321–2ap studies
mobility, social, 142, 151, 161, 205, OMC (open method of coordination),
253, 267 71–2
Subject Index 333
parallel development, 4–5, 112, 120, of social cohesion, 225–6, 247

304 universal, 15, 51, 65, 77, 88, 98,
participatory democracy, 120, 147 103; in PSEC, 161, 169–77, 171,
PISA Shock, 82 172, 175
PISA studies, 7, 20, 37–9, 42, 83, 149, Romania, 59, 62, 66
251
PSEC, 161, 177, 178n4 Scandinavia, see Nordic countries
school organisation, 23–8 school autonomy, 37, 39, 41, 98
skills outcomes, 29–36 and comprehensivisation, 211–14,
student attitudes, 184–6, 199–200, 212, 216
317ap
and ICCS, 207–9
post-compulsory education, 6, 8, 9,
school organisation, 23–8
46, 300
secondary education, see upper
see also compulsory education;
secondary education
training, continuing; vocational
skills outcomes, 29–36, 29, 30, 31, 32
education
social capital
pre-school education and care, see
and social cohesion, 153–7, 225,
PSEC
247–9, 248
Private Dependent Schools, 24, 38, 39
social cohesion
privatisation, 23, 24, 37, 41, 42
academic debates, 2
producer model, 190, 192, 193, 197,
198, 199, 305 access rates, 234–9, 235, 237, 238
see also collection model; total and decentralisation, 148
education model definition of, 141–4, 224–6, 247,
PSEC 267
ESCS data, 171–7, 173, 174, 175 and education, 226–30, 226,
female employment, 160–1, 177–8, 239–42, 250–6, 252, 258–61,
305; cross-national studies, 259, 267–8, 304–10
164–8, 167 and employment, 253–4, 254
and inequalities, 162, 168–70, indicators of, 230–2, 247–50, 271–8,
171–2, 176–8, 305 273, 274, 275, 276
public intervention, 78 and inequality, 142, 151–3, 227,
and social cohesion, 143, 144 249–52, 250, 252, 256–8, 257,
260–4, 260
QCA (qualitative comparative and integration, 146, 148, 157,
analysis), 47 250–2, 252
education vs. social cohesion, and labour, 226, 228, 242, 249–50,
232–3, 233 250, 268, 271–2, 276–8, 308,
on vocational education, 54–5 310
qualitative perspective, 6, 95, 121–3, and MLA, 6–7, 261, 262
134, 204, 264 and public intervention, 143,
quantitative perspective, 6, 16, 79, 194 147–50
QCA method, 232–3, 233
regimes, 5, 8, 50, 122, 124, 164 as societal phenomenon, 3
academic, 66, 77, 77, 84–7, 102, socioeconomic factors, 245–6,
111–13, 230 258–62, 260, 263–4
corporatist, 57, 59, 74–6, 75, 81–8, studies on, 223–4
103, 104, 106, 111, 112 social mobility, 142, 151, 161, 205,
LLL regimes, 72–90, 75, 77, 89–90n3 253, 267
334 Subject Index
Southern Europe, 5, 32, 48 and social cohesion, 231–8, 235,

see also France; Greece; Italy; Spain 237, 247–50, 249, 267–79, 273
Spain, 5, 20, 58, 60, 66, 124 see also tolerance
and social cohesion, 255
spending, educational, 21–3, 22, 24, UK, 20, 67
39, 71, 83, 109, 300 LLL regimes, 86–7, 88, 89, 94
Stages of Economic Growth (Rostow), 16 and PSEC, 168, 169, 171–7
student attitudes unemployment, 50, 80, 264n2
competitive vs. co-operative and LLL, 108–11
learning, 193–200, 194, 195, and PSEC, 164–5
196, 197 and social cohesion, 273, 276–7,
and curricula, 184–93; Cramér’s V, 275, 276, 321–2ap
186–7, 187 see also employment
Sweden, 58, 62, 301, 304 United States, 42, 86, 122, 124,
LLL education, 79–81, 88, 89 204
Switzerland and social cohesion, 151, 250,
higher education, 61–3 255–7
Systèmes d’enseignement et systèmes de universal regimes, 15, 51, 65, 77, 88,
pensée (Bourdieu), 182 98, 103
and PSEC, 161, 169–77, 171, 172,
175
tertiary education, 23, 24, 37–9,
universities, see higher education
271
upper secondary education
graduation rates, 30–2, 31, 32 access rates, 58–60, 58, 234–6, 235,
see also higher education 316ap
Thatcher administration, 86, 88, 89 and apprenticeships, 102–4
tolerance cross-national studies, 58–60,
and ethnicity, 288, 291, 320ap 99–105
multilevel model, 322–3ap graduation rates, 29–30, 30, 31
and social cohesion, 232, 235, and social cohesion, 234–6,
236, 237, 237, 248–50, 235
249
see also trust variance trends, 33–6, 33, 34, 35
total education model, 189–90, 193, Varieties of Capitalism, see VoC
197–200 VET, 8, 303, 311
training, continuing, 63–4, 64, hybridisation models, 79–89
74–6, 77–9, 83–4, 86–7, 108, VoC, 8, 50–1, 58, 67, 240
110 vocational education, 48–9, 60,
and social cohesion, 226–30, 61
238–42, 238 cross-national studies, 65–7
see also apprenticeships; LLL and curricula, 185, 187, 192
education; vocational education indicators on, 53–4, 316ap
trust and LLL education, 63–5, 63
institutional, 203, 204, 211–12, models, 81–6, 100, 103–4,
210, 212, 215, 216, 295, 229–30
319–20ap QCA method, 54–5
interpersonal, 227, 230, 254, see also higher education
268–71, 274, 274, 321–2ap Vocational Education and Training,
and social capital, 155–6 see VET
Subject Index 335
vocationalisation, 53–4, 56, 57, 62–3 criticisms of, 17–18, 116

and social cohesion, 234–9 national curricula, 117–20, 132
vocationalism, integrated, 84–6 WVS (World Values Survey), 3, 242n1,
246, 264n1, 285
World Culture theory, 16, 20, 37, 41 education and employment,
citizenship values, 284, 286, 294 269–71

(Education, Economy and Society) Jan Germen Janmaat, Marie Duru-Bellat, Philippe Méhaut, Andy Green - The Dynamics and Social Outcomes of Education Systems (2013, Palgrave Macmillan) PDF

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

(Education, Economy and Society) Jan Germen Janmaat, Marie Duru-Bellat, Philippe Méhaut, Andy Green - The Dynamics and Social Outcomes of Education Systems (2013, Palgrave Macmillan) PDF

Hochgeladen von

Copyright:

Verfügbare Formate

The Dynamics and Social Outcomes of Education Systems

Education, Economy and Society

Series Editors: Andy Green, Institute of Education, University of London, UK;

Education, Economy and Society

Jan Germen Janmaat

List of Tables vii

Notes on Contributors xii

Part I The Variation and Dynamics

1 The Dynamics of Education Systems: Convergent and

2 From Upper Secondary to Further Education: European

3 Lifelong Learning Regimes versus Vocational Education

4 Formal or Actual Convergence? Three Cases of

5 Curriculum Patterns in Citizenship Education:

Part II The Social Outcomes of Education Systems

7 Pre-School Education and Care: A ‘Win-Win’ Policy? 160

8 Educational Models and Their Impact on Student

9 Educational Differentiation and Inequalities of Civic

10 Lifelong Learning and Social Cohesion 223

11 Education and Social Cohesion in a Comparative

12 The Mismatch between Level of Education and Job

13 A Common Civic Culture in Europe: Has the Process of

Author Index 324

Subject Index 328

1.1 Student–teacher ratios by country groups, 2000, 2009 22

9.4 Educational differentiation and inequalities of civic

1.1 Total public expenditure on education as a percentage

5.5 The relation between democratic consolidation and

This book marks the result of an inspiring cooperation between British

Isabelle Dimeglio completed her doctoral research at the Institute of

Vanessa di Paola is Associate Professor at Aix-Marseille University,

François Dubet is Professor of Sociology at the University of Bordeaux,

Audrey Dumas is Assistant Professor (Maitre de Conference) in Eco-

Marie Duru-Bellat is Professor of Sociology at Sciences Po-Paris, France.

Andy Green is Professor of Comparative Social Science at the Institute of

Christine Han is Lecturer in Education at the Department of Life-

Bryony Hoskins is Senior Lecturer in Education at the University of

Jan Germen Janmaat is Reader of Comparative Social Science at the

Tom May is a researcher at the UK Institute of Leadership and Man-

Philippe Méhaut is Senior Researcher at the Institute of Labour Eco-

Nathalie Mons is Professor of Sociology at the University of Cergy-

Paul Morris is Professor of Education at the Institute of Education,

Tarek Mostafa is a research officer at the centre for Longitudinal Studies

Stéphanie Moullet is Labour Economist at the Labor Economics and

Noémie Olympio is a PhD student in labour economics at the Institute

Rebecca Ridley is a PhD student at the University of Southampton,

Yannick Savina is Research Engineer at the Observatoire Sociologique

Eric Verdier is Research Professor at the National Center for Scientific

Antoine Vérétout is a sociologist in the Department of Sociology at the

Interest in education has continued to grow along with the development

The first objective: Assessing the link between

studies showing the economic and sociocultural benefits that education

education other than attainment are properties of national systems (e.g.

The second objective: Exploring the variation

The cross-national comparative approach also necessitates a rigorous

claiming that the education systems of modern service-based economies

education systems will diverge. On the other hand, where there is a

Other distinctive aspects

The book is distinctive in three other respects. Firstly, we proceed from a

The structure of the book

In view of the possibility of system convergence, an analysis of the

by convergence theory) but in the opposite direction towards greater

on some indicators, they also find examples of divergence on others.

The ambiguity of a direct link between education and social cohesion is