You are on page 1of 27

Increasing Rejection of Intimate Partner Violence: Evidence of Global Cultural Diffusion

Author(s): Rachael S. Pierotti

Source: American Sociological Review, Vol. 78, No. 2 (April 2013), pp. 240-265
Published by: American Sociological Association
Stable URL:
Accessed: 18-08-2016 19:10 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact

American Sociological Association, Sage Publications, Inc. are collaborating with JSTOR to digitize,
preserve and extend access to American Sociological Review

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to

American Sociological Review

78(2) 240-265
Increasing Rejection of Intimate American Sociological
Association 2013

Partner Violence: Evidence of DOI: 10.1177/0003122413480363

Global Cultural Diffusion SAGE

Rachael S. Pierotti3


This study extends existing world society research on ideational diffusion by going beyond
examinations of national policy change to investigate the spread of ideas among nonelite
individuals. Specifically, I test whether recent trends in women's attitudes about intimate
partner violence are converging toward global cultural scripts. Results suggest that global
norms regarding violence against women are reaching citizens worldwide, including in some
of the least privileged parts of the globe. During the first decade of the 2000s, women in 23
of the 26 countries studied became more likely to reject intimate partner violence. Structural
socioeconomic or demographic changes, such as urbanization, rising educational attainment,
increasing media access, and cohort replacement, fail to explain the majority of the observed
trend. Rather, women of all ages and social locations became less likely to accept justifications
for intimate partner violence. The near uniformity of the trend and speed of the change in
attitudes about intimate partner violence suggest that global cultural diffusion has played an
important role.

attitudes, development, globalization, social change, social norms

A large literature explores the role of global world society on individual attitudes. This
norms in defining appropriate action for nation study examines trends in attitudes about inti
states (Meyer et al. 1997). Most studies focus mate partner violence; this form of violence
on changes to national policies in response to against women is commonly defined as "any
the transnational normative pressures of world behavior within an intimate relationship that
society. This study extends world society the causes physical, psychological, or sexual
ory by examining the diffusion of global norms harm to those in the relationship" (Heise and
to nonelite individuals. Normative global Garcia-Moreno 2002:89). It often occurs in
scripts, such as human rights, are not solely private as part of personal relationships and,
available to national governments. To the con as such, may be particularly resistant to exter
trary, global norms also provide moralized nal pressures, which makes it a good case for
guidelines for individual behavior (Meyer and examination of global influences. Also, norms
Jepperson 2000). To capture the extent of world
society's influence, it is important to go beyond "University of Michigan
studies of policy change.
Corresponding Author:
Recent transnational advocacy and devel
Rachael S. Pierotti, Population Studies Center,
opment programming on issues of violence University of Michigan, 426 Thompson Street,
against women provide an interesting and Ann Arbor, MI 48106-1248
important case for examining the influence of E-mail:

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
Pierotti 241

about interpersonal relationships (as opposed (NGOs), including advocacy groups, profes
to norms about institutional arrangements sional associations, and charities with global
such as the structure of education systems)reach (Boli and Thomas 1997; Merry 2006).
World society actors are influential because
are relevant to all individuals, irrespective of
social position. Finally, rapid dissemination of their ability to define and promote new
of global norms about violence against cultural scripts and models of policy and
women began in the mid-1990s and acceler activism (Keck and Sikkink 1998). Global
ated in the first decade of the 2000s. This cultural scripts draw on fundamental princi
timeframe means one can use recent survey ples of world society to define appropriate
data to examine whether attitudes became behavior for individuals, organizations, and
more consistent with world society's cultural nation-states (Boli and Thomas 1997; Meyer
scripts as those scripts were disseminated.and Jepperson 2000).
If dissemination of global norms is affectScholars have documented evidence of the
ing individuals worldwide, the proportioninfluence
of of global models on a variety of
people who view intimate partner violencenational
as policy domains: human rights
legitimate should be declining. I test this(Hafner-Burton and Tsutsui 2005; Koo and
hypothesis using nationally representative Ramirez 2009), democracy (Torfason and
data on women's attitudes in 26 countries,Ingram
all 2010), environmental protection
the countries for which relevant data were (Frank, Hironaka, and Schofer 2000), educa
available at two time points. I then combinetion (Meyer, Ramirez, and Soysal 1992), pop
demographic theories and methods withulation (Barrett and Tsui 1999), women's
world society theories about the disseminalabor force participation (Berkovitch 1999a),
tion of cultural scripts to derive and testwomen's voting rights (Ramirez, Soysal, and
explanations for aggregate-level trends inShanahan 1997), women's political participa
attitudes about intimate partner violence. tion (Paxton, Hughes, and Green 2006), gen
I find that within the first decade of the der mainstreaming (True and Mintrom 2001),
2000s, women in 23 of 26 countries became female circumcision (Boyle and Preves 2000),
more likely to reject intimate partner violence as and criminal regulation of sex (Frank, Camp,
a justifiable form of social control. Moreover, and Boutcher 2010). In each of these domains,
this aggregate trend was due to a rapid access to global models proves a powerful
increase in the likelihood of rejecting intimate predictor of national policy action. Moreover,
partner violence among women of all ages, dissemination of global cultural scripts explains
women with various levels of education andincreasing worldwide isomorphism in these
access to media, and women in urban and policy domains.
rural areas. This suggests the change wasThe analysis presented here pushes beyond
mostly due to the wide dissemination of existing studies of policy change by testing
global cultural scripts. Structural socioeco
whether individual-level attitudinal change is
nomic and demographic shifts in the populaalso consistent with the influence of global
tion account for little of the observed trend. cultural scripts. Case studies from around the
world demonstrate the complexities of when
and how global norms are adopted, resisted,
modified, and rejected by local actors work
World society theory holds that the ingglobal
within their own cultural logics (Boyle
arena is more than political and economic Levitt and Merry 2009; Rubtsova
exchanges among states (Meyer et al. 1997).These studies add great depth to our
World society is a distinct social spaceknowledge
consti of the social processes that accom
tuted by interactions among transnational
pany the introduction of new cultural scripts.
What the case study approach cannot detect,
actors, especially intergovernmental organi
zations, such as the United Nations, andhowever,
inter are broad trends that correspond
national nongovernmental organizations with the articulation and diffusion of global

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
242 American Sociological Review 78(2)

Global Cultural Scripts about

Violence against Women
International conferences

Donor pressure and incentives
Treaties and global normative pressures
NGO and activist mobilization
Donor-funded programming
NGO outreach campaigns
Enhanced media coverage National Policies that Address
Public discussions
School curricula Violence against Women
Interpersonal communication

Media coverage of policy debates
Government initiatives
Policy (threat of) enforcement

Individual Attitudes about

Violence against Women

Figure 1. Pathways of Influence: Global Cultural Scripts to Individual Attitudes

cultural scripts (Frank, Hardinge, and Wosick Figure 1 depicts the theorized pathways of
Correa 2009). That is the goal of this research.influence linking global cultural scripts with
If world society norms exert power at theindividual attitudes. Diffusion mechanisms
individual level as they do at the nationalconnect global cultural scripts directly to
policy level, we would expect the diffusion ofindividuals and operate indirectly through
global cultural scripts to produce relatively national political actions. The dramatic
rapid changes in individual attitudes. growth in the number of women's interna
Demographic research, especially studiestional NGOs and their efforts at cultural diffu
of changing family patterns, has long consid sion are largely credited with linking the
ered global ideational influences on individualglobal and the local (Berkovitch 1999b). In
level attitudinal and behavior change (Cleland this model, both national policy and individ
and Wilson 1987; Jayakody, Thornton, andual attitudes are affected by the diffusion of
Axinn 2008; Lesthaeghe 1983; Mason 1997;global cultural scripts. Attitudinal changes
National Research Council 2001; Thornton may be bolstered by policy action, but they do
2001, 2005; Yount and Rashad 2008). The not depend on national legal changes.
demographic literature describes mechanisms I simplified the model to emphasize the
for the diffusion of global family models,object of this study, and this requires some
including urbanization, education, and mass clarification. First, I build on the research
media, along with NGO outreach, governdescribed earlier that examines the influence
ment initiatives, development programs, and of global models. The focus on vertical diffu
religious institutions (Thornton 2005). Schol sion is not meant to deny that both policy
ars have also documented the role of thesemakers (Chimbwete, Watkins, and Zulu 2005)
mechanisms in disseminating messages aboutand individuals (Boyle and Carbone-Lopez
women's rights (Berkovitch 1999a; Merry2006; Htun and Weldon 2012) have agency in
2006). I bring together world society theoryshaping the dissemination and interpretation
and demographic theory about the diffusionof global cultural scripts. Second, the present
of cultural models to derive hypotheses aboutstudy does not include direct measures of
aggregate-level trends in attitudes about intiindividual links to world society and does not
mate partner violence. test the relative importance of each of these

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
Pierotti 243

mechanisms. These mechanisms operate dif violence against women, but the current influ
ence of global cultural norms cannot be
ferently in different places. The effect of
reduced to the unilateral imposition of West
laws, for example, will vary depending on the
socio-legal context and the salience of the ern values. Dissemination of global cultural
formal legal system in people's lives (Long scripts depends on the participation of activ
hofer and Schofer 2010). Testing these mech ists all over the world who vernacularize the
anisms is outside the scope of this study, butglobal
I models into language that makes sense
discuss them because they inform the hypoth in each local context (Merry 2006).
eses presented below. The dominant definition of violence
against women is enshrined in transnational
agreements and national laws. In 1979, the
General Assembly of the United Nations
adopted the Convention on the Elimination of
VIOLENCE AGAINST WOMEN All Forms of Discrimination against Women
(CEDAW), which demanded changes not
only in states' practices, but also required
In the community of transnational women's
rights activism, violence against womenstates to ensure private entities did not dis
is con
ceptualized as a human rights violationcriminate
that against women (Thompson 2002).
results from and reinforces systems of This declaration opened the door to discus
inequality (United Nations 2008). Violence
sions about violations of women's rights in
against women encompasses social practices
the private sphere.
such as intimate partner violence, sexual Thevio
debate about violence against women
lence, sexual harassment, female circumcision,
found its full expression at the 1994 Interna
tional Conference on Population and Develop
early or forced marriage, and human trafficking.
Depicting violence against women as a ment in Cairo and the 1995 Fourth World
Conference on Women in Beijing. At the Cairo
category of meaning facilitates the interpretation
of each social practice through the lens of gender gender equality and women's rights
were promoted
inequality and women's rights. For example, in on an international stage as key
the case of intimate partner violence,components
advocacy of improving reproductive health.
messages declare that wife beating isThe Beijing conference built on the success of
a rights
violation that is never justified. Cairo and produced a Platform for Action pro
The definition of violence against claiming
women that violence against women is a vio
lation of women's human rights (Tinker 2004).
is a cultural product; it is based on fundamen
tal principles of world society, includingIn 2008,
uni United Nations Secretary-General
versalism, liberal individualism, and Ban Ki-moon launched a campaign called
(Berkovitch and Bradley 1999; Boli andto End Violence against Women. The
campaign website called violence against
Thomas 1997). National and international
elites are encouraged to support the new women a "global pandemic" and quoted the
and institutional structures because of the Secretary-General saying, "there is one uni
association between human rights and gender versal truth, applicable to all countries, cul
tures and communities: violence against
equality on the one hand, and modernity and
development on the other hand (Berkovitch women is never acceptable, never excusable,
1999a; Luke and Watkins 2002; Thompson never tolerable" (UNiTE 2012). Global
2002; Thornton 2001). Feminist movements prominence of the issue of violence against
women has been increasing ever since the
in Western countries are an important source
of legitimating ideology and funding for 1994 and 1995 conferences in Cairo and Bei
efforts to highlight violence against womenjing. Transnational activists and policymakers
as a transnational issue. Global inequalities
share well-established cultural scripts defin
affect the construction of the definition of ing the problem, its scope, and its solutions.

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
244 American Sociological Review 78(2)

Figure 2. Worldwide Cumulative Number of Policy Actions That Address Intimate Partner

Evidence of the worldwide influence of reforms is noteworthy. The trend suggests

global cultural scripts about violence against
that national governments across the globe
women is seen in patterns of national legalresponded to pressures and incentives to
change. One study of rape law reform found address violence against women.
that global forces propelled a wave of recent All countries included in the present study
changes to rape laws (Frank et al. 2009). The implemented legal or policy changes since
changes reflected global norms that catego 1995 to prohibit, prevent, or punish violence
rize rape as a violation of individual liberty,
against women. Twelve countries enacted new
rights, and equality. Another study credited legislation specific to addressing intimate
global forces with inducing national governpartner violence: Bolivia (1995), Cambodia
ments to criminalize female circumcision
(2005), Dominican Republic (1997), Egypt
(Boyle and Preves 2000). (1998), Ghana (2007), India (2006), Indonesia
I gathered information from the United
(2004), Malawi (2006), Nepal (2009), Turkey
Nations Secretary-General's online database
(1998), Uganda (2009), and Zimbabwe
on violence against women to examine (2007). Another 10 countries passed legisla
whether there is also a global pattern in the tion or amended existing legislation regarding
enactment of national policies addressing other forms of violence against women: Arme
intimate partner violence.1 The findings, pre nia (2003), Benin (2003, 2006), Ethiopia
(2004), Kenya (2006), Madagascar (2000,
sented in Figure 2, are striking. Since 1975,
119 different countries worldwide have 2005), Philippines (2004), Rwanda (2008),
enacted approximately 260 national-level Senegal (1999), Tanzania (1998), and Zambia
legal changesnew legislation, amended (2005). In Haiti (2008), Jordan (2004), Mali
legislation, executive decrees, and constitu
(2006), and Nigeria (2007), national govern
tional provisionsto address intimate partner
ments have gotten as far as issuing policies
violence. Nearly 95 percent of those changes
regarding violence against women.
occurred since the 1995 Beijing conference.2In summary, during the late 1990s and the
Worldwide similarity in the timing of these
first decade of the 2000s, cultural scripts that

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
Pierotti 245

identified violence against women as a humanIn world society, the consensus is that a
rights violation became increasingly institu
husband's physical abuse of his wife is part o
tionalized within world society. A variety of
a broader social system that reinforces gende
world society actors, including NGOs inequality.
and For individual survey respondents
intergovernmental organizations, dissemi however, the meaning of violent acts may be
nated this cultural script. The dramatic expan
affected by more immediate concerns, such a
sion of policies addressing violence against the interactional context and individual
women is evidence of the broad diffusion and behaviors (Heise 1998). If individuals are
national-level influence of this global cultural unaware of the global discourse about vio
script. In the analyses that follow, I examine lence against women or find it irrelevant to
whether individual attitudes also became more their lives, there is no reason to anticipate
consistent with global cultural scripts during consistent trends in attitudes about intimate
this time period. In so doing, I test whether partner violence.
world society theory can be usefully applied On the other hand, global norms may have
to understanding trends in individual attitudes. symbolic power, independent of national gov
ernments' translation of those norms. National
and international discourse may be a cultural
resource, enabling individuals to change how
they perceive and evaluate violent acts. As
The analysis has three parts. I beginLevitt and Merry (2009:447) note, "that wom
by inves
tigating patterns in attitudinal trends. Second,
en's rights are codified into law endows them
I test mechanisms of diffusion of with
global cul and connects them to the state,
tural scripts as predictors of individual atti few women turn to formal legal
even though
tudes in cross-sectional analyses. arenas
Finally, I
to redress their grievances." Local elite
examine several potential explanations forgain
activists thelegitimacy and access to resources
aggregate trends. In this section,byI using
presentglobal norms to justify their work
(Tsutsui and Shin 2008); these activists then
hypotheses for each part of the analysis.
further disseminate global cultural scripts.
Moreover, access to global norms about
Aggregate Trends
violence against women is facilitated by the
First, I examine whether individual attitudes
work of many types of actors. Education sys
are becoming more consistent with tems, media, international donor program
ming, NGOs, and religious organizations also
scripts of world society. Despite international
elites' and national policymakers' heightened
disseminate global cultural scripts. NGOs and
attention to violence against women,development
it is con programs distribute printed
ceivable that global cultural scripts
never sponsor public discussions, and
offer training
reach nonelite women who are not engaged in to local professionals and com
transnational activism. Existing munity
groups on issues of gender equality
and violence against women. For example,
leads to skepticism regarding individual
access to global cultural scripts. World soci
during 2005 to 2007, a United States Agency
ety scholars have documented decoupling
for International Development (USAID) wom
between national policies based on global
en's rights program in Benin printed and dis
cultural models and implementation tributed
of those
15,000 booklets explaining women's
rights, placed
policies (Meyer et al. 1997). Furthermore, in 2,000 posters about women's
legally pluralistic societies, such asrights
many in of
city halls, produced films on wom
the countries in this study, the formal legal
en's rights and broadcast them to an estimated
system or transnational principles 20,000
of human
individuals, reached nearly 7,000 peo
ple through public meetings in rural commu
rights may not be salient cultural frameworks
in people's everyday lives. nities, and trained local paralegals, health

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
246 American Sociological Review 78(2)

often sites of social

professionals, civil authorities, and change (Ryder 1965).
(USAID Women's Legal Rights
Residents Initiative
of urban areas are more likely to be
familiar with
2007). This example is only one global
of norms
manybecause of their
development programs relative
to the elite
past organizations
ade. and individuals who disseminate those ideas.
Also, people
As described earlier, the in urban areas
diffusion oflive in socially
heterogeneous environments
cultural scripts about violence against and are exposed
began in earnest in the to alate 1990s
wide variety of life and
styles in intensi
terms of social
fied throughout the class,
2000s.cultural With
and religiousthe wide
orientations, and
political ideologies.
spread diffusion of global cultural Access to education and
scripts, I
hypothesize that individual
media has similarattitudes
effects of expanding are
becoming more consistent
zons andwith
exposure to global scripts
of world society: (Hornik and McAnany 2001). In many coun
tries, messages about women's rights, gender
Hypothesis 1: Within each and problems
country, of violenceare
women against
more likely to reject women are explicitly
intimate discussed invio
partner school cur
lence in the mid- to late-2000s than earlier ricula and in media programming.
in that decade.

Hypothesis 2: Individuals with greater access

After testing this hypothesis, additional sensi to global discourse through urban residence,
tivity analyses will assess whether attitudinal education, or access to media are more like
trends were similar among married and ly to reject intimate partner violence than
unmarried women, and among men. those with more limited access to the global

Mechanisms of Diffusion: Predictors

of Attitudes about Intimate Pariner Explanations for the Aggregate Trend
From theories of diffusion, demographers have
proposed two explanations for aggregate
World society theory predicts that people
level trends in attitudes: attitudinal shifts
most closely connected to the global institu
tional environment will be most likely within
to existing population groups and changes
reflect the values promoted in that environin population composition (Firebaugh 1992).
ment (Meyer 2010). Previous research in subChanges within population groups are consis
Saharan Africa on attitudes about intimate tent with widespread diffusion of cultural
partner violence confirms this prediction: onmodels and rapid aggregate-level change.
average, people in urban areas, those with
Population compositional changes occur as
some subgroups grow in relative size and
more education, and those with greater access
to media are more likely to reject intimate become a larger proportion of the population.
partner violence (Uthman, Lawoko, and Attitudinal trends due to population composi
Moradi 2009). If the dissemination of globaltional shifts are the result of more narrow
norms about violence against women is at cultural diffusion and should produce more
least partly responsible for individual attigradual change. Analyses of the demographic
tudes, we should find a cross-sectional assoprocesses driving attitudinal change thus pro
ciation between access to messages emanatingvide further tests about whether these trends
from the world cultural environment and are consistent with the widespread influence
attitudes about intimate partner violence. of global cultural scripts.
I test the influence of three mechanisms of I take two approaches to understanding
global norm diffusion: urban residence, edu attitudinal trends. First, I investigate the
cation, and access to media. Urban centers are extent to which changes in population

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
Pierotti 247

compositionin terms of urban living, edu Hypothesis 3: Within-population group attitu

cation, and media accessexplain the change dinal shifts explain most of the change in
in attitudes about intimate partner violence. the percent of women who reject intimate
Second, I examine how much of the observed partner violence. Increased urban living,
trend is due to attitudinal changes within birth education, and media access (compositional
cohorts measured in both survey waves ver change) explain only a small part of the ag
sus change due to replacement in the survey gregate trend.
population of older cohorts by younger
cohorts who are more likely to reject intimate Within-cohort change and cohort
partner violence. replacement. Social scientists often look to
cohort replacement as an explanation for
Increasing urban living, education,
social change (Alwin 1990; Lesthaeghe and
and media access. Assuming that tests of
Surkyn 1988; Ryder 1965). Cohort replace
Hypothesis 2 show that people in urban areas ment suggests that social change occurs as
with more education and media access are younger generations with distinct beliefs and
values replace older generations in the popu
more likely to reject intimate partner violence,
the next question is to what extent changes lation. A main assumption of cohort
in population composition account for the replacement theory is that individuals' atti
aggregate-level trend in attitudes about intitudes are formed early in life and remain
relatively stable over their life course.
mate partner violence. Most of these countries'
populations are becoming more urban andAn alternative model of social change
have increasing access to education andassumes that individuals change their atti
media. Some of the observed changes in atti tudes as adults (Alwin 1990). That adults may
tudes, therefore, may result from changing change their opinions is the operating assump
population composition. tion of activists and development practition
Although I expect changes in urban living,ers who expend resources attempting to
education, and media access to explain some change how individuals think about violence
against women. This is within-cohort change
of the attitudinal change, I anticipate that the
broad dissemination of global cultural scripts
and it may result from agingas respondents
get older their opinions changeor from
produced a shift in attitudes that is greater and
more rapid than what is explained by popula
period effectsif there is something different
about later periods that causes people to
tion compositional changes. Urban living,
education, and media access increase expo
change their opinions.
sure to global norms, but these are not the Transnational dissemination of global cul
tural scripts about violence against women
only ways of gaming access to global cultural
scripts. Messages are also disseminated influences both the context in which new
through personal networks and diffused generations are socialized and the cultural
through many other mechanisms by agents of context in which already socialized adults
world society. These efforts to diffuse global make meaning of their actions and interac
norms have the potential to lead to attitudinal tions. I hypothesize that both types of change
change within population groups defined by will be operative, but within-cohort change
place of residence, education, and media will explain more of the observed trend. If the
accesschange that is above and beyond cause of the aggregate-level trends were lim
shifts due to differences in population compo ited to cohort replacement, it would mean
sition. Tests of this hypothesis will evaluate either that only young people have exposure
whether patterns of attitudinal change are to global cultural scripts on violence against
consistent with relatively broad dissemination women (perhaps through school), or that only
of global cultural scripts. young people are receptive to those messages.

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
248 American Sociological Review 78(2)

corresponds with the period

Given the variety of mechanisms of increasing
of diffusion
global attention
of global cultural scripts, to issues ofpeople
I expect violence against
all ages to have access women.4
to those scripts. And,
while it is conceivable that older women are Sampling for each DHS study was con
somewhat more resistant to external cultural ducted by randomly selecting primary sam
influences, global cultural scripts should be pling units (PSUs) and then randomly
powerful enough to influence people during sampling households within selected PSUs. In
all periods of the life course. most countries, all women between the ages of
15 and 49 years were eligible to participate. In
Hypothesis 4: Within-cohort attitudinal shifts Egypt, India, Indonesia, Jordan, Nepal, and
explain most of the change in the percent Turkey, only ever-married women were eligi
of women who reject intimate partner vio ble.5 This inconsistency in sampling is one
lence. Cohort replacement explains only a reason results are comparable within country
small part of the aggregate trend. and across time but not across countries. All
analyses use sampling weights to adjust for
variation in the probability of selection.

Dependent Measure
Data for this study come from 52 Demographic
and Health Survey (DHS) datasets, including
The outcome variable derives from a question
two from each of 26 countries.3 All countries
that asked respondents whether it is okay for
with two waves of DHS data on women's a man to hit or beat his wife under certain
attitudes about intimate partner violence are Specifically, the most com
included in the analysis. These countries areform of the question asked, "Sometimes
Armenia, Benin, Bolivia, Cambodia, Dominican
a husband is annoyed or angered by things
which his wife does. In your opinion, is a
Republic, Egypt, Ethiopia, Ghana, Haiti, India,
Indonesia, Jordan, Kenya, Madagascar, Malawi,
husband justified in hitting or beating his wife
Mali, Nepal, Nigeria, Philippines, Rwanda,
in the following situations?" The five scenar
Senegal, Tanzania, Turkey, Uganda, Zambia,
ios presented to respondents were (1) if she
and Zimbabwe. Half of the countries with goes out without telling him, (2) if she
data meeting the inclusion criteria are in neglects
sub the children, (3) if she argues with
Saharan Africa. As such, inclusion or excluhim, (4) if she refuses to have sex with him,
and (5) if she burns the food. Given the sensi
sion of variables and models from the analyses
is largely driven by their relevance in tive
the nature of this survey question, differ
African context. Nonetheless, although notences in question wording have the potential
to influence the distribution of responses.
representative of the globe, these data come
Based on research by Yount and colleagues
from a wide range of low- and middle-income
countries. (2011), I conducted an analysis of question
The datasets come from nationally repre wording and determined that small differ
sentative, repeated cross-sectional surveys.ences
In across surveys in eight countries could
not account for the observed trends.6
each country, the first wave of data collection
was in the early- to mid-2000s and the second
Moreover, the findings presented below do
was in the mid- to late-2000s. For most coun not depend on inclusion of the countries with
tries, the two waves of data collection were question wording differences.
five years apart. The shortest interval between There is some variation between countries,
waves was three years in Egypt and the long however, in the scenarios presented as poten
est was seven years in India. Table 1 shows tial justification for intimate partner violence.
survey years and sample sizes for each of the To ensure comparability within country and
26 countries. The timing of these data collections across time, only scenarios that were asked in

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
Pierotti 249

Table 1. Survey Years and Sample Sizes

Wave 1 Wave 2

Year Sample Size Year Sample Size

Armenia 2005 6,566 2010 5,922
Benin 2001 6,219 2006 17,794
Bolivia 2003/04 17,654 2008 16,939
Cambodia 2005/06 4,168 2010/11 18,754
Dominican Rep. 2002 23,384 2007 27,195
Egypt 2005 19,474 2008 16,527
Ethiopia 2000 15,367 2005 14,070
Ghana 2003 5,691 2008 4,916
Haiti 2000 10,159 2005/06 10,757
India 1998/99 90,303 2005/06 93,724
Indonesia 2002/03 29,483 2007 32,895
Jordan 2002 6,006 2007 10,876
Kenya 2003 8,195 2008/09 8,444
Madagascar 2003/04 7,949 2008/09 17,375
Malawi 2000 13,220 2004/05 11,698
Mali 2001 12,849 2006 14,583
Nepal 2001/02 8,726 2006/07 8,640
Nigeria 2003 7,620 2008 33,385
Philippines 2003 13,633 2008 13,594
Rwanda 2000 10,421 2005 11,321
Senegal 2005 14,602 2010/11 15,688
Tanzania 2004/05 10,329 2010 10,139
Turkey 1998 6,152 2003/04 8,075

Uganda 2000/01 7,246 2006 8,531

Zambia 2001/02 7,658 2007 7,146
Zimbabwe 1999 5,907 2005/06 8,907

both waves of data collection were included that indicates whether the respondent broadly
in the dependent variable. The dependent rejects wife beating. This variable is coded 1
for respondents who rejected all scenarios
variable should not be compared across coun
tries, however, because the scenarios are notsuggested by the survey interviewer as rea
the same in each country. The standard sce sonable justification for a husband to beat his
narios (listed earlier) were used in the depend
wife and 0 otherwise.8 For the sake of brevity,
ent variable for 20 of the countries.7 I will interpret this throughout the analysis as
Examining trends in responses to each a rejection of intimate partner violence. Note
that this is an over-simplification; some
scenario individually provides interesting
information about shifting criteria for the respondents may reject all of the suggested
establishment of just cause for violence justifications but may believe there are other
instances in which a husband is justified in
(Heise 1998). In this analysis, however, I am
interested in the degree to which attitudes beating his wife. Across the 52 datasets, an
conform to global norms about violence average of 51 percent of respondents rejected
against women, which emphasize rejectionall ofscenarios. To ensure results are not an arti
fact of the dependent variable coding, I also
all justifications for violence. For that reason,
I created a dichotomous dependent variable conducted analyses using an additive scale (0

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
250 American Sociological Review 78(2)

to 5) for the number of scenarios

Pritchett that
(2001), I used principal a
respondent deemed reasonable justification
analysis (PCA) to construct wealth indices for
for intimate partner each survey sample.
violence. Given the available
Results (not
measures, however,
shown) are largely consistent the resulting
across specifi index was
cations of the dependent correlated with urban living. Among
the African countries, the correlation between
the wealth score and urban residence ranged
Predictors of Attitudes about Intimate
from a low of .55 in Benin to a high of .86 in
Partner Violence
Ethiopia. Because the wealth index is not a
The analyses require measures of several great measure, and because it is highly cor
demographic characteristics. Urban resi related with urban living, I ultimately dropped
dence, educational attainment, and mediait from the multivariate analyses.
access are the three main variables of interest. Religious leaders and organizations are a
Urban living is a dichotomous variable common source of cultural scripts, so I
defined by the survey teams in each country. included a dichotomous variable to indicate
Educational attainment is coded as a categor whether a respondent identified as Muslim.
Empirical evidence suggests there are sub
ical variable that captures the highest level of
schooling attended by the respondent: no stantial cultural differences between Muslim
schooling, primary, secondary, or higher. and non-Muslim societies on issues of gender
Media access is a composite measure of equality (Norris and Inglehart 2002). I
access to newspapers, radio, and television. included this as a control rather than a varia
Respondents who came in contact with news ble of primary interest, however, because
papers, radio, or television at least once per religious context varies substantially across
week were given a code of 1 for access to the countries in this analysis and because it is
media. Respondents with less frequent access beyond the scope of this article to examine
were coded 0. the complex relationship between religion
In addition to the main predictors, I tested and cultural scripts about gender.
several other variables for a relationship with A final set of control variables measure
attitudes about intimate partner violence.women's status within marriage, which may
First, I included continuous measures of age also be associated with attitudes about inti
and age-squared in each model as controls.mate partner violence. Women who have
The analysis of cohort replacement closelynever married may be more independent and
examines the effect of age on attitudes aboutless tolerant of intimate partner violence than
intimate partner violence. I also investigatedtheir married peers. In countries with data
the role of wealth. Wealth may increase accessfrom all women, I coded women who have
to national and international elites who par ever been married as 1 and those who have
ticipate in global culture and provide means never married as 0.
for social engagement outside the home. Women who are relatively dependent on
Wealth may therefore serve as an additional their marital relationship for their livelihoods
mechanism of diffusion of global culture. may be more tolerant of abuse than women
Given the populations under study, it is with more power in their relationship. Previ
difficult to measure wealth, and availableous research in low-income countries has
measures indicate a large amount of clumpingfound that women who marry at young ages
on the poor end of the wealth scale, especiallyand women who have fewer years of school
in Africa (Vyas and Kumaranayake 2006).ing than their husbands are more likely to
The DHS does not include income or expend justify intimate partner violence (Kishor and
iture data; instead, the surveys collect infor Subaiya 2008; Yount 2005; Yount and Li
mation on household characteristics and 2009). Women who marry young generally
household assets. Following Filmer and
complete fewer years of school, have less say

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
Pierotti 251

in choosing their spouse, and have less timethere was a statistically significant change
to form a sense of self outside of their marital
between the first and second waves in the pro
relationship. Women with lower levels of portion of respondents in each country who
education than their husbands may have rela reject intimate partner violence.
tively lower bargaining power within their
relationship, making them more dependent Hypothesis 2. To test Hypothesis 2, I
and more likely to justify intimate partner examined predictors of rejection of intimate
violence.9 partner violence. I used chi-square tests to
I treated women's age at marriage as a examine the bivariate relationship between
categorical variable, dividing ever-married each predictor variable and rejection of inti
women into those who were first married at mate partner violence (results available from
age 15 years or younger, those married author). I then tested the multivariate relation
between age 16 and age 19, and those married ship between the predictor variables and
at age 20 or older. I tested several specifica rejection of intimate partner violence using
tions for capturing educational gaps between logistic regression.
wives and husbands. A variable with three
categorieshusband and wife have the same Hypothesis 3. The next part of the analy
sis examines the extent to which aggregate
level of education, husband attended a higher
level of education, wife attended a higher level trends in attitudes about intimate partner
level of educationcaptured the most meanviolence are explained by shifts in population
ingful variation among respondents. A sepacomposition. I used direct standardization to
rate variable for husbands' education was examine how much of the change over time
coded the same way as the respondent's wasedu due to increases in urban living, educa
cational attainment variable. tional attainment, and media access. Direct
The weighted distribution of each of the inde standardization allows for estimation of a
pendent variables by country and wave, as wellcounterfactual scenario. What percent of the
as results of adjusted Wald tests for significant population in wave 2 would have rejected
change between survey waves, are available in intimate partner violence if attitudes about
Tables Sla and Sib of the online supplement intimate partner violence had remained
( unchanged but the population composition
Unfortunately, direct measures of individual had shifted? For each survey wave in each
access to global cultural scripts, or the influ country, I divided the population into sub
ence of those scripts, are not available in largegroups by urban/rural residence, highest level
national samples from many countries. There of education attended, and media access/no
fore, I cannot test the direct influence of world access. This resulted in a distribution of
society on individual attitudes. As detailed in respondents across 16 subgroups. For each
the previous section, however, I have the datasubgroup, I estimated the percent who rejected
to investigate whether world society theory intimate partner violence (rate of rejection).
may be applied to explain the observed trends Standardization showed whether the percent
in attitudes about intimate partner violence. The who reject intimate partner violence in the
next section describes how I test each of the population would have changed if the rate of
hypotheses described earlier. rejection for each subgroup had remained the
same. So, for each country, I multiplied the
subgroup distribution from wave 2 by the
rates of rejection from wave 1.

Hypothesis 4. The final part of the analy

Hypothesis 1. The first part of the analysis
describes the trends in attitudes about intimate
sis tests how much of the change in population
partner violence. To examine Hypothesis 1,1 on the dependent variable can be
explained by within-cohort change and how
used adjusted Wald tests to analyze whether

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
252 American Sociological Review 78(2)

much can be explained by

within cohort
a span replace
of only five years. Figure 3
ment. To do this, I used linear
shows the regression
proportion of women who reject all
decomposition (Firebaugh 1989). for This
partner violence in
estimates change due toeachcohort
survey wave replacement
for each country: the top
net of within-cohort change (the
bar is the percent in combined
wave 1 and the middle bar
is the and
effects of age and period) percent in wave
vice 2. (TheFor
versa. standardized
each country, I pooled wave
the2 twobar will be discussed
waves below.) The
of data
and regressed my outcome
change isvariable
statistically on
survey at = .05 in
all countries.
year and respondent's birth year:In 23 out of 26 countries, there
is a significant increase in the proportion of
y = bQ + survey year + b2 birthrejectingyear
intimate+ partner
e violence.
Twelve of the countries saw more than a 10
The slopes from thispercentage
regression equation
point increase in the rate of rejec
represent the expectedtion. Nigeria hadin
change the largest
the change,
percent with a
of respondents who reject point increase
intimate in the propor
violence (y) for everytionone-year
of women who rejectincrease in
intimate partner
survey year or meanviolence,
birth followed by Zambia,
year. For Kenya,
country, I multiplied theand Armenia with
slopes approximately
from the
regression by the actual change point
survey in the rejection
and mean birth year.10
of intimateThepartnerelapsed
violence. time
between survey years and India, Bolivia,
the and difference
Ethiopia had the small
est increases in rates of rejection,
mean birth year was approximately five years but the
for each country. Thischanges are all still statistically
produced estimates significant
and in
the within-cohort and the expectedreplacement
cohort direction. Indonesia,
components of the change and the
Madagascar are clear outliers.
percent who
reject intimate partnerThose three countries had a statistically sig
nificant decrease in the proportion of women
Within-cohort changewho
= reject intimate partner
b^ (wave 2 violence.
In the online
survey year - wave 1 survey supplement, Table
year) S2 pre
sents responses by survey year for each sce
Cohort replacement = b2 incorporated into theyear
birth outcome measure.
in wave 2 - mean birthThe detailedin
year story is largely
wave 1)the same as the
picture in Figure 3. Twenty countries had
This method assumes that the
significant two
decreases components
in the acceptability of at
are orthogonal and that least
the effects
three are
of the scenarios linear.for
as justification I
examined the reasonableness of
intimate partner these
violence. assump
In Turkey, Uganda,
tions by testing howand closely
Zimbabwe, thethe predicted
trend appears to be more
change from the two components
tenuous. The trend is in sum to
the opposite the
actual observed change.
for three or four of the scenarios in Indonesia,
Jordan, and Madagascar.
Additional tests examined whether trends
were similar among women who had married
Aggregate Trends and those who had not, as well as among
men. In all countries where both ever- and
The data show striking changes in thenever-married
propor women were surveyed, the
tion of women in each country who observed
rejecttrend was in the same direction and
intimate partner violence. The changes are magnitude regardless of marital
of similar
impressive because of the nearly uniform
status (results not shown). There were two
waves of data available for men's attitudes
direction of the trend and their magnitude

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
Pierotti 253





Dominican Rep.













0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Wave 1 Actual Wave 2 Actual o Wave 2 Standardized

Figure 3. Percent Who Reject All Justifications for Intimate Partner Violence

about intimate partner violence for 15 of Overall,

the I find a great deal of support for
countries included in this study. As shown in
Hypothesis 1. With the exception of Indonesia,
Figure SI in the online supplement, trends
Jordan, and Madagascar, women were sub
among men mirror those among women. stantially more likely to reject intimate partner
Twelve countries had significant and sub in the more recent wave of data col
stantial increases in the percent of lection.
men These substantial changes occurred
rejecting intimate partner violence. In
a span of only five years, and the observed
Dominican Republic, there was no signifi
change was comparable among women who
cant change in the percent of men rejecting
have ever married and those who have not, as
intimate partner violence. In Indonesia and
well as among men. The trend in attitudes
Madagascar, as among women, there about
was aintimate partner violence is converging
significant decrease in the percent of menthe global cultural script that prescribes
rejecting intimate partner violence. the rejection of violence against women.

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
254 American Sociological Review 78(2)

Mechanisms of Diffusion:
were more likely Predictors
to reject intimate partner vio
of Attitudes about lence in 17 of the countries.
Intimate Partner In many of these
Violence countries the effect is curvilinear; the age
squared term is significant, indicating that the
Next, I examine whether people with greater positive effect of age diminishes among the
oldest respondents. I will discuss the effect of
access to global cultural scripts are more likely
to endorse the values embedded in those scripts.age further during the examination of cohort
I tested three primary mechanisms linking indi
replacement versus intra-cohort change.
viduals to global cultural scripts as predictors of In 14 of the 20 countries that did not
women's attitudes about intimate partner viorestrict the sample based on marital status,
lence: urban residence, education, and mediawomen who had never married were more
access. Tables 2a and 2b present results of likely than those who had married to reject
intimate partner violence. Moreover, in 12
multivariate logistic regression as odds ratios.
These models tested the significance of the countries, women who married at older ages
three mechanisms of diffusion simultaneously,were more likely than women who married at
while controlling for survey wave, Muslim age 15 or younger to reject intimate partner
violence. This indicates that women's status
identity, age, age-squared, marital status, age at
marriage, and husband's education. The first
in marriage is associated with attitudes about
intimate partner violence, even after control
coefficient in the top left of the first table means
that in Armenia, women in wave 2 had 1.99ling for key demographic characteristics.
greater odds of rejecting intimate partner vio In the final model, I eliminated the cate
lence than did women in wave 1, controlling
gorical variable capturing spousal differences
for the other factors in the model. in education and replaced it with a direct
Results presented in Tables 2a and 2b are measure of husband's educational attainment.
again remarkable for their consistency. In 22Models that include variables for spousal
countries, women who lived in urban areas educational differences indicate that, contrary
had higher odds of rejecting intimate partner to expectation, compared to women in cou
ples with the same educational level, women
violence, controlling for all other factors. The
association is not significant in Mali andin marriages with husbands who had more
Zambia and it goes in the opposite directioneducation were more likely to reject intimate
in Madagascar and Nepal. Women who had
partner violence, and women in marriages
attended at least some secondary school hadwhere they had more education than their
much higher odds of rejecting intimate parthusbands were less likely to reject intimate
ner violence in all but four countries. The partner violence. Models that include hus
effect of higher education is even larger and is educational attainment reveal that the
significant in each country. Even controlling
positive relationship between education and
for urban living and education, media access
attitudes about intimate partner violence for
is still associated with higher odds of reject
both women and men is stronger than the
ing intimate partner violence in 14 countries.
effects of relative spousal education. In nearly
Overall, each of these mechanisms for diffuall countries, a husband's educational attain
ment had a positive association with his
sion of global norms has a substantial, inde
pendent effect on the odds of rejecting wife's attitudes about intimate partner vio
intimate partner violence. This is strong lence,
sup independent of her own educational
port for Hypothesis 2: women with more attainment.
access to global discourse are more likely to
reject intimate partner violence.
Explanations for the Aggregate Trend
It is also worth noting effects for some of the
Increasing urban living, education, and
other variables in the final models. Controlling
media access. Because urban residence,
for all other factors in the model, older women

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
cd *
>> *

a IN CO O in CM CM
05 o cd ^ q rH CM

cm tJH rH rH rH rH rH

* * * # *
fl * * * * * *
cd * * * * * * * * *

T3 cd CO CO rH CO O 5* m O CD fx
q q q q o
q q in
rH rH <N rH rH rH rH CM*

* * *
* * *
* * * * *
co CO CO LO io rH CO CO rH
O tN. o o o
q q rH
05 Z CM
rH rH rH rH rH rH

* * * * * * *
* * * * * * *
#cd * * * * * * * * *
*3 LO CO cm CD CO O 5? cd LO CO rH
o o o
q 9 2 q rH
r-l rH rH rH CM rH rH rH

* * * *
* * * *
* * * * * *
LO <N CD O. o t-N in m CM
rf CO o
q rH
q q q
r-H rH rH CM C3* rH rH rH

* * * * * * *
cd * * * * * * *
a * * * * * * * *
cd CO CO 05 N O CO CO cm CM 05
q q q O Ol CM cm
q q
o rH rH rH rH rH rH rH rH rH

* * *
* * * * * * *
Oh * * * * * * * *
O in cd <N o in cm CO CM CO

2 q
cm cm CM
q q rH rH
rH r-< rH CM IS." rH rH rH rH rH

* * * * * * * * *
* * * * * * * * *
^ o <
* * * * * * * * *

>> CD CD CM o CO CO cn CD
DO in in CSJ
q CO
q . z q q cm
W rH rH rH CM rH rH rH rH

* * * * * * * *
* * * * * * *
* * * * * * * * *
P o
q q o CM
q rH

cm rH rH CM in rH rH rH rH CM*

* * * * *
X5 * * * * *
o * * * *
[X t>.
q o
s rH rH rH rH CD rH
* * *
* * * *
IX o
*> *
CM t>s fl m co
rH cm rH
q rH
q Poo
rH rH rH rH
* * * * * * * fl * 05 * * *
* * * * * * * * * * *
* * * * * * * f-i * 05 * * *
cd CD in C5 CO 05
fl rH
o o
q q q q q 05
CM II q q
CQ rH cm' o rH rH rH rH rH fl rH rH cm"

oo u
fl fl
Q5 K 3< T3
cn CD o 0
"*> ,
cd fl

X5 op 05 rfl _: ^
2 ffl ?t5
cd CO ^ <H -O T3

^ !
oo .!
g a s p s a
frt W ~ w
TJ 2 m m >
a, w X
^ ^ 00 00 03 ao 3 Cl, co X
Table2.PredictorsfRejctionfItimaePrtneViolenc(OdsRatios) S 2 < < z < X Note:DR=mincaRepubli.PrmayscholeduatinwsedathrfencgoupfrAmeniabcusveryfwspondethaoductin. *p<.05;*p<.01;* p<.01(two-tailedts ).


This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to

05 N M Ifi
rH rQ

O M 03 O) ^
N N O ^ O Ol Ol


0 * * <?
M O) O) <


lO O rH


03 COOrHLOCDt^OO co CO 03 O CM
Nco^cocoorom eg O CO CD



.tn CD

rH <N r-H

a o O 05 CO O CO CM
tv. o o tx o co o
3? Z
r-l rH rH rH rH

Ph *
o ^
* rf1
ffilOCDOlNOON f-4 O
CD ^
P fl

fn 03 05 rH 03 O ^
II " '
_0 rH rH
(30 O
cd 3 T5 *

03 *
" IfiOCOOl-tlflOlri 1^03^05
T3 U Cjqqinq^ooicM O 03 rH CO
cd P rH f-H r-H f-H ^0 s^ 3. ^
3 4-J
t3 cd
"d r3 * 03
o Ct3 *f-i ~ - b-o
03 a as
"2 * a r
rH -"d T3
CO pj CD B cd 03
S o xt
.9 o w
n o a
T3 2 s J .1 s -a
CO O "is h <D 3
&, w K js lt: w ojj cu rH M J Ph W K
Table2.PredictorsfRejctionfItimaePrtneViolenc(OdsRatios) w 2S<<Z a *p<.05;*p<.01;* p<.01(two-tailedts ).


This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
Pierotti 257

education, and media access are strongly

portion of the observed change between
associated with attitudes about intimate part
waves 1 and 2. Turkey is the exception, where
ner violence at the individual level, it ispopulation compositional changes account
important to examine how much of the aggre for two-thirds of the change in attitudes. Oth
erwise, population compositional changes
gate-level trend is explained by changes in the
population composition on these characterisaccount for one-third of the change in Ethio
pia and India, one-quarter of the change in
tics. A first indication of the role of population
compositional changes is found in Tables 2aArmenia and Tanzania, and 20 percent or less
and 2b. The coefficient on wave 2 remains of the change in all other countries. This pro
statistically significant in every countryvides support for Hypothesis 3: increasing
except for India and Turkey, despite inclusionurbanization, educational attainment, and
of a number of demographic predictors. Thismedia access explain some, but only a small
is evidence that urbanization, increasing eduportion, of the change in the percent of women
cational attainment, expanding media access,who reject intimate partner violence. This
and improvements in women's status in marfinding is consistent with widespread diffu
riage do not explain all of the change in attision of global cultural scripts through a vari
tudes about intimate partner violence. ety of mechanisms.
I further tested the magnitude of the effect
of compositional changes by estimating a Within-cohort change and cohort
counterfactual: the percent of women whoreplacement. The analysis of within-cohort
would have rejected all justifications for viochange versus cohort replacement examines
lence in wave 2 if rates of rejection of inti how much of the observed change is due to
mate partner violence had remained the samechanges within the birth cohorts represented in
as they were for each population group both waves of data collection versus change
in wave 1. As described earlier, I defined due to entrance into the population of new
population groups by urban/rural residence,cohorts of women with different attitudes than
level of education, and media access because their predecessors. As hypothesized, within
these characteristics are consistently associ cohort change has greater explanatory power
ated with attitudes about intimate partner than does cohort replacement: increases within
violence. In Figure 3, the first and second birth cohorts in the percent rejecting intimate
bars show the actual percent of women whopartner violence explain the population-level
rejected intimate partner violence in waves 1trends in attitudes. In a surprising finding, how
and 2. The third bar for each country showsever, in most countries, cohort replacement has
what the rejection percentage would havea negative effect on the percent of women who
been in wave 2 if the population composition rejected intimate partner violence.
had changed but the rates of rejection within Table 3 presents results of the regression
each population group had remained the decomposition. The left panel (columns 1 and
same. 2) presents results already displayed in Figure 3:
In almost all countries, the counterfactual
the percent of women in each survey who reject
scenario produces a rejection percentage intimate partner
that violence. The third column
is remarkably close to the actualshows wave the1 percentage-point
per change between the
cent who rejected intimate partner violence.
two time points, or the percent in wave 2 minus
The wave 2 standardized bar looks like the the percent in wave 1. Columns 4 and 5 show
wave 1 bar. This means that had there not results from the regression decomposition anal
ysis described earlier. Column 4 presents the
been within-group attitudinal changes, there
would have been little change between wavesamount (in percentage-points) of predicted
in the percent who reject intimate partnerchange due to cohort replacement, and column
violence. In most countries, population com5 shows the amount of predicted change due to
positional changes account for only a smallwithin-cohort change.

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
258 American Sociological Review 78(2)

If this linear regression intimatefits
model partnerthe
violence. Column 8
the predicted change in
shows columns
that 4 and
within-cohort change 5
in these coun
should sum to equal tries
the observed
accounts change
for more than 100 percent of the
shown in column 3. The model fits in cases observed change. Based on these results, I find
where both survey year and a respondent's support for only part of Hypothesis 4. Within
birth cohort have independent linear effects cohort change is the driving factor behind the
on the likelihood of rejection of intimate partincrease in the percent of women who reject
ner violence. The percentages in column 6intimate partner violence. Contrary to expecta
indicate how closely the predicted changetions, however, in most countries cohort replace
mirrors the actual change. Agreement is close ment does not explain any of the observed trend.
to 100 percent in almost all countries, indicat This finding relates to the somewhat sur
ing that this is an appropriate method for
prising relationship between age and rejection
decomposing within-cohort change versusof intimate partner violence. Cohort replace
cohort replacement. In India, Indonesia, Madment explains none of the observed trend in
agascar, Nepal, and Zimbabwe, the agree most countries because young women were
ment is not as strong, so the decomposition less likely than older women to reject intimate
results should be interpreted with caution." partner violence. The logistic regression mod
Finally, columns 7 and 8 show that withinels presented in Tables 2a and 2b show that
cohort change is overwhelmingly driving theodds of rejecting intimate partner violence
dominant trend at the aggregate level. Col
generally increase with age. Although this does
umn 7 shows the percent of the observednot conform to expectations that youth are
change attributable to cohort replacement andmore likely to adapt to new social norms, it is
column 8 presents the percent attributable toconsistent with the idea that age is a source of
within-cohort change. In outlier countriesempowerment for women. Women generally
that had a decrease in rejection of intimate gain self-confidence, respect, and freedom as
partner violence, cohort replacement explainsthey age. This may translate into increased
a substantial portion of the aggregate trend. Inrejection of intimate partner violence.
Indonesia and Madagascar, nearly 40 percent Within-cohort change may be due to
of the change can be attributed to cohortrespondent aging or a period effect that makes
replacement. Similarly, in Jordan, over 20
respondents of all ages in wave 2 more likely
percent of the decrease in the percent ofto reject intimate partner violence than
women who reject intimate partner violencerespondents in wave 1. I posit that the period
was due to cohort replacement. As seen ear effect attributable to dissemination of global
lier, in all other countries, women were morecultural scripts plays a major role. Evidence
likely to reject intimate partner violence infor this interpretation is provided by the fact
wave 2 than in wave 1. Of those countries,that women of all ages were more likely to
cohort replacement contributed the most toreject intimate partner violence in wave 2. In
the attitudinal trend in Ethiopia: 21 percent of
a separate analysis (not shown), I found that
the change was due to cohort replacement. in most countries, a majority of cohorts
In the other 22 countries, cohort replacementshowed a significant increase in the percent
explains almost none of the change. The percentwho rejected intimate partner violence. The
of change due to cohort replacement is negativetrend is consistent across ages. In summary,
for many of these countries: had there been only within-cohort
change is overwhelmingly
cohort replacement and no within-cohortdriving the dominant trend at the aggregate
change, the trend would go in the opposite level. Cohort replacement explains some of
direction. If cohort replacement was the onlythe negative change in Indonesia, Jordan, and
factor driving the change, a smaller percentageMadagascar, but otherwise it is not a good
of women in wave 2 than in wave 1 would haveexplanation.

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
s-S <-.
O CT CO O O O CO o o

Q S 2"
><? u

05 ,,

0 o a CO
co co ^ in co rH tv "j oi cs m co o o lO tO CD CT O O
0 CM rH rl CO CO CM I CO CO I I I rH rl CO
P o cd
.o U



3 ^
o S 0 0

u > ,,

fH in CD ^ ffl co n OinNCOONOCOCO
CD 0 h co 05 ro m t-hooctcoctcooo
n Jp-g o
o .1*3 s

1, 4- CD
n ^ oo CO CT CO in CT CO O O rH CM in CO CM CO in CO CO o CO

3.3 s CT CO
in IS. CO CM is CM in CO ts rH CO CT in CT CO rH o in CT
LO Csj CM CM O CO CM CO CM CM CM CO 1 CT CO tsl o O in CO tsl CO
S r9 rH rH r-t rH rH rH 1 1 rH rH CM rH rH rH



tn Q
I* I I I f r
""3. I I I I I I I

T3 .. ^
05 05 ,-t
> 00 ^ CT
CO o
K in N o
05 S 1
CO S ^
CO CO CM CO rH o co cm in 05 o ^ cd in ^ in cd rf
rH rH rH rH rH rH
rO 71 N
0 O u
-M C
u 0
. O
Ph ^
o ^h ts
O 0 05
CO in O
CM co
in CT rH in O CO LO CO CO CO CM o CO
CO rH o CM LO o rs rH [s in LO in CT CM CD CM CO CT in IS." co o

B *
I CO in CO in CT CO rH CD CO CO CM CO CD CM tv. LO CO CO in CM CO in

0 CD
S Id
0 d

0 0 o CT N. CO o CO CO
CD CO t-N. CO CT rH in in ts. CM CM CD CD IS. CO tv CO CO CO in CM CM

03 03

TJ cd
cd cd X!
O S, 9 > X5 cd
00 cd 03
g '# .2 O
a > O
"I 8 -h cd 03 N ^
"i s
d rl
& .2?
Table3.CohrtReplacmentversu Withn-CohrtChange
O cd
m m u
00 .
.i V ^ V 5 * - ajrco
W2222Z n< to H I 00 cd .5
D N N Note:gaivprcentsdiahtferadonlybechrtplaemndowithn-crage,thrndiatuesbointmaepr violnce wouldbemovingintheop sitedirection.


This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
260 American Sociological Review 78(2)

DISCUSSION of the 2000s, which corresponds with increas

ing global attention to issues of violence
against women
This study extends world society theory by (Merry 2006). During this
testing whether it can be usefullytime period,
applied toworld society agents sought to
reach individuals
understanding trends in attitudes among non directly through activities
elite individuals. Previous worldsuchsociety
as development programming and media
research documents the power of global Transnational advocates also pres
tural scripts and models to influence national governments for relevant pol
policymakers. Scholars have foundicy action.
that theSuch pressure led to a dramatic
common pressures of world societyincrease
explainin the number of countries that
otherwise inexplicable isomorphism in passed legislation addressing forms of vio
national policies (Meyer et al. 1997). lence against women. At the same time,
Global cultural scripts are not just relevant national political debates likely contributed to
to national policymakers; they also provide the spread of global cultural scripts.
moralized individual behavioral scripts that The near uniformity of the observed trend
can influence individuals outside the halls of and the speed of change in attitudes about inti
government. To capture the extent of world mate partner violence suggest something
society's influence, it is therefore important to occurred in the past decade that affected women
study individual attitudes and behaviors. I (and men) globally. Absent a macro-level influ
examined trends in individual attitudes and ence, there is no reason to expect such extraor
found substantial support for the hypothedinary consistency in attitudinal trends. I posit
sized influence of global cultural scripts. that diffusion of global cultural scripts about
The study examined data on individual women's rights, gender equality, and the ills of
attitudes about intimate partner violence durviolence against women was an important
ing a period of heighted transnational activ macro-level factor that influenced national pol
ism on this issue. Results show that after onlyicymakers and people at the grassroots.
five years, a significantly larger percentage of Further analysis shows that the demo
women in Armenia, Benin, Bolivia, Cambo graphic processes underlying the attitudinal
dia, Dominican Republic, Egypt, Ethiopia,trends are consistent with the proposed theo
Ghana, Haiti, India, Kenya, Malawi, Mali,retical framework that links world society
Nepal, Nigeria, Philippines, Rwanda, Senewith nonelite individuals through broad diffu
gal, Tanzania, Turkey, Uganda, Zambia, and sion of global cultural scripts. The changes in
Zimbabwe rejected intimate partner violence. attitudes about intimate partner violence
An increase in the rejection of intimate partoccurred too rapidly to be explained by struc
ner violence was evident in 23 of the 26 coun tural socioeconomic or demographic shifts,
tries studied and the size of the change was both of which are expected to occur over the
robust in 12 of those countries. For example, course of generations. Neither shifts in popu
in Nigeria, the percent of women who rejected lation composition in terms of urban living,
intimate partner violence increased from 33 educational attainment, and access to media
percent in 2003 to 52 percent in 2008. Nearly nor cohort replacement explain much of the
a 20-percentage point change within only five observed increase in rejection of intimate
years is evidence of a very rapid shift in atti partner violence.
tudes. This trend is the same among ever- and In the cross-sectional analysis, I found that
never-married women. In countries for which women with greater access to global cultural
data are available, men's attitudinal trends scripts through urban living, education, or
follow the same patterns. access to media were more likely to reject
The increase in rejection of intimate part intimate partner violence. Between the two
ner violence occurred during the first decade survey waves, some countries experienced an

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
Pierotti 261

empowerment (as discussed by Kishor and

increase in these key demographic indicators.
These population compositional changes Subaiya [2008]), this finding suggests that the
account for some of the increase in the rejec
relationship between age and intimate partner
tion of intimate partner violence, but theyviolence attitudes deserves further research.
explain only a small portion of the total This study inspires several other avenues
change. In the cross-section, attitudes aboutfor future research. First, although I examined
intimate partner violence were also associated whether trends in attitudes about intimate
with women's status within marriage. How partner violence are converging toward global
ever, changes in marital gender relations fail cultural scripts, I did not directly test the
influence of global discourse. An important
to explain the majority of the shift in attitudes
about intimate partner violence. next step is the collection of individual-level
The aggregate-level trend is produced by an data on access to global cultural scripts to
increase across socioeconomic groups in thedirectly test the influence of world society.
likelihood of rejecting intimate partner vio Second, the focus of this study was on time
lence. This is consistent with the diffusion of trends in attitudes about intimate partner vio
global cultural scripts through a wide variety lence. Data were comparable across time
of mechanisms. Cultural scripts are spread notwithin country but could not be compared
only through narrow channels of cultural diffuacross countries. Further research is needed to
sion, such as education, but also through mechinvestigate cross-national differences in levels
anisms such as development programming, of acceptance of intimate partner violence.
NGO outreach, and interpersonal networks. Analyses examining differences in the speed of
That women in a variety of socioeconomicchange in attitudes about intimate partner vio
positions have all become more likely to rejectlence will also contribute to our understanding
intimate partner violence suggests a general of the effects of global cultural diffusion.
cultural shift. Third, future research should expand tem
Further evidence of a general cultural shiftporal and geographic coverage. The DHS con
is provided by the analysis of within-cohorttinues to collect data and recently released a
change versus cohort replacement. Resultsthird wave of data on attitudes about intimate
show that within-cohort change explainspartner violence from five countries. The trend
nearly all of the observed change. Betweenof increasing rejection of intimate partner vio
survey waves, women of all ages became lence accelerated in Ethiopia, Malawi, Nepal,
more likely to reject intimate partner vioand Zimbabwe. Rwanda saw a slight reversal
lence. This finding, like the finding that popuof the trend. As more data become available,
lation compositional changes do not explain aadditional analyses should test the longevity of
majority of the attitudinal trend, is also conthe current trend in attitudes about intimate
sistent with the wide dissemination and influpartner violence. In addition, more data is
ence of global cultural scripts. The results needed to study the relationship between attitu
provide evidence of rapid cultural diffusion, dinal trends in low-, middle-, and high-income
not slow demographic change. countries. Detailed studies of outlier countries
The cohort replacement analysis also led to should examine why there is increasing accept
the surprising finding that younger women ance of intimate partner violence in Indonesia,
were less likely than their elders to reject inti Jordan, and Madagascar. Data used in this
mate partner violence. The popular image of study provide few clues as to why attitudinal
social changeinspired by young people com shifts in those countries did not follow the
ing to adulthood with liberal attitudesdoes dominant trend.
not apply in this case. In most countries Finally, it is important to note that the out
included in this study, the youngest respondents come investigated here was reported attitudes
were among the least likely to reject intimate about intimate partner violence. With these
partner violence. Although it may result from data, there is no way to distinguish real attitu
the positive relationship between age and women's dinal change from increasing influence of

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
262 American Sociological Review 78(2)

social desirability. For Funding

this study, I argue that
the distinction is unimportant; respondents'
This material is based on work supported by the National
Science Foundation
knowledge of the socially Graduate Research response
desirable Fellowship
is in itself evidence of[Grant
the No. DGE 0718128], The author gratefully
diffusion of global
acknowledges use of the services and facilities of the
norms. It does matter, however, whether
Population Studies Center at the University of Michigan,
women actually believe it Center
funded by NICHD is Grantokay for hus
R24 HD041028.
bands to beat their wives.
Moreover, studies of the influence of Acknowledgments
world society on national policies highlight I am grateful for contributions from Sarah Burgard,
decoupling between form and function result Arland Thornton, Kiyo Tsutsui, Barbara Anderson, and
ing from adoption of policy models regard Apoorva Aekka, as well as the anonymous reviewers and
ASR editors.
less of how well they fit a nation-state's needs
and intentions (Meyer et al. 1997). The same
pattern of decoupling is possible within indi Notes
viduals. Decoupling may manifest as a dis 1. The online database is located at http://webapps01.
connect between stated attitudes and actions. I collected infor
mation from the database from November 2011
Individuals will adopt or adapt some aspects
through February 2012.
of the global cultural discourse and reject oth2. Although it is plausible that reporting of national
ers. Similarly, individuals may selectively policies regarding violence against women
apply the logic of the global cultural dis improved as global attention to this issue grew, it is
course. Still, as Hafner-Burton and Tsutsui unlikely that the dramatic increase in the number of
policy actions can be entirely explained by changes
(2005) found, conformity with world society
in reporting practices.
that begins as an "empty promise" may have3. DHS data are available for download at http://www.
long-term consequences by providing lan

guage used to promote social change. Addi4. Ideally, wave 1 data for each country would have
tional in-depth research on how individuals been collected before the diffusion of global scripts
about violence against women, but such data are not
make use of global cultural scripts about available. The collection of data on attitudes about
violence against women will help to advance violence against women is itself part of world soci
the findings of this study. ety activism on this issue. Despite the timing of data
Investigation of connections between collection, however, it is clear that violence against
global cultural scripts and individual attitudes women gained prominence at the level of world
society before there was coordinated worldwide
is an important extension of world society mobilization at the national and individual levels.
research. Global cultural models matter not
Following institutionalization of the issue of vio
only because they influence elites worldwide lence against women in world society, local
but also because of their potential to influence activism began influencing national and interna
millions of individuals in less privileged tional responses, but the development and
dissemination of a global cultural script facilitated
social locations. During the first decade of the this local mobilization.
2000s, as violence against women became an 5. In cases where the sample consisted of all women in
increasingly prominent issue worldwide, one wave and ever-married women in the other

women across the globe became more likely wave, I restricted samples from both waves to ever

to reject intimate partner violence. Results are married women for the present analysis.

consistent with the influence of cultural diffu 6. There are small question wording differences across
waves in eight countries. Based on Yount and col
sion, not structural socioeconomic or demo
leagues' (2011) findings, the wording differences
graphic changes. Previous case study research may be causing an underestimate of the change over
has documented specific instances of the time in the Dominican Republic, India, and Turkey.
effect of the global on the local. This study Wording differences could be the source of some
overestimation of the change over time in Rwanda
found evidence that the diffusion of global
and Senegal, although effects are likely to be small.
cultural scripts can produce widespread trends Because of small sample sizes, Yount and colleagues
in attitudes at the individual level. (2011) were unable to estimate effects of the

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
Pierotti 263

Berkovitch, Nitza. 1999a. From Motherhood to Citizen

across-time wording variation found in Benin, Haiti,
and Mali, although these effects are also likely to be
ship: Women s Rights and International Organizations.
small. (Analysis available from the author.) Baltimore, MD: The Johns Hopkins University Press.
7. The India surveys excluded the third and fourth
Berkovitch, Nitza. 1999b. "The Emergence and Trans
standard scenarios and included the following two formation of the International Women's Movement."
scenarios instead: if she shows disrespect for inPp. 100-126 in Constructing World Culture: Interna
tional Nongovernmental Organizations Since 1875,
laws and if he suspects her of being unfaithful. The
Jordan surveys excluded the fourth standard sceedited by J. Boli and G. M. Thomas. Stanford, CA:
nario and included the following two additional Stanford University Press.
scenarios: if she insults him and if she disobeys
Berkovitch, Nitza and Karen Bradley. 1999. "The Global
him. Nigeria included one extra scenario in bothization of Women's Status: Consensus/Dissensus in
waves: if the food is not cooked on time. The first World Polity." Sociological Perspectives 42:481-98.
standard scenario was not asked in Turkey; instead,
Boli, John and George M. Thomas. 1997. "World Culture
a scenario asked if she wastes money. Finally, in in the World Polity: A Century of International Non
Cambodia the fifth scenario was excluded and in Governmental Organization." American Sociological
Zambia the first scenario was excluded, both Review 62:171-90.

because of variation in the wording of the scenario Boyle, Elizabeth H. 2002. Female Genital Cutting: Cul
across survey waves. tural Conflict in the Global Community. Baltimore,
8. To be conservative, only respondents who answered MD: Johns Hopkins University Press.
no to each question were assigned a 1 on this depenBoyle, Elizabeth H. and Kristin Carbone-Lopez. 2006.
dent variable. Respondents who said "don't know" "Movement Frames and African Women's Explana
in reply to any of the scenarios were coded 0. tions for Opposing Female Genital Cutting." Interna
9. Researchers have also found mixed evidence of an tional Journal of Comparative Sociology 47:435-65.
association between spousal age differences and Boyle, Elizabeth H. and Sharon E. Preves. 2000.
attitudes about intimate partner violence (e.g., "National Politics as International Process: The Case
Kishor and Subaiya 2008). In the data used here, of Anti-Female-Genital-Cutting Laws." Law & Soci
there was only a weak association between spousal ety Review 34:703-737.
Chimbwete, Chiweni, Susan Cotts Watkins, and Eliya
age gaps and attitudes about intimate partner vio
lence. For the sake of parsimony, I dropped this Msiyaphazi Zulu. 2005. "The Evolution of Population
variable from the analysis. Policies in Kenya and Malawi." Population Research
10. I also conducted this analysis by estimating the and Policy Review 24:85-106.
marginal effects from a logistic regression andCleland, John and Christopher Wilson. 1987. "Demand
using the marginal effects in place of the coeffi Theories of the Fertility Transition: An Iconoclastic
cients from OLS regression. Results were the same, View." Population Studies 41:5-30.
Filmer, Deon and Lant H. Pritchett. 2001. "Estimating
so I opted to include the OLS regression methodol
ogy and results for ease of interpretation. Wealth Effects without Expenditure DataOr Tears:
11. In these five countries, the effect of survey wave (the An Application to Educational Enrollments in States
contribution of within-cohort change) depends on of India. " Demography 38:115-32.
the respondent's birth cohort. Effects of survey yearFirebaugh, Glenn. 1989. "Methods for Estimating Cohort
and birth year are not independent so linear regres Replacement Effects." Sociological Methodology
sion decomposition cannot isolate their contributions 19:243-62.

Firebaugh, Glenn. 1992. "Where Does Social Change

to the change. Nonetheless, the overall picture
gleaned from the decomposition is consistent with a Come From? Estimating the Relative Contributions
closer examination of the data: in Indonesia and of Individual Change and Population Turnover." Pop
Madagascar, cohort replacement made a substantial ulation Research and Policy Review 11:1-20.
contribution to the decrease in the percent of womenFrank, David John, Bayliss J. Camp, and Steven A.
who reject intimate partner violence; in India, Nepal, Boutcher. 2010. "Worldwide Trends in the Criminal
and Zimbabwe, cohort replacement accounted for Regulation of Sex, 1945 to 2005." American Socio
logical Review 75:867-93.
none of the increase in the percent of women who
reject intimate partner violence. Frank, David John, Tara Hardinge, and Kassia Wosick
Correa. 2009. "The Global Dimensions of Rape-Law
Reform: A Cross-National Study of Policy Out
comes." American Sociological Review 74:272-90.
Alwin, Duane F. 1990. "Cohort Replacement and
Frank, David John, Ann Hironaka, and Evan Schofer.
Changes in Parental Socialization Values." Journal of 2000. "The Nation-State and the Natural Environ
Marriage and Family 52:347-60. ment over the Twentieth Century." American Socio
Barrett, Deborah and Amy OngTsui. 1999. "Policy as Sym logical Review 65:96-116.
Hafner-Burton, Emilie M. and Kiyoteru Tsutsui. 2005.
bolic Statement: International Response to National
Population Policies." Social Forces 78:213-34. "Human Rights in a Globalizing World: The Paradox

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
264 American Sociological Review 78(2)

Merry, Sally
of Empty Promises." American Engle. 2006.
Journal ofHuman Rights and Gender
110:1373-1411. Violence: Translating International Law into Local
Heise, Lori L. 1998. "Violence against Women: An Inte
Justice. Chicago: University of Chicago Press.
grated, Ecological Framework." Violence against
Meyer, John W. 2010. "World Society, Institutional The
Women 4:262-90. ories, and the Actor." Annual Review of Sociology
Heise, Lori L. and Claudia Garcia-Moreno. 2002. "Vio 36:1-20.
lence by Intimate Partners." Pp. 87-121 in World Meyer, John W., John Boli, George M. Thomas, and
Report on Violence and Health, edited by E. G. Krug, Francisco O. Ramirez. 1997. "World Society and
L. L. Dahlberg, J. A. Mercy, A. B. Zwi, and R. Loz the Nation State." American Journal of Sociology
ano. Geneva: World Health Organization. 103:144-81.

Hornik, Robert and Emile McAnany. 2001. "Mass Media Meyer, John W. and Ronald L. Jepperson. 2000. "The
and Fertility Change." Pp. 208-229 in Diffusion Pro 'Actors' of Modern Society: The Cultural Construc
cesses and Fertility Transition: Selected Perspectives, tion of Social Agency." Sociological Theory 18:100
edited by J. B. Casterline. National Research Council, 120.
Committee on Population, Division of Behavioral Meyer, John W., Francisco O. Ramirez, and Yasemin N.
and Social Sciences and Education. Washington, DC: Soysal. 1992. "World Expansion of Mass Education,
National Academy Press. 1870-1980." Sociology of Education 65:128-49.
Htun, Mala and S. Laurel Weldon. 2012. "The Civic National Research Council. 2001. Diffusion Processes
Origins of Progressive Policy Change: Combat and Fertility Transition: Selected Perspectives, edited
ing Violence against Women in Global Perspective, by J. B. Casterline. Committee on Population, Divi
1975-2005." American Political Science Review sion of Behavioral and Social Sciences and Educa
106:548-69. tion. Washington, DC: National Academy Press.
Jayakody, Rukmalie, Arland Thornton, and William Norris,
G. Pippa and Ronald Inglehart. 2002. "Islamic Cul
Axinn, eds. 2008. International Family Change: Ideture and Democracy: Testing the 'Clash of Civiliza
ational Perspectives. New York: Taylor & Francistions' Thesis." Comparative Sociology 1:235-69.
Group. Paxton, Pamela, Melanie M. Hughes, and Jennifer L.
Keck, Margaret E. and Kathryn Sikkink. 1998. Activists Green. 2006. "The International Women's Movement

beyond Borders: Advocacy Networks in International and Women's Political Representation." American
Politics. Ithaca, NY: Cornell University Press. Sociological Review 71:898-920.
Kishor, Sunita and Lekha Subaiya. 2008. Understanding Ramirez, Francisco O., Yasemin Soysal, and Suzanne
Women's Empowerment: A Comparative Analysis of Shanahan. 1997. "The Changing Logic of Political
Demographic and Health Surveys (DHS) Data. DHS Citizenship: Cross-National Acquisition of Women's
Comparative Reports No. 20. Calverton, MD: Macro Suffrage Rights, 1890 to 1990." American Sociologi
International. cal Review 62:735-45.
Koo, Jeong-Woo and Francisco O. Ramirez. 2009. Rubtsova, Anna A. 2011. "How to Feed a Baby: Global
"National Incorporation of Global Human Rights: Processes and Individual Choices." Emory University,
Worldwide Expansion of National Human Rights Atlanta, Georgia (
Institutions, 1966-2004." Social Forces 87:1321-53.
Lesthaeghe, Ron. 1983. "A Century of Demographic and Ryder, Norman B. 1965. "The Cohort as a Concept in
Cultural Change in Western Europe: An Exploration the Study of Social Change." American Sociological
of Underlying Dimensions." Population and Devel Review 30:843-61.
opment Review 9:41135. Thompson, Karen Brown. 2002. "Women's Rights Are
Lesthaeghe, Ron and Johan Surkyn. 1988. "Cultural Human Rights." Pp. 96-122 in Restructuring World
Dynamics and Economic Theories of Fertility Politics: Transnational Social Movements, Networks,
ChangPopulation and Development Review 14:1 and Norms, edited by S. Khagram, J. V. Riker, and K.
45. Sikkink. Minneapolis: University of Minnesota Press.
Levitt, Peggy and Sally Eagle Merry. 2009. "Vernacular Thornton, Arland. 2001. "The Developmental Paradigm,
ization on the Ground: Local Use of Global Women's Reading History Sideways, and Family Change."
Rights in Peru, China, India and the United States." Demography 38:449-65.
Global Networks 4:441 -61. Thornton, Arland. 2005. Reading History Sideways: The
Longhofer, Wesley and Evan Schofer. 2010. "National Fallacy and Enduring Impact of the Developmental
and Global Origins of Environmental Association." Paradigm on Family Life. Chicago: University of
American Sociological Review 75:505-533. Chicago Press.
Luke, Nancy and Susan Cotts Watkins. 2002. "Reactions Tinker, Irene. 2004. "Introduction: Women Transforming
of Developing-Country Elites to International Popu International Development." Pp. xiii-xxx in Develop
lation Policy." Population and Development Review ing Power: How Women Transformed International
28:707-733. Development, edited by A. S. Fraser and I. Tinker.
Mason, Karen Oppenheim. 1997. "Explaining Fertility
New York: The Feminist Press at the City University
Transitions." Demography 34:443-54. of New York.

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to
Pierotti 265

Torfason, Magnus Thor and Paul Ingram. 2010. "Inter Comparative Analysis of 17 Sub-Saharan Countries."
governmental Organization and the Global Rise ofBMC International Health and Human Rights.
Democracy." American Sociological Review 75:355 Vyas, Seema and Lilani Kumaranayake. 2006. "Con
77. structing Socio-Economic Status Indices: How to Use
True, Jacqui and Michael Mintrom. 2001. "Transnational Principal Components Analysis." Health Policy and
Networks and Policy Diffusion: The Case of Gender Planning 21:459-68.
Mainstreaming." International Studies Quarterly Yount, Kathryn M. 2005. "Resources, Family Organiza
45:27-57. tion, and Domestic Violence against Married Women
Tsutsui, Kiyoteru and Hwa-Ji Shin. 2008. "Global
in Minya, Egypt." Journal of Marriage and Family
Norms, Local Activism and Social Movement Out 67:579-96.
comes: Global Human Rights and Resident Koreans
Yount, Kathryn M., Nafisa Halim, Michelle Hynes, and
in Japan." Social Problems 55:391-418. Emily R. Hillman. 2011. "Response Effects to Atti
UNiTE to End Violence against Women. 2012. "About
tudinal Questions about Domestic Violence against
UNiTE." United Nations Secretary-General's CamWomen: A Comparative Perspective." Social Science
Research 40:873-84.
paign. Retrieved March 7, 2012 (
en/women/endviolence/about. shtml). Yount, Kathryn M. and Li Li. 2009. "Women's 'Justifi
United Nations. 2008. "Good Practices in Legislation on
cation' of Domestic Violence in Egypt." Journal of
Violence against Women: Report of the Expert Group Marriage and Family 71:1125-40.
Meeting." Vienna, Austria: United Nations Division
Yount, Kathryn M. and Hoda Rashad, eds. 2008. Family
for the Advancement of Women & United Nations in the Middle East: Ideational Change in Egypt, Iran,
Office on Drugs and Crime. and Tunisia. Oxford, UK: Routledge.
United States Agency for International Development
(USAID) Women's Legal Rights Initiative. 2007.
Rachael S. Pierotti is a doctoral candidate in sociology
Women s Legal Rights Initiative Final Report. Wash
ington, DC: Chemonics International Inc. Retrieved
and a Population Studies Center trainee at the University
of Michigan. Her research interests include the effects of
March 7, 2012 (
cross-cutting_programs/wid/dg/wlr.html). gender inequality on population health and economic
Uthman, Olalekan ., Stephen Lawoko, and Tahereh
development, and the influence of demographic shifts
Moradi. 2009. "Factors Associated with Attitudes and efforts to promote development on the social con
towards Intimate Partner Violence against Women:struction
A of gender.

This content downloaded from on Thu, 18 Aug 2016 19:10:53 UTC
All use subject to