Journal of Cleaner Production xxx (2012) 1e10

Contents lists available at SciVerse ScienceDirect

Journal of Cleaner Production

journal homepage: www.elsevier.com/locate/jclepro

Approach to a causal model between attitudes and environmental behaviour.

A graduate case study
Luis M. Rodríguez-Barreiro a, Rosario Fernández-Manzanal b, *, Luis M. Serra c, José Carrasquer b,
María B. Murillo d, María J. Morales e, José M. Calvo b, Javier del Valle f
a
Department of Science Education, Universidad de Zaragoza, San Juan Bosco, 7, 50009 Zaragoza, Spain
b
Beagle - Department of Didactic of Experimental Sciences, Universidad de Zaragoza, San Juan Bosco, 7, 50009 Zaragoza, Spain
c
GITSE, Aragon Institute of Engineering Research (I3A), Universidad de Zaragoza, Zaragoza, Spain
d
GPT, Aragon Institute of Engineering Research (I3A), Universidad de Zaragoza, Zaragoza, Spain
e
Department of Didactic of Experimental Sciences, Universidad de Zaragoza, Zaragoza, Spain
f
Department of Geography and Land Management, Universidad de Zaragoza, Zaragoza, Spain

a r t i c l e i n f o a b s t r a c t

Article history: This paper offers a study of the environmental attitudes of graduates from the University of Zaragoza and
Received 13 November 2011 their relationship with pro-environmental behaviour. These attitudes were analysed using a validated
Received in revised form Likert scale, along with a questionnaire on environmental behaviours. A set of 20 items in the scale were
20 September 2012
grouped into four categories: Formation, Outreach activities, Conservation and Intention to act. Also, a set
Accepted 21 September 2012
Available online xxx
of 12 items were used to evaluate environmental behaviour. Structural equation modelling was used to
test a causal model relating attitude to behaviours. The model shows the connection between two of the
attitudinal factors e conservation and intention to act e with pro-environmental behaviour. The most
Keywords:
Environmental attitudes
relevant to the causal model is that the Conservation factor is the one that has shown the closest relation
Environmental behaviour with Behaviour and this relation is shown through the Intention to act. The study involved 60 graduates
Structural equation modelling with AMOS that belong to departments that offer either compulsory or optional courses concerning the environ-
ment. This study suggests that conservation perspective should be considered in the university studies,
both in interdisciplinary education programs and research programs. Moreover, the results obtained in
this research work suggest that providing in the University adequate resources and habits in everyday
academic life turns to be a very significant aspect for fostering pro-environmental behaviour.
Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction The recognition of the significant role played by human activi-

Education has always been seen as key to improving quality of
life, not just of individuals but also collectively for humankind
(Tilbury, 2012).
Higher education is, unquestionably, an important tool for
building the future. In the words of Lozano (2006), “the future
leaders, decision-makers and intellectuals of the social, political,
economic and academic sectors are created, formed and shaped
within the world’s higher education institutions”. Universities are
starting to realise the impact they have on the environment, but
according to authors such as Lidgren et al. (2006), Lozano (2006,
2012), Desha and Hargroves (2010), Ferrer-Balas et al. (2010) or
Jones et al. (2010), this awareness is still relatively new in most
universities.
* Corresponding author.
E-mail address: roferma@unizar.es (R. Fernández-Manzanal).
ties in the degradation of the environment should be associated
with an increased awareness of environmental problems. If this
relationship of human activitieseenvironmental degradation is
widely accepted, it will be possible to also address the need to
achieve more sustainable lifestyles. Researchers interested in
finding out how people view environmental problems are gradu-
ally paying more attention to environmental conscience and to new
“attitudinal goals” associated with the environment (Dunlap et al.,
2000).
Several attitudinal studies have been carried out in social and
behavioural sciences: a reason for this interest is the belief that
people make evaluative judgements about a wide variety of objects
and rely on these judgements, or attitudes, in deciding among
several possible courses of action in the future (Crawley and
Koballa, 1994).
A common definition of attitude describes it as including the
three components of cognition, affect, and behaviour (Crano and

0959-6526/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jclepro.2012.09.029

Please cite this article in press as: Rodríguez-Barreiro, L.M., et al., Approach to a causal model between attitudes and environmental behaviour.
A graduate case study, Journal of Cleaner Production (2012), http://dx.doi.org/10.1016/j.jclepro.2012.09.029
2 L.M. Rodríguez-Barreiro et al. / Journal of Cleaner Production xxx (2012) 1e10
Prislin, 2006). As Ajzen (2001) pointed out, when we have an
attitude, we judge something along emotional dimensions, such as
good or bad, harmful or beneficial, pleasant or unpleasant and
likable or dislikable. It is important to notice that these evaluative
judgements are always focused towards something, often called the
attitude object. Herein, the attitude object refers to the environ-
mental judgement made by university graduates, and evaluated
through their cognition, affect and willingness to act.
Systematic research of environmental attitudes started to be
carried out in the early seventies (Wiegel and Wiegel, 1978) and is
now a clearly defined field, as shown by several studies like Stern
et al. (1999), Dunlap et al. (2000) or Schultz (2001) among others.
The study of environmental attitudes is interesting, above all,
because of the possible influence on human behaviour. Environ-
mental attitudes provide a good understanding of the set of beliefs,
interests or rules that influence pro-environmental action. Conse-
quently, numerous theoretical frameworks have been developed to
explain the relationship between attitudes and pro-environmental
action (Schwartz, 1977; Ajzen and Fishbein, 1980, 2000; Triandis,
1980; Hines et al., 1986/87).
Key researches that establish the potential relationship between
environmental attitudes and behaviours are presented next.
Many studies were conducted based on prediction of behaviour
from attitudinal variables within the framework of the theory of
reasoned action described by Ajzen and Fishbein in 1980. This
theory states that attitudes are important to the behaviour but do
not determine behaviour directly; rather, attitudes influence
behavioural intentions, which in turn shape our actions.
Theories in social and health psychology assume that intentions
cause behaviours. The relevance of intention to act or willingness to
act is justified in many research works, e.g., Armitage and Conner
(2001), Bamberg and Schmidt (2003), Webb and Sheeran (2006),
Bamberg and Möser (2007) or Rivis et al. (2009). These studies use
modern statistical methods to synthesise results of a body of
primary studies, and have provided convincing empirical evidence
for the utility of intention as predictor of environmental behaviour.
Other models have been developed throughout the search for
variables associated with responsible environmental behaviour
(Stern, 2000; Courtenay-Hall and Rogers, 2002; Kollmus and
Agyeman, 2002; Juárez-Nájera et al., 2010; Lin and Huang, 2012),
which have highlighted the influence of internal factors, called
environmental consciousness (such as values and attitudes,
knowledge, feelings, emotional involvement or psychological
benefit) and external factors (such as infrastructure, social and
cultural factors, economic situation, etc.).
Stern (2000) indicates the types of variables that may influence
behaviour and analyses the strength and implications. Apart from
attitudes, Stern considers three variables to be important: contex-
tual aspects, personal skills, and habits or routines. It is also stated
that environmental behaviour cannot be explained solely by the
action of variables, such as attitudes or social and cultural variables.
According to the Stern, context plays a key role in understanding
and assessing behaviours, and attitudinal factors have an important
predictive function for behaviour when not constrained by context
or personal abilities. However, when a suitable context comes
about, both institutional factors (some actions only take place with
the presence of adequate infrastructures) and economic factors
intervene. In the present paper, several aspects suggested by Stern
have been considered when analysing the environmental behav-
iours of the university graduates, based on context, habits or
institutional factors.
In a higher education level, there is a lack of models analysing
the connection between environmental attitudes and behaviours.
Nevertheless, a model recently published by Juárez-Nájera et al.
(2010) focuses on German and Mexican universities and explores
Please cite this article in press as: Rodríguez-Barreiro, L.M., et al., Approach to a causal model between attitudes and environmental behaviour.
A graduate case study, Journal of Cleaner Production (2012), http://dx.doi.org/10.1016/j.jclepro.2012.09.029
values and moral norms grounded within individuals, rather than
grounded in rational choice and self-interest. This model considers
motivation, largely related to a personal sense of responsibility, as
a stimulus to perform specific pro-environmental actions.
Considering the aforementioned research studies, it can be
concluded that the scenario is quite complex and it is essential to
further examine the relationship between environmental attitudes
and behaviours.
Research at the University of Zaragoza verified whether
university education has equipped students with environmental
attitudes, and the extent of these attitudes in connection with
environmental behaviour. This paper analyses the environmental
attitudes of university graduates, according to the definition of
attitudes by Ajzen (2001). The scale of attitudes used herein (see
Appendix 1) was developed in an earlier work (Fernández-
Manzanal et al., 2007) to obtain insight in the graduates’ environ-
mental attitudes (Environmental Attitudes of the University, EAU
Scale). The use of the same scale of attitudes will help discover if
university environmental education has provided students with
environmental attitudes.
Furthermore, an analysis of everyday environmental behaviours
explored through a self-report is also presented (Questionnaire on
Environmental Behaviours, see Appendix 2).
Results of these questionnaires are applied to the analysis and
search for a possible influence of attitudes on proenvironmental
behaviour. A causal model is proposed, in which attitudes and
environmental behaviours in everyday life are related. Following
this analysis, it is explained how the proposed causal model reflects
the relationship between environmental attitudes and behaviours.
The model was developed using structural equation modelling.
2. Case study: the University of Zaragoza
Environmental evaluation of graduate students aims at estab-
lishing the extent of environmental attitudes provided by univer-
sity education and verifying whether such attitudes are displayed
in everyday pro-environmental behaviour.
Addressing such issues allows the validity and reliability of the
proposed questionnaires to be contrasted, evaluating whether
university education is promoting sustainable behaviour through
its education, and searching for a relationship between environ-
mental attitudes and behaviours, i.e., find a causal model that
enables the establishment of a relationship between the attitudes
examined and everyday environmental behaviours.
The following questions were part of the questionnaire:
1. What are the characteristics of the environmental attitude
displayed by University of Zaragoza graduates?
2. What environmental behaviours in everyday life are displayed
by University of Zaragoza graduates?
3. Is there a causal relationship between environmental attitudes
and behaviours? Which factors of the environmental attitude
explain the causal model between environmental attitudes and
pro-environmental behaviours?
3. Method
3.1. Sample
The sample consisted of 60 university graduates who obtained
their degrees from various faculties of the University of Zaragoza,
Spain. The participants belong to departments that offer either
compulsory or optional courses concerning the environment or
environmental training. The faculties and centres involved in this
 L.M. Rodríguez-Barreiro et al. / Journal of Cleaner Production xxx (2012) 1e10 3

study and their corresponding percentage of participants follow: 0.70 and 0.80 are respectable and between 0.80 and 0.90 are very
Faculty of Education (11.7%), Faculty of Humanities and Education good.
(18.3%), Faculty of Humanities and Social Sciences (13.3%), Faculty The distribution of the items into the previously referred factors
of Law (6.7%), Faculty of Economics and Business Administration or categories was obtained by means of exploratory factor analysis
(13.3%), Faculty of Geography (6.7%), Faculty of Sciences (5%), and confirmatory factor analysis (Fernández-Manzanal et al., 2007).
Polytechnic Centre of Engineering (25%). The participants pre- The objective of exploratory factor analysis is to find the number of
sented the following characteristics: separate components that might exist for a group of items. The
confirmatory factor analysis (Bollen, 1989; Arbuckle and Wothke,

a course with environmental content was completed in their 1999; Arbuckle, 2003) provides, along with other advantages,
graduate or further studies (for instance, PhD courses); a suitable statistical framework to evaluate the validity and reli-

they are currently working in schools or companies and the ability of each item, instead of only providing global assessments.
work period ranges from 2 months to 5 years; After these analyses, the items in the scale were sorted into 4

they have freely accepted to participate in this research. categories as previously referred. This scale has also been used by
other authors (Yurt et al., 2010) and a repeated factor analysis on
The sample can be defined as intentional, as the researchers the new data can confirm validity.
contacted possible participants due to the existence of a previous The application of this scale to the sample provided the results
relationship, which facilitated the possibility of direct contact. shown in Table 1.
Possibility of direct contact is the main reason for using this type of The value obtained for Cronbach’s alpha is 0.78, slightly lower
sample (aside from filling out questionnaires, interviews were also than the value obtained with undergraduate students.
carried out, however not the scope of this paper). Categories with the highest rating: C1, C2, C3.
Categories with the lowest rating: C4.
3.2. Measures With respect to the behavioural questionnaire (see Appendix 2),
a list of questions was asked in order to define which everyday
The first and second questions of the problem (Section 2) were activities correspond to those that might have arisen throughout
answered by attitude and behaviour questionnaires used in the respondents’ academic life, such as reading or learning about
a previous research, in which 952 undergraduate students from the environmental issues from magazines or newspapers (behaviour 1,
University of Zaragoza participated (Fernández-Manzanal et al., 3, 11, 12). Other environmental actions analysed by this question-
2007). naire are those that can be implemented because they have insti-
The attitude questionnaire is a scale (Environmental Attitudes of tutional support, such as the proximity of recycling bins, or the
the University, EAU Scale, Appendix 1) in which environmental availability of recycled paper at the workplace (behaviour 2, 5, 7);
attitude is explored through 20 items divided into 4 different other behaviours are related to the habits acquired in the family
categories: Category 1, Formation, explores the participants’ feel- home or throughout life, such as turning out the lights when
ings regarding environmental education at the university and the leaving an empty room (behaviour 4, 6, 9) and, finally, behaviours
importance given to such education, comprising items i1, i8, i9, i17 related to transportation are explored along with the consumption
and i18. Category 2, Outreach activities, with items: i2, i7, i12, i15 of “green” products (behaviour 8, 10). The results for this ques-
and i20, examines how the participants value the fieldwork or tionnaire are shown in Table 2, where the possible answers were
outreach work. Category 3, Conservation, with items: i4, i6, i10, i14 YES or NO.
and i19, examines the importance given to conservation of the Behaviours with the highest percentage: b1, b4, b5, b6, b7, b9
environment. Finally, Category 4, Intention to act comprising the and b11.
following items: i3, i5, i11, i13 and i16 explores the willingness to Behaviours with the lowest percentage: b2, b3, b8, b10 and b12.
act or intention of maintaining certain pro-environmental
behaviours.
The EAU Scale represents a popular format for a Likert-type
scale, which consists of asking respondents to mark an X in one of Table 1
five possible responses: Strongly Agree, Agree, Indifferent, Average for the items in the scale.
Disagree, and Strongly Disagree. Agreement response choices are N Minimum Maximum Mean Std. deviation
usually bipolar. The graduates were asked to show their level of i1 60 2 5 4.40 0.67
agreement, indifference or disagreement with the statements. Five i2 60 2 5 4.17 0.72
points were attributed to “Strongly Agree” and one point for the i3 60 2 5 4.20 0.69
“Strongly Disagree”, in the case of favourable statements. The i4 60 1 5 4.40 0.85
i5 60 1 5 3.17 0.99
opposite was applied to unfavourable statements. Intermediate
i6 60 2 5 3.57 0.98
positions between these two extremes were attributed different i7 60 2 5 4.30 0.67
marks. Favourable statements correspond to items: i2, i3, i5, i7, i8, i8 60 4 5 4.65 0.48
i9, i12, i15, i17 and i19, with the remaining items referring to i9 60 1 5 3.52 0.83
unfavourable statements. i10 60 4 5 4.47 0.50
i11 60 2 5 3.37 0.96
Data were subject to an item analysis to choose the set of items i12 60 2 5 4.32 0.65
that formed an internally consistent scale. Internal consistency is i13 60 2 5 3.72 0.85
a measurable property of items that implies that they measure the i14 60 1 5 3.95 0.91
same object of attitude. This analysis includes the item-total i15 60 2 5 4.27 0.63
i16 60 1 5 4.12 0.98
correlation for each item and Cronbach’s coefficient alpha. All
i17 60 2 5 4.33 0.82
these correlations are reasonably strong. Internal consistency i18 60 1 5 4.50 0.81
among a set of items suggests they share common variance. The i19 60 2 5 4.15 0.78
Cronbach’s coefficient alpha value of 0.84 was not unexpected for i20 60 3 5 4.55 0.62
undergraduate students. According to DeVellis (1991) research Note: The minimum values indicate the lowest value given by the respondents to
scales between 0.65 and 0.70 are minimally acceptable, between the referred item.

Please cite this article in press as: Rodríguez-Barreiro, L.M., et al., Approach to a causal model between attitudes and environmental behaviour.
A graduate case study, Journal of Cleaner Production (2012), http://dx.doi.org/10.1016/j.jclepro.2012.09.029
4 L.M. Rodríguez-Barreiro et al. / Journal of Cleaner Production xxx (2012) 1e10
Table 2
Percentage of responses per question.
b1 b2 b3 b4
Percentage, yes 63.6 57.4 29.1 92.7
Percentage, no 36.4 42.6 70.9 7.0
a
b6 and b8 are unfavourable statements.
The obtained results are similar to those obtained with the
undergraduates, except in the case of b8 that presents a worse
valuation in this sample (graduate case study).
The third question of the problem (see Section 2) was addressed
by focussing on the feasibility of constructing a causal model that
reflects the relationship between attitudes and environmental
behaviour. Two types of analyses e which are presented below e
were carried out.
3.3. Data analysis
The last part of the research was developed into two different
phases, which correspond to an exploratory bias and a confirma-
tory bias. In the exploratory phase, two types of analyses were
carried out. The first one was achieved by means of multiple linear
regressions and the second one using the Specification search tool
in AMOS. Both analyses led to a unique plausible model that was
tested in the second phase, in which SEM (Structural Equation
Modelling) methodology was applied from a confirmatory
perspective.
In relation to SEM, Byrne (2001) pointed out that it is a statistical
methodology that takes a confirmatory approach, rather than an
exploratory, to the data analysis. It constitutes a powerful tool for
multivariate analysis. The term structural equation models
expresses two basic aspects of the procedure: a) that the causal
relationships are represented by a series of structural equations;
and b) that these structural relations can be modelled graphically e
in the form of a diagram e to enable a clearer conceptualisation of
the theory under study.
SEM has become a widely used methodology for experimental
research (Byrne, 2001). As noted by Mueller (1996) and Murillo-
Luna et al. (2011), structural equation modelling should not be
seen as a mere statistical technique: its use is similar to a research
process. According to Arbuckle (2003), the use of SEM has grown
enormously over the last three decades. Without doubt, the
development of diverse programmes to estimate and validate these
models, e.g., AMOS (Analysis of Moment Structures), has contrib-
uted to this growth.
3.3.1. First phase. Exploratory model analysis
Analysis of the potential influence of the factors comprising the
scale of attitudes, i.e., their predictive power regarding environ-
mental behaviour, has been studied using multiple linear regres-
sion analysis. This technique allows for two broad classes of
applications, depending on whether they focus on prediction or on
explanation. On the predictive side of this technique, the linear
combination of the independent variables e the factors on the scale
e is engineered to maximise the estimation of the dependent
variable e behaviour e and is a predictor of the explanatory power
of the dependent variable by the independent variables (Batista
and Coenders, 2000).
Regarding the regression model, the stepwise method was used
to select the independent variables. With this method, the variables
are examined at each stage to see whether they are included in the
model. Table 3 shows the best model, which contains three inde-
pendent variables (C4, Intention to act; C3, Conservation, and C2,
Outreach activities). The value of the adjusted R square indicates
Please cite this article in press as: Rodríguez-Barreiro, L.M., et al., Approach to a causal model between attitudes and environmental behaviour.
A graduate case study, Journal of Cleaner Production (2012), http://dx.doi.org/10.1016/j.jclepro.2012.09.029
b5 b6a b7 b8a b9 b10 b11 b12
85.5 3.7 90.9 46.3 67.3 9.1 80.0 49.1
14.5 96.3 9.1 53.7 32. 90.9 20.0 50.9
that these predictors explain the 43.4% of the variability of the
criterion. Moreover, verification of the hypothesis R ¼ 0 shows
significant differences and, as a consequence, it suggests that
a relationship exists between the referred components of the scale
and environmental behaviour. The contribution of the fourth vari-
able (C1) is not relevant.
The tool in AMOS, Specification search, helps compare several
different models. To this end, AMOS provides instruments that
enable to adjust the model proposed, as well as to select one of the
different models maintaining three of their properties: fit, parsi-
mony1 and significance. In this phase, the structural model shown
in Fig. 1 was taken as starting point.
The independent variables (Conservation, Formation, Outreach
activities and Intention to act) are measured using the EAU Scale
items (Appendix 1). The dependent variable Behaviour compre-
hends the 12 items included in the questionnaire (Appendix 2).
Specification search allows for a priori verification of the
dispensable nature of some of the relationships between variables,
which are marked by dashed arrows. When there is a dashed arrow,
AMOS adjusts the model either with this feature or without it. If
several arrows have been drawn e three in our diagram 2 e the
software adjusts the model 23 ¼ 8 times by using all the possible
subsets of optional arrows.
Once the estimation of parameters is completed, AMOS provides
a series of indexes with comparative purposes. One of the most
widely used index is the rescaled version of the Bayes Information
Criterion BIC0 (Raftery, 1995). According to this criterion, the best
model among all structural models compared is the one that
includes only factors Conservation and Intention to act and the
dependent variable Behaviour.
As a conclusion, the two exploratory analyses led to the model
proposal shown in Fig. 2 as the most plausible model. The figure
shows how the Intention to act factor plays a mediating role
between the Conservation and the Behaviour.
Specification search also shows that the items that eventually
enable the evaluation of the two latent or independent variables
are i4, i6, i10 and i14 for Conservation, and i3, i11, i13 and i16 for
Intention to act.
3.3.2. Second phase. Confirmatory model analysis
The second phase of the investigation focused exactly on testing
the model through SEM. Once the model was specified (Fig. 2),
parameter estimation must be established, so the possible modi-
fications in order to interpret the final model could be analysed.
By default, the estimation of parameters was based on the
maximum likelihood (ML) method (Satorra and Bentler, 1994;
Bentler and Dudgeon, 1996; Garson, 2011). As states Garson (2011),
ML calculates estimates based on maximising the probability that
the observed covariances are drawn from a population assumed to
1
In agreement with the principle of parsimony, a phenomenon should be
described with minimum possible number of elements.
2
Even though they are not included in the diagram, dashed arrows have been
also drawn in measurement models, from independent variables to items or
dependent variables that make measurement possible. Furthermore, this allows
setting aside some items. It is helpful with SEM models to distinguish between
independent variables and those that are dependent.
 L.M. Rodríguez-Barreiro et al. / Journal of Cleaner Production xxx (2012) 1e10 5

Table 3
Test of the hypothesis R ¼ 0 using ANOVA.b
Analysis of Sum of DF Mean F Significance
variance squares square
Regression 74.043 3 24.681 13.791 0.000a
Residual 84.114 47 1.790
a
Input variables: C4, C3, C2.
b
Dependent variable: Behaviour.
be the same as that reflected in the coefficient estimates. That is, ML
picks estimates which have the greatest chance of reproducing
covariances in the observed data.
It is important to note that with use of ML estimation, it is
assumed that the hypothesised model is valid, and the scale of the
dependent variables is continuous (Byrne, 2001).
Global statistics was used to confirm the goodness of the model,
which usually verifies the capacity of the estimated model to
reproduce the master parameter matrix. Robust statistics and
indices found in the psychometric literature have been used, such
as: the Number of parameters (NPAR), the Minimum Discrepancy
(CMIN) most commonly expressed as a Chi Square (c2), the Normed
Fit Index (NFI), the Incremental Fit Index (IFI), the Comparative Fit
Index (CFI), the Akaike’s Information Criterion (AIC) with the
BrowneeCudeck Criterion (BCC) and the Expected Cross-Validation
Index (ECVI). These indexes are described by Akaike’s (1987), Bollen
(1989), Browne and Cudeck (1989, 1993), Bentler (1995) or Jöreskog
and Sörbom (1996) between others.
4. Results
4.1. Parameters
The various aspects concerning estimated parameters (feasi-
bility errors and significance) are presented in this section, along
with potential defects in specification.
The section of the AMOS in Table 4 presents separately the factor
loadings (as regression coefficients), and in Table 5, the variances
(in this case, for both factors and measurement errors). Table 6
shows the corresponding standardised weights and the Squared
Multiple Correlations. In relation to parameters, listed to the right
of each parameter is presented its estimated value (column 1),
standard error (column 2), and critical ratio (column 3). An exam-
ination of this unstandardised solution reveals all estimates to be
both substantively reasonable. All standard errors appear also to be
in good order. All variances are also significant except for two,
belonging to measurement Error e4 and e3. The reason may lie
perhaps in the size of the sample which will be analysed below.
Conservation Intention
Behaviour
Adjusted
R square
0.434 Fig. 2. The most plausible model.
The squared multiple correlations (Table 6) is a useful statistic
information that is independent of all units of measurement and
represents the proportion of variance that is explained by the
predictors of the analysed variable. Squared multiple correlations
for variables are, Intention ¼ 0.835 and Behaviour ¼ 0.307. From
such data, it is concluded that the variable Behaviour is explained
by the model as 30.7%. Intention to act, for its part, stands at 83.5%.
The causal model proposed, which depicts the causal relation-
ship between environmental attitudes and behaviours, is shown in
Fig. 3. The diagram shows one exogenous variable, Conservation,
one endogenous variable, Intention to act, and the dependent
variable, Behaviour. Each one of the latent factors is estimated by
their respective indicators (the items of the scale enclosed in
rectangles), from which the arrows are sent to their indicators. Size
errors appear enclosed in circles, unseen random variables that join
all the unconsidered effects in the system and that can affect the
measurement of the variable it is influencing (Ruiz, 2000).
4.2. Interpretation of the model as a whole. Goodness of fit statistics
The goodness of the model was verified by the information of
different robust statistics and cited indices, directly from the AMOS
output (Byrne, 2001). For each set of fit statistics, the AMOS soft-
ware provides three rows. The first row focuses on the hypoth-
esised model under test (i.e., proposed model); the second row
focuses on the saturated model, and the third row, on the inde-
pendence model. These three models were considered as repre-
senting points on a continuum, with the independence model at
one extreme, in which all correlations among variables are zero and
is the most restricted. The saturated model is located at the other
extreme, in which the number of estimated parameters equals the
number of data points, the least restricted; the hypothesised or
proposed model is somewhere in between.
Focussing on the first set of fit statistics, the labels NPAR
(number of parameters), CMIN (minimum discrepancy) are most
commonly expressed as a chi-square statistic (c2), DF (Degrees of
Freedom), P (Probability Value), and CMIN/DF. The minimum
discrepancy function (CMIN) is a direct consequence of adjusting to
the procedure of maximum credibility. If this function is weighted
by the sample size, the chi-square statistic is obtained, which can be
used to verify the null hypothesis, that is, where all the residuals are
zero. Reduced values, such as in this case, represent lesser
discrepancy between what is observed and what is predicted by the
model, and therefore, a better fit (Table 7).

Conservation Table 4
Regression coefficients.

Estimate S.E. C.R. P

Intention ) Con 0.417 0.216 1.934 0.053
Formation Intention i4 ) Con 0.401 0.198 2.028 0.043
Behaviour i6 ) Con 0.848 0.238 3.564 ***
i10 ) Con 0.338 0.119 2.833 0.005
i14 ) Con 1.000
i3 ) Int 2.119 0.999 2.122 0.034
i11 ) Int 1.352 0.750 1.801 0.072
Outreach i16 ) Int 1.000
activities
i13 ) Int 1.145 0.643 1.782 0.074
Behaviour ) Int 3.476 1.766 1.968 0.049

Fig. 1. Structural model. S.E., standard error; C.R., critical ratio; P, probability.

Please cite this article in press as: Rodríguez-Barreiro, L.M., et al., Approach to a causal model between attitudes and environmental behaviour.
A graduate case study, Journal of Cleaner Production (2012), http://dx.doi.org/10.1016/j.jclepro.2012.09.029
6 L.M. Rodríguez-Barreiro et al. / Journal of Cleaner Production xxx (2012) 1e10

Table 5 0; 0;
Variances. 1 1
e4 i4 i3 e3
Estimate S.E. C.R. P 0;
1 0;
0; 1
Con 0.396 0.152 2.609 0.009 e6 i6 0
i11 e11
Ee4 0.014 0.020 0.677 0.499 0;
1
Con Int 0;
e4 0.643 0.121 5.293 ***
e10 i10 i13
1
e13
e10 0.204 0.040 5.111 *** 1 1
0;
e6 0.661 0.138 4.776 *** 1 1 0;
e14 i14 1
e14 0.418 0.111 3.757 *** 0; i16 e16
e11 0.763 0.147 5.196 ***
e13 0.595 0.113 5.256 *** Ee4
e3 0.090 0.059 1.520 0.129
e16 0.854 0.160 5.342 ***
e22 2.244 0.485 4.628 ***

0;

e22
1 Behaviour
Researchers have addressed the chi-square limitations by
developing goodness-of-fit indexes that take a more pragmatic
approach to the evaluation process. In any case, the chi-square is Fig. 3. Final model. Note: i, Item of the scale; Con, Conservation factor; Int, Intention to
highly dependent on the size of the sample, and hence the need to act factor; e, Error.
address other indices, which are also of a global nature, but more
descriptive and less dependent on the number of respondents
taking part in the research. between the fitted covariance matrix in the analysed sample and
The next set of goodness-of-fit statistics is presented now, which the expected covariance matrix that would be obtained in another
can be classified as incremental or comparative indexes of fit sample of equivalent size. The model presenting the lowest ECVI
(Table 8). value exhibits the greatest potential for replication. The value of
Bentler (1990) revised the NFI (normed fit index) to take sample 1.825 is compared to the values for the saturated and independent
size into account and proposed the comparative fit index (CFI). models; as 1.825 is lower, it can be concluded that the proposed
Values for both the NFI and CFI range from zero to 1.00 point for model offers the best fit for such data.
a well-fitting model (Table 8). Values close to one for the NFI and The Conservation factor has an indirect effect on Behaviour,
CFI indices would indicate that the evaluated model is consistent since such an influence is measured by the Intention to act factor.
with the theoretical model generating the structures observed The proposed model exhibits features that make it very suitable:
among the variables. The interest lies in the fact that it tends to consistency with the theory, good fit and parsimony.
reject correct models when the sample is small in size. The same
thing happens with the IFI (incremental fit index). The IFI was
5. Discussion
developed to address the issues of parsimony and sample size,
which was known to be associated with the NFI. As such, its
The obtained results are analysed in light of other relevant
computation is basically the same as NFI, with the exception that
research works.
degrees of freedom are taken into account. Thus, it is not surprising
The EAU Scale was initially developed for the detection of
that of the value found herein IFI ¼ 0.767 is consistent with that of
possible different environmental attitudes of the university
the CFI in reflecting a well-fitting model.
undergraduate students taking environmental courses with respect
The statistics Akaike’s information criterion (AIC) and the
to those without environmental education. The validity and reli-
BrowneeCudeck criterion (BCC) reflect the extent to which
ability of the EAU Scale, that was already tested in the first study
parameter estimates from the original sample will cross-validate in
(Fernández-Manzanal et al., 2007), is clearly endorsed as the
future samples. The basic difference among these indexes is that
assessment of the items is seen to have the same orientation as
BCC imposes greater penalties than AIC for model complexity. Both
before. The scale of this research work offers higher scores for many
criteria address the issue of parsimony in the assessment of model
of the statements, and these values are slightly higher than those
fit; as such, statistical goodness of fit and number of estimated
obtained in the research work developed with undergraduate
parameters are taken into account.
students (Fernández-Manzanal et al., 2007). The categories or
The expected cross-validation index (ECVI) is central to the next
factors with a better valuation were C1 (Formation) and C2
cluster of fit statistics. Specifically, it measures the discrepancy
(Outreach activities). This was not surprising, since the graduates
from this sample received education on environmental issues and
have undergone some external training. It is thus possible to say
Table 6 that university graduates have assimilated the environment as part
Standardised regression weights and squared multiple correlations. of their values and concerns. Other studies (Schultz and Zelezny,
Estimate Squared 1999; Amérigo and González, 2000; Zelezny and Schultz, 2000)
Intention ) Con 0.914 0.835
i4 ) Con 0.300 0.090
i6 ) Con 0.549 0.301 Table 7
i10 ) Con 0.427 0.182 CMIN (minimum discrepancy function).
i14 ) Con 0.697 0.486
i3 ) Int 0.897 0.805 Model NPAR CMIN DF P CMIN/DF
i11 ) Int 0.406 0.165 Proposed model 28 52.026 26 0.002 2.001
i16 ) Int 0.297 0.088 Saturated model 54 0.000 0
i13 ) Int 0.392 0.154 Independence model 9 137.778 45 0.000 3.062
Behaviour ) Int 0.554 0.307
NPAR, number of parameters; DF, degrees of freedom; P, probability value.

Please cite this article in press as: Rodríguez-Barreiro, L.M., et al., Approach to a causal model between attitudes and environmental behaviour.
A graduate case study, Journal of Cleaner Production (2012), http://dx.doi.org/10.1016/j.jclepro.2012.09.029
 L.M. Rodríguez-Barreiro et al. / Journal of Cleaner Production xxx (2012) 1e10
Table 8
NFI/IFI/CFI/AIC/BCC/ECVI.
Model NFI Delta1 IFI Delta2 CFI AIC BCC ECVI
Proposed model 0.622 0.767 0.719 108.000 119.455 1.825
Saturated model 1.000 1.000 1.000 108.026 130.041 1.831
Independence model 0.000 0.000 0.000 155.778 159.451 2.640
NFI, normed fit index; IFI, incremental fit index; CFI, comparative fit index; AIC,
Akaike’s information criterion; BCC, BrowneeCudeck criterion; ECVI, expected
cross-validation index.
coincide in indicating the interest displayed for the environment by
various social agents, irrespective of age or educational
background.
The value obtained for Cronbach’s alpha is 0.78, slightly lower
than the value obtained with undergraduate students and in this
case confirmed the predictions of DeVellis (1991), alpha may
decrease somewhat when the scale is administered to a sample
other than the one used for its development. Likewise, in a recent
paper published (Yurt et al., 2010) in which this scale was used,
validity and reliability results are presented for both the statements
and the factors or categories.
The sample employed in the present research work is different
(all participants were graduated) from the sample that provided
the first results (Fernández-Manzanal et al., 2007), however, the
number and features of the items have not been changed. This
criterion is in agreement with Hawcroft and Milfont (2010), who
state that research into psychometrics suggests that an inconsistent
use of scales may well be problematic. For example, it is known that
even a small change in the wording of one item can have
a substantial effect on how people respond to a scale. Changes in
the response format (e.g., variation in the number of points on
a Likert scale offered to participants) are also known to effect
responses.
Nevertheless, these results suggest that future researchers
should be aware of the fact that samples from different socioeco-
nomic backgrounds may respond differently to the EAU Scale.
With respect to the second question on environmental actions
(see Section 2), the most common answers regard water and energy
wastage. Also important is the percentage of respondents that take
actions associated with recycling into consideration. We interprete
that such behaviours are facilitated by significant support at an
institutional level: both the Zaragoza City Council and the
Government of Aragon’s Ministry of the Environment have helped
provide the means for people to participate in such actions (recy-
cling bins have been brought closer to people’s homes and water
and energy-saving campaigns have been organised). These results
are similar to those obtained with the sample consisting of
undergraduate students, although they also shown favourable
results with respect to the use of public transportation.
Other environmental behaviours that obtained lower scores in
the questionnaire, such as the use of recycled paper, the utilisation
of public transportation, the use of ecological consumption prod-
ucts, search of environmental information, or the use of free time to
carry out activities for the protection of the environment, may be
due to the lack of adequate resources and the effort required to put
them into practice. In this sense, a high score for attitude is not
consistent with environmental behaviour. One aspect that this
study has in common with previous researches is: “. however,
although Europeans are aware of the need to protect the environ-
ment, their green attitudes do not always translate into environ-
mentally friendly behaviour and concrete actions” (European
Commission, 2008). Stern (2000) explains that pro-
environmental behaviours are very complex and have an intricate
network of influences, with attitudes being but a part of this
Please cite this article in press as: Rodríguez-Barreiro, L.M., et al., Approach to a causal model between attitudes and environmental behaviour.
A graduate case study, Journal of Cleaner Production (2012), http://dx.doi.org/10.1016/j.jclepro.2012.09.029
7
network. For Stern, “attitudinal factors have an important predic-
tive function for behaviour when they are not greatly constrained
by context or personal abilities”. A reduction of the public trans-
portation in the case of the graduates, can be explained in the
context of a higher availability of private car.
These results also suggest, as in the case of environmental
attitudes, that future researchers should be aware of the fact that
samples from different socioeconomic backgrounds and different
formation may respond differently to the behavioural
questionnaire.
The last aspect addressed is the one that provides the title for this
paper. A causal model has been tested for the relationships between
environmental attitudes and behaviours. Its diagram shows the
presence of an indirect effect from one of the factors on the scale,
Conservation, on the dependent variable, Behaviour. Such an influ-
ence is modulated by a second factor, Intention to act or willingness
to engage in such behaviour. The model meets the appropriate
characteristics in terms of theoretical significance, fit and parsi-
mony. According to the data, Behaviour variance is explained as
being 30.7%. Intention is explained in the model as being 83.5%.
These percentages seem respectable in light of the theories dis-
cussed concerning the influence of attitudes on behaviours, in
particular the theory of Ajzen and Fishbein (2000) on the intention
of behaviour as an intermediate position between the attitudes and
the behaviours. Regarding the meaning of items in EAU Scale cate-
gories that defined the proposed model (Conservation and Intention
to act), these results are in agreement with the conclusions of the
work of Juárez-Nájera et al. (2010) that indicate that “pro-environ-
mental behaviour is defined as actions contributing to environ-
mental conservation, or human activity intended to protect natural
resources, or at least reduce environmental deterioration. These
definitions include a deliberate component, or intentionality”.
Other authors (Armitage and Conner, 2001; Bamberg and
Schmidt, 2003; Webb and Sheeran, 2006; Bamberg and Möser,
2007; Rivis et al., 2009) have analysed different models in order
to study the relationship between intention and behaviour.
Different results have been obtained because of different conditions
(the applied models, the considered variables, the samples, etc. are
different). Nevertheless, all of these studies show that pro-
environmental behavioural intention mediate the impact of all
other psychosocial variables on pro-environmental behaviour.
Correlational studies with modern statistical methods show that
correlation among intention and behaviour is in the range from 27%
(Bamberg and Möser, 2007) to 66% (Webb and Sheeran, 2006). The
result obtained herein shows that intention predicts 30.7% variance
of behaviour and intention is explained in 83.5% by conservation
factor. The variance of behaviour indicates that, like in referred
works and as stated Bamberg and Möser (2007), “the processes
contributing to the actual enactment of pro-environmental
behavioural intention are not fully understood”.
With respect to the independent variables of attitudes used in
this model, it can be concluded that the variable conservation is the
most important in the causal relationship of the studied environ-
mental behaviours. Obviously, the environmental formation is
a necessary, however not a sufficient precondition for developing
pro-environmental behaviour (Lozano García et al., 2006; Stephens
and Graham, 2010).
Considering the importance of conservation factor in the model
and that all of respondents have received environmental formation
and have participated in outreach activities (field activities in the
EAU scale, see Appendix 1), it is suggested that environmental
formation could include conservation and preservation aspects, as
for instance integrity of animal and plant species, as well as indi-
vidual and social well-being, and safety of present and future
human generations as proposed by Juárez-Nájera et al. (2010).
8 L.M. Rodríguez-Barreiro et al. / Journal of Cleaner Production xxx (2012) 1e10

Although the model has to be replicated with further samples,
as done with the scale of attitudes, its provisional validation can be
taken as empirical evidence in favour of the theories proposed, in
particular, those proposed regarding the connection between atti-
tudes and behaviours. According to MacCallum (2003), if the model
is modified and it is found to fit properly, it should then be validated
with further data.
6. Conclusions
The first conclusion from the previous results indicates that the
environmental attitude of the University of Zaragoza graduates
presents a high valuation, slightly higher than that obtained by
undergraduate students of the same university. The highest values
correspond to the Formation and Outreach activities factors. These
results indicated the great importance given by graduates to
education on environmental issues. Furthermore, these results are
also relevant in terms of environmental attitudes.
The environmental behaviour related to daily life has a distinct
valoration, since it is the most practiced in relation to family habits
or to the existence of recycling containers close to home. As it has
been shown throughout this study and in view of the data obtained,
as a second conclusion it can be pointed out that, although envi-
ronmental training is indeed needed, is not sufficient to sustain
environmental behaviour. Once environmental formation is
acquired e as it is the case of the graduates in the sample e
providing adequate resources and habits in everyday academic life
turns to be a very significant aspect for environmental behaviour.
With respect to the third conclusion, referred in the title of this
work, the most conclusion to the causal model is that the
Conservation factor is the factor that has shown the closest rela-
tionship with Behaviour. This relationship is shown through the
Intention to act and, in the model, this intermediate point explains
a high percentage of environmental behaviour. Environmental
behaviour is considered to be a set of effective, deliberate, and
anticipated actions aimed at accepting responsibility for conser-
vation and preservation of nature and natural resources.
The obtained results provide a positive empirical evidence of the
relationship between attitudes and behaviour, and represent
a stimulus to continue investigating into the heart of the matter and
renewed theoretical framework, which is the extended version of
the reasoned action. The causal model confirms the relationship
between the expected wish, the intention of sustainable behaviour,
and the real behaviour.
In light of the previous comments, the university community
shall incorporate the conservation perspective in all of its fields of
study, both in interdisciplinary education programmes and
research programmes. Considering that environmental education is
directly linked to pro-environmental behaviour, the university shall
provide a favourable set of opportunities, resources and situations,
so that learning and investigation of environmental behaviour can
be put into practice.
Acknowledgements
This research team thanks the Vice-Chancellor’s Office for
Research at the University of Zaragoza for the opportunity to carry
out this project. Thanks are extended to the graduates for their
participation in completing the questionnaires.

Appendix 1. EAU scale

Put an X in the chosen option

Strongly agree Agree Indifferent Disagree Strongly disagree
1. Environmental education for people cannot help
to resolve environmental problems, only
technology can do this.
2. Universities should schedule more field activitiesa
because they help to understand the matter better.
3. I am willing to consume less and go without some
comforts if it helps to protect the environment.
4. I believe that environmental problems are exaggerated,
nature balances out over time.
5. When I buy a product, I assess the type of packaging
and choose one that is recyclable.
6. The progress of a district should not be held up with
the excuse of protecting some birds.
7. I like to participate in field activitiesa because it is a
good way of understanding the environment in which
I live.
8. I believe that information is increasingly necessary to
be aware of the effects our actions have on the environment.
9. I try to choose subjects that deal with matters related to
the environment because I feel that I do not know enough.
10. Nowadays, the laws and government regulate and control
so much that there is very little contamination.
11. If I have to choose between the construction of a motorway
and the protection of a plant species, I choose the motorway.
12. Field activitiesa help to be more in touch with nature.
13. In order to have a more technologically developed society,
I am willing to tolerate noise caused by vehicles.
14. The benefits of modern consumer products are
more important than the contamination caused by their
production and use.
15. Field activitiesa help to increase awareness of
environmental matters.

Please cite this article in press as: Rodríguez-Barreiro, L.M., et al., Approach to a causal model between attitudes and environmental behaviour.
A graduate case study, Journal of Cleaner Production (2012), http://dx.doi.org/10.1016/j.jclepro.2012.09.029
 L.M. Rodríguez-Barreiro et al. / Journal of Cleaner Production xxx (2012) 1e10 9

(continued )

Strongly agree Agree Indifferent Disagree Strongly disagree

16. Even if public transport were more efficient than
it currently is, I would still prefer to use my own car.
17. In my opinion, the more people know about the natural environment
the better their defence attitude.
18. Environmental education activities are only useful for children.
19. We should try to conserve the earth’s plants and animals, even though
it is expensive.
20. Field activitiesa are a waste of time, the most important thing is class work.
a
Field activities refer also to outreach activities in companies as well as those who have developed field activities directly connected with contents of the curriculum. The
original term of field activities has been maintained in agreement with the discussion of results presented in Section 5.

Appendix 2. Questionnaire on environmental behaviours

Yes No
1. During my studies, I have chosen optional subjects or supplementary activities related to the environment.
2. I use recycled paper.
3. On at least one occasion I have used my free time or my holidays to carry out activities related to protecting the environment.
4. If I come across an empty room with the lights on, I always turn them off.
5. I put waste glass and plastic containers in recycle bins.
6. I throw cigarette ends and other rubbish down the toilet that should not be disposed of in that way, as long as it won’t block the drain.
7. I put waste paper in recycle bins.
8. I use my car to get to work even though I could use other means of transport.
9. I turn off the shower whilst I soap myself.
10. Before I buy something I read the product information to check the type of product and I buy the most ecological one.
11. I try to keep up to date on environmental issues.
12. I read and look out for information related to the environment.

References
Ajzen, I., 2001. Nature and operation of attitudes. Annual Review of Psychology 52,
27e58.
Ajzen, I., Fishbein, M., 1980. Understanding Attitudes and Predicting Social Behav-
iour. Prentice-Hall, Englewood Cliffs, N.J.
Ajzen, I., Fishbein, M., Stroebe, W., Hewstone, M., 2000. Attitudes and the attitude-
behavior relation: reasoned and automatic processes. European Review of
Social Psychology. Wiley, Chichester, England.
Akaike, H., 1987. Factor analysis and AIC. Psychometrika 52, 317e332.
Amérigo, M., González, A., 2000. Los valores y las creencias medioambientales en
relación con las decisiones sobre dilemas ecológicos ([Environmental values
and beliefs and their relationship with decisions in ecological dilemmas]).
Estudios de Psicología 22 (1), 65e73.
Arbuckle, J.L., 2003. Amos 5.0 Update to the Amos User’s Guide. SmallWaters,
Chicago.
Arbuckle, J.L., Wothke, W., 1999. Amos 4.0 User’s Guide. SmallWaters, Chicago.
Armitage, C.J., Conner, M., 2001. Efficacy of the Theory of Planned Behaviour:
a meta-analytic review. The British Journal of Social Psychology 40, 471e499.
Bamberg, S., Schmidt, P., 2003. Incentives, morality, or habit? Predicting students’
car use for university routes with the models of Ajzen, Schwartz, and Triandis.
Environmental and Behavior 35, 264e285.
Bamberg, S., Möser, G., 2007. Twenty years after Hines, Hungerford, and Tomera:
a new meta-analysis of psycho-social determinants of pro-environmental
behaviour. Journal of Environmental Psychology 27, 14e25. http://www.
sciencedirect.com/science/article/pii/S0272494406000909.
Batista, J.M., Coenders, G., 2000. Modelos de ecuaciones estructurales (Structural
equations models). La Muralla, Madrid.
Bentler, P.M., 1990. Comparative fit indexes in structural models. Psychological
Bulletin 107, 238e246.
Bentler, P.M., 1995. EQS Structural Equations Program Manual. Multivariate Soft-
ware, Inc., Encino, CA.
Bentler, P.M., Dudgeon, P., 1996. Covariance structure analysis: statistical practice
theory and directions. Annual Review of Psychology 47, 563e592.
Bollen, K.A., 1989. Structural Equations with Latent Variables. John Wiley and Sons,
New York.
Browne, M.W., Cudeck, R., 1989. Single sample cross-validation indices for covari-
ance structures. Multivariate Behavioral Research 24, 445e455.
Browne, M.W., Cudeck, R., 1993. Alternative ways of assessing model fit. In:
Bollen, K., Long, J.S. (Eds.), Testing Structural Equation Models. Sage Publica-
tions, Newbury Park, CA, pp. 136e162.
Byrne, B.M., 2001. Structural Equation Modeling with Amos. Basic Concepts,
Applications and Programming. Lawrence Erlbaum Associates, Mahwah, NJ.
Courtenay-Hall, P., Rogers, L., 2002. Gaps in mind: problems in environmental
knowledge-behaviour modeling research. Environmental Education Research 8
(3), 283e297.
Crano, W.D., Prislin, R., 2006. Attitudes and persuasion. Annual Review of
Psychology 57, 345e374.
Crawley, F.E., Koballa, T.R., 1994. Attitude research in science education: contem-
porary models and methods. Science Education 78 (1), 35e55.
Desha, C.J., Hargroves, K., 2010. Surveying the state of higher education in energy
efficiency, in Australian engineering curriculum. Journal of Cleaner Production
18, 652e658.
DeVellis, R.F., 1991. Scale development. Theory and applications. In: Applied Social
Research Methods Series, vol. 26. Sage, Newbury Park, CA.
Dunlap, R.E., Van Liere, K.D., Mertig, A.D., Jones, R.E., 2000. Measuring endorsement
of the New Ecological Paradigm: a revised NEP scale. Journal of Social Issues 56
(3), 425e442.
European Commission, March, 2008. Special Eurobarometer 295.Attitudes of
Europeans Citizens towards the Environment. Available from: http://ec.europa.
eu/public_opinion/archives/ebs/ebs_295_en.pdf (accessed 04.01.11.).
Fernández-Manzanal, R., Rodríguez Barreiro, L. Ma, Carrasquer, J., 2007. Evaluation
of environmental attitudes. Analysis and results of a scale applied to university
students. Science Education 91, 988e1009.
Ferrer-Balas, D., Lozano, R., Huisingh, D., Buckland, H., Ysern, P., Zilahy, G., 2010.
Going beyond the rhetoric: system-wide changes in universities for sustainable
societies. Journal of Cleaner Production 18 (7), 607e610.
Garson, D., 2011. Structural Equation Modeling. http://faculty.chass.ncsu.edu/
garson/PA765/structur.htm (accessed 12.10.11).
Hawcroft, L.J., Milfont, T.L., 2010. The use (and abuse) of the new environmental
paradigm scale over the last 30 years: a meta-analysis? Journal of Environ-
mental Psychology 30, 143e158.
Hines, J.M., Hungerford, H.R., Tomera, A.N., 1986/87. Analysis and synthesis of
research on responsible environmental behaviour: a meta-analysis. Journal of
Environmental Education 18, 1e8.
Jones, P., Selby, D., Sterling, S., 2010. Sustainability Education. Perspectives and
Practice across Higher Education. Earthcan, London.
Jöreskog, K.G., Sörbom, D., 1996. Lisrel 8: User’s Reference Guide. SSI Scientific
Software International, USA.
Juárez-Nájera, M., Rivera-Martínez, J.G., Hafkamp, W.A., 2010. An explorative socio-
psychological model for determining sustainable behavior: pilot study in
German and Mexican Universities. Journal of Cleaner Production 18, 686e694.
Kollmus, A., Agyeman, J., 2002. Mind the Gap: why do people act environmentally
and what are the barriers to pro-environmental behavior? Environmental
Education Research 8 (3), 239e260.
Lin, P.-C., Huang, Y.-H., 2012. The influence factors on choice behavior regarding
green products based on the theory of consumption values. Journal of Cleaner
Production 22, 11e18.
Lidgren, A., Rodhe, H., Huisingh, D., 2006. A systemic approach to incorporate
sustainability into university courses and curricula. Journal of Cleaner Produc-
tion 14 (9e11), 797e809.
Lozano, R., 2006. Incorporation and institutionalization of SD into universities:
breaking through barriers to change. Journal of Cleaner Production 14,
787e796.
Lozano, R., 2012. Organizational learning and creativity as means to foster
sustainability in universities. Higher education for sustainability: a global

Please cite this article in press as: Rodríguez-Barreiro, L.M., et al., Approach to a causal model between attitudes and environmental behaviour.
A graduate case study, Journal of Cleaner Production (2012), http://dx.doi.org/10.1016/j.jclepro.2012.09.029
10 L.M. Rodríguez-Barreiro et al. / Journal of Cleaner Production xxx (2012) 1e10
overview of commitment and progress. In: Higher Education in the World 4.
Higher Education’s Commitment to Sustainability: From Understanding to
Action. GUNi Series on the Social Commitment of Universities, vol. 4. CPI Wil-
liam Clowes, Great Britain.
Lozano García, F.J., Kevany, K., Huisingh, D., 2006. Sustainability in higher education:
what is happening? Journal of Cleaner Production 14 (9e11), 757e760.
MacCallum, R.C., 2003. Working with imperfect models. Multivariate Behavioral
Research 38 (1), 113e139.
Mueller, R.O., 1996. Basic Principles of Structural Equation Modeling. An Introduc-
tion to LISREL and EQS. Springer-Verlag, New York.
Murillo-Luna, J., Garcés-Ayerbe, C., Rivera-Torres, P., 2011. Barriers to the adoption of
proactive environmental strategies. Journal of Cleaner Production 19 (13),
1417e1425.
Raftery, A.E., 1995. Bayesian model selection in social research. In: Mariden, P.V.
(Ed.), Sociological Methodology 1995. Jossey-Bass, San Francisco, pp. 111e163.
Rivis, A., Sheeran, P., Armitage, C.J., 2009. Expanding the affective and normative
components of the theory of planned behavior: a meta-analysis of anticipated
affect and moral norms. Journal of Applied Social Psychology 39, 2985e3019.
Ruiz, M., 2000. Introducción a los modelos de ecuaciones estructurales (An Intro-
duction to Structural Equation Modeling). UNED, Madrid.
Satorra, A., Bentler, P., 1994. Corrections to test statistics and standard errors in
covariance structure analysis. In: von Eye, A., Clogg, C.C. (Eds.), Latent Variables
Analysis: Applications for Developmental Research. Sage, Thousand Oaks, CA,
pp. 399e419.
Schwartz, S.H., 1977. Normative influences on altruism. In: Berkowitz, L. (Ed.),
Advances in Experimental Social Psychology, vol. 10. Academic Press, San Diego,
pp. 221e279.
Schultz, P.W., 2001. The structure of environmental concern: concern for self, other
people and the biosphere. Journal of Environmental Psychology 21, 327e339.
Please cite this article in press as: Rodríguez-Barreiro, L.M., et al., Approach to a causal model between attitudes and environmental behaviour.
A graduate case study, Journal of Cleaner Production (2012), http://dx.doi.org/10.1016/j.jclepro.2012.09.029
Schultz, P.W., Zelezny, L., 1999. Values as predictors of environmental attitudes:
evidence for consistency across 14 countries. Journal of Environmental
Psychology 19, 255e265.
Stern, P.C., 2000. Toward a coherent theory on environmentally significant behavior.
Journal of Social Issues 56 (3), 407e424.
Stern, P.C., Dietz, T., Abel, T., Guagnano, G.A., Kalof, L., 1999. A value belief-norm
theory of support for social movements: the case of environmental concern.
Human Ecology Review 6, 81e97.
Stephens, J.C., Graham, A.C., 2010. Toward an empirical research agenda for
sustainability in higher education: exploring the transition management
framework. Journal of Cleaner Production 18, 637e644.
Tilbury, D., 2012. Higher education for sustainability: a global overview of
commitment and progress. In: Higher Education in the World 4. Higher
Education’s Commitment to Sustainability: From Understanding to Action.
GUNi Series on the Social Commitment of Universities, vol. 4. CPI William
Clowes, Great Britain.
Triandis, H.C., 1980. Values, attitudes, and interpersonal behavior. In: Howe, H.E.,
Page, M.M. (Eds.), Nebraska Symposium on Motivation 1979. University of
Nebraska Press, Lincoln, pp. 195e259.
Webb, T.L., Sheeran, P., 2006. Does changing behavioral intentions engender
behavior change? A meta-analysis of the experimental evidence. Psychological
Bulletin 132, 249e268.
Wiegel, R.H., Wiegel, J., 1978. Environmental concern: the development of
a measure. Environmental and Behavior 10, 3e15.
Yurt, O., Cevher-Kalburana, N., Kandir, A., 2010. Investigation of the environmental
attitudes of the early childhood teacher candidates. Procedia Social and
Behavioral Sciences 2 (2), 4977e4984.
Zelezny, L., Schultz, P.W., 2000. Psychology of promoting environmentalism:
promoting environmentalism. Journal of Social Issues 56 (3), 365e371.