10.

1177/0022022102239152 JOURNAL OF CROSS-CULTURAL PSYCHOLOGY Spini / MEASUREMENT EQUIVALENCE OF VALUES

MEASUREMENT EQUIVALENCE OF 10 VALUE TYPES FROM
THE SCHWARTZ VALUE SURVEY ACROSS 21 COUNTRIES
DARIO SPINI
Universities of Lausanne and Geneva
Equivalence of measurement across cultures is a prerequisite for the generalization of an instrument. The
measurement equivalence of 10 value types derived from Schwartzs structural model of values and mea-
sured with the Schwartz Value Survey questionnaire is evaluated in 21 countries. Based on previous research
by Schwartz and colleagues, the measurement equivalence of the 10 value types is tested separately using
nested multigroup confirmatory factor analyses. Results indicate that it is possible for most value types to
reach acceptable levels of configural and metric equivalence; only the dimension of Hedonismis rejected at
these two levels of equivalence. Four value types (Benevolence, Conformity, Self-Direction, and Universal-
ism) also showfactor variance equivalence. The hypotheses of scalar and reliability equivalence are rejected
for all value types. Indications are also given of the number of items to be included for measuring the value
types at the different levels of equivalence.
Comparison is one of the main activities of cross-cultural research. However, psychologi-
cal measures have no intrinsic properties warranting the assumption that they can be mea-
sured in an equivalent manner across cultures. Equivalence of structure (number and rela-
tionships among psychological constructs or latent variables) and measurement
(relationships between latent variables and their indicators) have to be conceived at different
levels so that factors, means, variances, or covariances can be compared in different settings
(Van de Vijver & Leung, 1997a). Unfortunately, researchers in cross-cultural psychology
often compare means or correlations across cultural groups without testing the equivalence
of the constructs and measures. The major goal of this article is to provide a test of equiva-
lence of an instrument that has been widely used in cross-cultural psychology, namely, the
Schwartz Value Survey, and to provide an application of multigroup structural equation
models (Bollen, 1989b; Jreskog, 1971) in order to test different levels of equivalence using
an international survey administrated to student populations living in 21 countries.
3
AUTHORS NOTE: The data presented here were collected with the help of the following colleagues, students, and friends: Daniel
Roselli (Argentina), Velina Topalova (Bulgaria), Monica Herrera and Marguerite Lavalle (Canada), Victor Espinoza, Isabell
Kempf, and Pablo Salvat Bologna (Chile), Danilo Prez Zumbado (Costa Rica), Yapo Yapi (Cte dIvoire), Maaris Raudsepp (Esto-
nia), Eva Green (Finland), Joane Cotasson and Elena Lozano (France), Jyoti Verma (India), Emda Orr and Yossi Mana (Israel), Luisa
Campanile and Annamaria Silvana de Rosa (Italy), Kazuhisa Kobayashi and Masaaki Orii (Japan), Araceli Otero de Alba (Mexico),
Cecilia Gastardo-Conaco (Philippines), Jorge Correia Jesuino (Portugal), Aliou Sall (Senegal), Alex Amati and William Onzivu
(Uganda), Glynis Breakwell (United Kingdom), and Gordana Jovanovic (Yugoslavia). I ammost grateful for the work they agreed to
do. This study was supported financially by the Socit Acadmique de Genve. Parts of the analyses were carried out during a visit
to the Institute for Research in Social Science, University of North Carolina at Chapel Hill, during which I benefited fromthe super-
vision of Kenneth Bollen (under a fellowship fromthe Swiss National Science Foundation for young researchers). My thanks go also
to ShalomSchwartz and his collaborators for providing the Schwartz Value Survey in several languages and for their comments on a
previous manuscript. Finally, the editorial assistance of Ian Hamilton was greatly appreciated, as were the thoughtful comments of
Paolo Ghisletta, of two anonymous reviewers, and the critical and encouraging feedback of Fons van de Vijver during the process of
revision. Please address correspondence to Dario Spini, Institute for Life Course and Life Styles Studies, Btiment Provence, Uni-
versity of Lausanne, 1015 Lausanne, Switzerland; e-mail: [email protected]
JOURNAL OF CROSS-CULTURAL PSYCHOLOGY, Vol. 34 No. 1, January 2003 3-23
DOI: 10.1177/0022022102239152
2003 Western Washington University
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
VALUES ACROSS CULTURES
Value theory has been an important issue in cross-cultural psychology since Rokeachs
(1973) seminal work. Values have since then been used as independent variables to under-
stand attitudes and behavior and as dependent variables of basic differences among social
groups and categories. This last property has encouraged cross-cultural psychologists to
seek common dimensions of values and to study differences among cultures.
Hofstede (1980, 1982, 1983) took an important first step in the study of values across cul-
tures in his study on work-related values across 50 countries. Concerning differences among
countries and cultures, a series of studies followed up on Rokeachs (1973) and Hofstedes
(1980) work. Ng et al.s (1982) study, for example, used the Rokeach Value Survey
(Rokeach, 1967) with some additional items in order to investigate cultural differences
among 9 Asian countries. The Rokeach Value Survey instrument has, however, been consid-
ered as biased toward Western values (Hofstede &Bond, 1984) and limited in the number of
dimensions assessed (Braithwaite & Law, 1985). This is one reason why several studies
aimed at determining the number and content of value dimensions needed to account for
interindividual and cross-cultural differences appeared in the 1980s (see Bond, 1988; Chi-
nese Culture Connection, 1987).
Research inspired by the works of Rokeach (1973) and Hofstede (1980) was vital to the
development of a cross-cultural theory of values, as it revealed the possibility of obtaining
valid etic dimensions of values. It also allowed the development of appropriate research
methods still in use today (see Van de Vijver & Leung, 1997a). Nevertheless, these studies
remained data driven and did not propose an integrative and universal theory of values. It was
in the early 1990s that another important step forward was made with Schwartzs formula-
tion of a theory describing the universal content and structure of values.
A THEORY OF CONTENT AND STRUCTURE OF VALUES
Schwartz (1992, 1994) defined a model that further detailed the content and structure of
values on the basis of empirical cross-cultural studies (Schwartz, 1992; Schwartz & Bilsky,
1987, 1990). Values were defined as desirable transsituational goals, varying in importance,
that serve as guiding principles in the life of a person or social entity (Schwartz, 1994, p. 21).
Three universal requirements were thought to be at the root of values: needs of individuals as
biological organisms, requisites of coordinating social interaction, and requirements for the
functioning of society and the survival of groups.
Fromthese three basic goals, 10 motivational value types were derived: (a) Achievement:
personal success through the demonstration of competence according to social standards; (b)
Benevolence: concern for the welfare of close others in everyday interaction; (c) Confor-
mity: restraint of actions, inclination, and impulses likely to upset or harmothers and violate
social expectations or norms; (d) Hedonism: pleasure and sensuous gratification for oneself;
(e) Power: attainment of social status and prestige, and control or dominance over people and
resources; (f) Security: safety, harmony, and stability of society, of relationships, and of the
self; (g) Self-Direction: independent thought and action; (h) Stimulation: excitement, nov-
elty, and challenge in life; (i) Tradition: respect, commitment and acceptance of the customs
and ideas that ones culture or religion impose on the individual; (j) Universalism: under-
standing, appreciation, tolerance, and protection for the welfare of all people and for nature.
These 10 value types were measured using the Schwartz Value Survey, an instrument
4 JOURNAL OF CROSS-CULTURAL PSYCHOLOGY
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
composed of 57 value items (see Table 1). This set of values is supposed to cover the main
comparable value dimensions across cultures.
Schwartzs (1992) theory also defines the structure of the motivational types in a
bidimensional space. The concept of structure refers here to the conflicts and compatibilities
among value types. As this structural part of the model is not tested here, its description will
therefore not be developed.
EQUIVALENCE OF VALUE TYPES ACROSS CULTURES
Schwartz tested his theory using similarity structure analysis (Borg & Lingoes, 1987;
Guttman, 1968), an application of multidimensional scaling. This technique makes the
assumption that the similarities and dissimilarities among a set of objects can be represented
in a dimensional space. In the case of similarity structure analysis, the similarities are often
measured by the correlations among items. The best representation of the starting matrix is
then to be found and a stress measure is provided to compare the structural representation
obtained with the original similarities (see Kruskal & Wish, 1990). In the case of Schwartz
(1992), a two-dimensional space (sometimes three, especially in the case of Japan) was cho-
sen as representative of the correlations among the value items. Taking into account
Schwartzs indications on the reliability of the Schwartz Value Survey items (see Table 1),
Spini / MEASUREMENT EQUIVALENCE OF VALUES 5
TABLE 1
List of Value Items by Motivational Value Types
Value Type Value Item
Achievement Successful (93/84), capable (84/75), ambitious (82/74), influential (74/68),
intelligent (64/56), self-respect (35/31)
Benevolence Helpful (95/86), honest (91/82), forgiving (85/77), loyal (80/71),
responsible (77/72), true friendship (63/58), a spiritual life (55/52),
mature love (51/47), meaning in life (41/39)
Conformity Politeness (92/84), honoring parents and elders (90/81), obedient (88/79),
self-discipline (82/75)
Hedonism Pleasure (95/86), enjoying life (94/85), self-indulgent (-)
Power Social power (97/88), authority (94/86), wealth (92/83), preserving my public
image (62/57), social recognition (60/54)
Security Clean (84/75), national security (82/74), social order (79/71), family
security (78/71), reciprocation of favors (73/68), healthy (55/49),
sense of belonging (54/48)
Self-Direction Creativity (92/85), curious (89/80), freedom (81/74), choosing own
goals (79/71), independent (76/68), private life (-)
Stimulation Daring (93/85), a varied life (93/85), an exciting life (87/79)
Tradition Devout (93/85), accepting portion in life (87/78), humble (79/72),
moderate (74/68), respect for tradition (74/66)
Universalism Protecting the environment (90/84), a world of beauty (90/82), unity with
nature (87/80), broad-minded (83/75), social justice (75/69),
wisdom (75/69), equality (74/68), a world at peace (73/66),
inner harmony (47/44)
NOTE: Numbers in parentheses are correct empirical locations of value items in 97 samples (Schwartz, 1994, p. 33)
and in 88 samples (Schwartz &Sagiv, 1995, pp. 99-101), respectively. (-) indicates value items not included in a pre-
vious 56-item Schwartz Value Survey. Items in italics have more than 75% of correct locations.
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
some values emerge consistently in their theoretical value type (for example, social power,
pleasure, helpful) whereas others seem to have weak links with them (for example, self-
respect, meaning in life; see Schwartz, 1992, 1994). Athreshold of 75%of correct locations
across the samples is proposed in order to select reliable and valid items for the measurement
of the 10 value types (Schwartz, 1994; Schwartz & Sagiv, 1995) (see Table 1).
This statistical procedure appears to be really efficient to describe the structural part of the
model. But the information provided by Schwartz in his works is usually limited to the fit of
the whole structure to the matrix of original dissimilarities and to the number of correct loca-
tions in the structures, whereas little or no information is supplied concerning the relation-
ships between the items and their underlying dimension. To get this last-mentioned informa-
tion, Schwartz (1992, 1994; Schwartz & Sagiv, 1995) compared the solutions obtained
across countries to conclude whether the items are or are not comparable, valid, and reliable
measures of the various value types. An exception to this procedure is to be found in Schmitt,
Schwartz, Steyer, & Schmitt (1995). Here, the measurement part of the value model is
assessed and confirmed using multitrait-multimethod analyses within a particular culture.
In this article, a complementary test of levels of equivalence across countries in the mea-
surement part of the model is assessed. This analysis is developed on the basis of structural
equation models with latent variables (e.g., Bollen, 1989b). This statistical technique
enables the test of the hypothesis of unidimensionality of the 10 value types across nations
and the relations between the value types and the various items defined to measure them. It is
also particularly suited for testing measurement equivalence in cross-cultural research as
already demonstrated by different applications and recent methodological developments
(Caprara, Barbaranelli, Bermdez, Maslach, &Ruch, 2000; Cheung &Rensvold, 2000; Lit-
tle, 1997; Steenkamp & Baumgartner, 1998).
Multigroup structural equation models are carried out here for the different types of val-
ues separately. This means that the structural part of the model (the relationships among the
latent variables) is not tested.
1
Inferences are made on the equivalence of measurement (the
relationships among the observed and latent variables) for each value type across cultures.
These analyses, in addition to the fact that they test assumptions of the universal theory of the
content of values, are particularly relevant for evaluating Schwartzs suggestions concerning
the number of items per value type. The validity and reliability of the value items as indica-
tors of the 10 value types is of prime concern to those who wish to use the value types cross-
culturally or within a given culture, as did Feather (1997), Sagiv and Schwartz (1995), Spini
and Doise (1998), and Devos, Spini, and Schwartz (in press).
LEVELS OF EQUIVALENCE
In summary, two types of information are provided by the analyses performed by
Schwartz and colleagues: (a) the degree to which the measurements of the value types, fol-
lowing Schwartzs (1992, 1994; Schwartz &Sagiv, 1995) indications, are similar across the
21 samples; and (b) the number of items that should be discarded from the Schwartz Value
Survey in order to use the value types measured in cross-cultural research. These indications
are indeed useful, but insufficient to claim for equivalence of measurement in the Schwartz
Value Survey or any other instrument. The main reason for this critical claimis that there are
different types of equivalence to be distinguished and tested across different groups.
Van de Vijver and Leung (1997b) distinguished three main types of equivalence: struc-
tural, measurement, and metric. Structural equivalence refers to the similarity of psycho-
metric properties of data sets from different cultures (p. 261). This level of equivalence is
6 JOURNAL OF CROSS-CULTURAL PSYCHOLOGY
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
primarily concerned with the similarity of correlations among the observed measures and by
the comparability of the factor structure underlying the instrument being compared across
cultures. Measurement unit equivalence is the case when the unit of measurement is identical
across cultural groups, but the scales do not have a common origin. This level of equivalence
enables the comparison of difference scores both within and across groups, whereas the
scores themselves can only be compared within cultures. If the additional equivalence of the
origin of scores can be demonstrated, one can assume scalar equivalence or full score compa-
rability. In this case, scores can be compared meaningfully both within and across cultural
groups. This level of equivalence is not easily obtained with psychological measurements in
cross-cultural research, whereas other types of measurement such as height, age, or weight
reach this last level of equivalence.
These different levels of equivalence have been defined in more detail by Steenkamp and
Baumgartner (1998). These authors proposed a more refined differentiation of the levels of
equivalence and also a procedure for assessing measurement invariance in cross-national
research using successive tests of equivalence using structural equation models.
To understand howstructural equation models can be used to test levels of equivalence of
each value type across different cultural groups, three equations are presented hereafter.
These equations express the basic structural and measurement relationships of the uni-
dimensional models across different sample groups (Bollen, 1989b, Steenkamp &
Baumgartner, 1998):
x
g
=
g
+
g

g
+
g
(1)

g
=
g
+
g

g
(2)

g
=
g

g
+
g
(3)
Assuming p items and one latent factor in each country, Equation 1 defines the relation-
ship between the observed variables x (p 1 vector) in the different groups or countries g (g =
1, . . . ., G) and the latent variable (11vector). represents the vector (p1) of coefficients
indicating the magnitude of the expected change in the observed variable for a one-unit
change in the latent variable. These coefficients are regression coefficients (factor loadings)
for the effects of the latent variable on the observed variable. The vector (p 1) represents
the items intercepts, and is the vector (p 1) of errors of measurement for x. Equation 2
defines the mean structure, with , the vector (p 1) of the itemmeans, and corresponding
to the vector (here 1 1) of latent means. Residuals are assumed to have means of 0. Equation
3 is the covariance structure with , the variance-covariance matrix of x, where is the vari-
ance-covariance matrix of the latent variable and the variance-covariance matrix of
(usually constrained to be a diagonal matrix). Taking into account these components of the
measurement model, Steenkamp and Baumgartner (1998) precise a procedure with succes-
sive steps corresponding to different levels of equivalence across groups or countries:
0) A first preliminary analysis tests the assumption that the different samples have different
covariances () and means () vectors. If this is not the case, the samples can be pooled and a test
of equivalence has no purpose. Basically, all cultures would stem from the same population.
Usually, equivalence cannot be assumed at this general level. Moreover, one can obtain an initial
idea of the differentiated impact of the inequivalence of means or covariances testing their
equality across the countries separately. Thus, three models are tested here, one pooling the
Spini / MEASUREMENT EQUIVALENCE OF VALUES 7
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
effects of means and covariance variances (), one testing the invariance of covariances (),
and one testing the invariance of means () across the countries.
1) Configural invariance: The goal here is to demonstrate that the items constituting the mea-
surement instrument exhibit the same configuration of salient (different from zero) and not-
salient (fixed at zero) factor loadings () across the countries. Unidimensional models of values
are evaluated using Schwartzs (1994) indications concerning the empirical location of each
value. These first models are considered here as the basic models (H
base
) for the test of
inequivalence. This type of model estimates factor variance, error variances, and factor load-
ings, with the exception of one factor loading (linking the same value itemto its latent variable in
all the samples) set at unity in order to anchor the scale of the latent variable.
2) Metric invariance: At this level, structural equation models can be used to test the hypothesis
(H) that metrics or scale intervals are equal across countries. This requires that configural
equivalence be accepted. To test the hypothesis of metric equivalence, constraints on the factor
loadings are defined in such a way as to constrain these parameters to be equal across the sam-
ples (
1
=
2
. . . =
G
). If metric invariance is confirmed, this results in the fact that difference
scores on an itemcan be meaningfully compared across samples and that these difference scores
are indicative of similar cross-national differences in the latent variable. Metric invariance is a
very important step in the evaluation of equivalence, as this level of invariance is a prerequisite
for other levels to be tested.
3) Scalar invariance: In many studies, it is of primary importance to compare means across
countries. To test such comparisons meaningfully, one must establish scalar invariance that
results in the possibility of comparing the observed means of the underlying construct across
groups. This test requires providing the analyses with the observed means in addition to the
covariances and to impose additional constrains of equality on the intercepts (
1
=
2
. . . =
G
).
This model (H

) is nested in the H

model and can thus be directly compared to it.

4) Factor variance invariance: The fourth model tests the equivalence of the factor variance
across the samples (H

). Here, the factor variance is fixed to be equal across the samples (

1
=

2
. . .
G
), in addition to the parameters already fixed in the H

model. The H

is nested in the H

model.
5) Error variance invariance: The last model (H

) adds to the H

model equal con-

straints across the samples on the error terms (
1
=
2
. . .
G
). These additional constraints make
it possible to test the hypothesis of an equal amount of error measurement across the samples. If
items are metrically invariant and if the error variances and factor variances are cross-nationally
invariant, the items are equally reliable across the samples.
At this point, we have defined two possible sequences of nested models. One is concerned
with covariationsH
base
> H

> H

> H

whereas the other focuses on meansH

base
>
H

> H

. If two nested models fit the data equally well, the most constrained model (hence
parsimonious) is accepted. If not, the hypothesis of equivalence is rejected and the least con-
strained model is retained. Subsequent model comparisons need not be evaluated (see Van de
Vijver & Leung, 1997a).
It is also important to note that other sequences could be defined on the basis of the
researchers interest. In particular, the last two tests, concerning variance and error
invariance, can be performed in the inverse order (H

) or separately (H

, H

), as they do
not build on each other as the others do (Steenkamp &Baumgartner, 1998). In this article the
sequence H
base
> H

> H

> H

is proposed, as it allows testing the reliability of the value

types measurements across the samples. Considering that the models applied in this study
have only one latent dimension, the invariance of factor covariance across samples, an addi-
tional test of equivalence mentioned by Steenkamp and Baumgartner (1998) is not of interest
here. It is also important to note that partial equivalence could also have been tested to assess,
for example, if the different levels of equivalence could be reached by the value types in a
8 JOURNAL OF CROSS-CULTURAL PSYCHOLOGY
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
smaller number of countries. But as Schwartzs theory of values is intended to be universal,
only full invariance is considered.
SELECTION OF ITEMS
The second type of information examined concerns the number of items composing the
value types. Therefore, different models, including a different number of items, are tested in
order to check the indications provided by Schwartz (1992, 1994). Thus, a first model is
composed of all the items of the Schwartz Values Survey linked to a given value type. Then,
items are progressively excluded from the model one by one, starting with the item that has
the smallest number of correct locations (see Table 1). This procedure is followed for value
types that have at least four items in their complete formto test the fit of the model. In princi-
ple, four items are considered a minimumfor testing a single dimension with structural equa-
tion models (see Mulaik & Millsap, 2000). The maximum number of items per value type
will be included in the final models in order to cover all the aspects of the value types and to
favor scales with a higher reliability, and three-itemmodels are only evaluated when models
including a greater number of items will be rejected. These models comprising a different
number of items are not nested. This means that the comparison of models is based on abso-
lute fit indices, and no direct comparison is carried out between the models, including a dif-
ferent number of items.
METHOD
SAMPLE
In total, 3,859 questionnaires were collected from student samples in 21 different coun-
tries (see Table 2). To ensure a certain degree of variability within each national sample and
comparability across samples, we requested the participation of 100 university students in
psychology and 100 in law. When this was not possible, samples were completed with stu-
dents from other major disciplines (languages, environmental studies, and natural sciences
in Finland, political sciences and English in India, economics in Israel, natural sciences and
management in the United Kingdom, and various branches in Senegal). Of course, samples
are not representative of the national populations. The goal here is to have comparable sam-
ples across a large variety of countries.
From this first sample, 3,787 participants were selected. These participants were those
who had at most two missing answers on the 57 value items. Missing values were then
replaced by the mean value of the country sample from which the respondent came and
rounded to unity to represent a possible answer and respect the distribution of the variables.
Five hundred forty-six answers were replaced in this way, which corresponds to 0.25%of the
total number of answers in the remaining sample. This procedure enabled us to retain
98.13%of the original sample, whereas a complete listwise deletion of missing cases would
have reduced the sample to 86.63% of the original.
MATERIAL AND PROCEDURE
The 57-itemSchwartz Value Survey was used as the first part of a questionnaire made up
of three parts: (a) Schwartz Value Survey, (b) organizing principles of position taken toward
Spini / MEASUREMENT EQUIVALENCE OF VALUES 9
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
human rights, and (c) factual variables (age, gender, socioeconomic status, religious prac-
tice, social activism, etc.) (see Spini, 1997; Spini &Doise, 1998). Most questionnaires were
completed between the end of 1995 and the beginning of 1997, but the last sample was
included in the study in January 1998.
Translation of questionnaires from English to French, Hebrew, Italian, Japanese, Portu-
guese, and Spanish was done partly by Schwartz and colleagues (see Schwartz, 1992) and
partly by ourselves.
2
For the other languages, the persons in charge of data collection in the
different countries were urged to make a double translation of the material (see Brislin,
Lonner, & Thorndike, 1973).
RESULTS
The analyses presented here are multigroup confirmatory factor analyses (see Bollen,
1989b, pp. 355-369) of unidimensional structural equation models. Lisrel 8.30 software
(Jreskog &Srbom, 1999) and maximumlikelihood estimation procedures are used on the
basis of variance-covariance and mean matrices.
3
As most variables were not multivariate
normal, one asymptotic covariance matrix was calculated on the basis of the total sample (K.
Jreskog, personal communication, April 11, 2001). This matrix (which needs a large num-
ber of observations) allows for the calculation of the Satorra-Bentler scaled chi-square value
10 JOURNAL OF CROSS-CULTURAL PSYCHOLOGY
TABLE 2
Number and Type of Students, Median Age, Percentage of Women and
Questionnaire Language
Type of Studies
Country n Psych. Law Other N.S. Age % Women Language
Argentina 221 92 129 22 73.9 Spanish
Bulgaria 209 149 60 22 72.9 Bulgarian
Canada 189 97 92 22 75.1 French
Chile 126 33 90 3 20 53.5 Spanish
Costa Rica 100 56 42 2 22 71.7 Spanish
Cte dIvoire 180 89 82 9 26 22.6 French
Estonia 210 59 88 63 21 70.5 Estonian
Finland 253 30 145 78 23 67.3 Finnish
France 181 93 88 22 63.0 French
India 200 97 35 68 23 38.0 Hindi
Israel 292 193 99 23 75.3 Hebrew
Italy 127 92 35 25 57.4 Italian
Japan 152 152 20 28.5 Japanese
Mexico 216 106 110 22 63.1 Spanish
Philippines 202 104 98 20 64.6 English
Portugal 102 30 70 2 22 76.5 Portuguese
Senegal 143 66 77 25 25.3 French
Switzerland 181 87 94 22 71.4 French
Uganda 200 100 100 23 41.1 English
United Kingdom 116 67 49 23 43.5 English
Yugoslavia 187 103 83 1 22 73.4 Serbo-Croat
Total sample 3,787 1,762 1,673 335 17 23 60.0
NOTE: Psych. = psychology; N.S. = not specified.
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
(Satorra & Bentler, 1994), a corrected chi-square for nonnormal data (see West, Finch, &
Curran, 1995).
For each model, the item that had the greatest number of correct locations in its a priori
value region across the samples was used to anchor the dimension fixing its factor loading
to 1 (see Table 1). Then, two types of overall fit indices were used: overall model fit measures
and comparative fit indices (see Bollen, 1989b).
Hu and Bentler (1999) recommended the association of two indices: the standardized root
mean squared residual (SRMR) (Bentler, 1995), which evaluates misspecifications in factor
covariances or latent structure of the model, and one among other types of fit indices such as
the root mean square error of approximation (RMSEA) (Steiger & Lind, 1980), the Tucker-
Lewis Index or Nonnormed Fit Index (Tucker-Lewis, 1973), the Incremental Fit Index (IFI)
(Bollen, 1989a), the Relative Noncentrality Index (RNI) (McDonald &Marsh, 1990) equiv-
alent to the Comparative Fit Index (CFI) (Bentler, 1990) or the Centrality Index (MCI)
(McDonald, 1989) in order to evaluate misspecifications in the measurement model.
Among this second list of fit indices two are used here, the IFI and the RMSEA. This
choice is dictated by redundancy in the indications given by the other fit indices. Values equal
or inferior to 0.08 for the SRMR are considered to indicate a good fit to the data (Hu &
Bentler, 1999). Values equal or superior to 0.95 for the IFI and below 0.06 for the RMSEA,
with a confidence interval limit below 0.08, are used as thresholds for not rejecting a model
(Browne & Cudeck, 1993; Hu & Bentler, 1999). In summary, models with both SRMR and
RMSEA or both SRMR and IFI values indicating a good fit will not be rejected.
In addition to these overall fit indices, comparative fit indices are also used when the dif-
ference between nested models is evaluated (see Bollen, 1989b). Two ways of directly com-
paring models are used: the difference in chi-squares between the compared models (Bollen,
1989b) and the difference in RMSEA (Browne & Dutoit, 1992) calculated using FITMOD
(Browne, 1992), a software providing point and interval estimates for RMSEA differences.
One of these two comparative indices should indicate that the nested models are equivalent in
order to accept the most constrained model. The same thresholds defined for the RMSEAare
used for the difference in RMSEA.
4
TESTING THE LEVELS OF EQUIVALENCE OF THE VALUE TYPES
Covariance and mean vectors. The evaluation of the models testing the invariance of
covariances () and means () is reported in Table 3. The value types with the maximum
number of items were used. As the goal of this analysis is to check the assumption that there
is variance at the mean and covariance levels across the samples, only a selected number of fit
indices are presented: the Satorra-Bentler
2
, the SRMR, RMSEA, and IFI. Results were
unambiguous; there is indeed mean and covariance variance across the sample for the 10
value types. These results justify a multisample analysis. Moreover, the separate analyses for
mean and covariance variance indicate that item means rather than item covariances are
major factors of variance across the samples.
Configural invariance. Unidimensional factor models across the 21 samples allowed us
to evaluate the configural invariance hypothesis (H
base
) for each value type in function of the
selected items (see Table 4). This hypothesis was not evaluated for Hedonism and Stimula-
tion, as these value types comprised only three items.
Spini / MEASUREMENT EQUIVALENCE OF VALUES 11
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
The three Achievement basic models were all acceptable, considering the SRMR and
RMSEA fit indices. Values of the IFI also indicated globally a very good fit, even if the six-
and five-itemmodels had values just belowthe 0.95 threshold. Having said this, it was diffi-
cult to choose one model over the others, and the three models were thus accepted.
The six models tested for Benevolence values merit two preliminary considerations. First,
the model with nine items could be rejected on the basis of an SRMRfit index superior to the
0.80 threshold. Second, maybe at the exception of the four-item model, all the models had
relatively low IFI goodness-of-fit values. Thus, it was on the basis of the association of
acceptable SRMRand RMSEAvalues that the eight-itemmodel could be taken into consid-
eration as the most inclusive model for more constrained tests of equivalence.
Only one model was tested for Conformity values, but this model fitted the data extremely
well and could therefore be accepted. Concerning Power values, the five-item model could
be rejected on the basis of both IFI and RMSEAindices. The hypothesis of configural equiv-
alence was accepted in the case of the four-itemmodel for which both the SRMRand the IFI,
contrary to the RMSEA, indicated a good fit.
12 JOURNAL OF CROSS-CULTURAL PSYCHOLOGY
TABLE 3
Absolute Fit Indices for Invariance of Means and Covariations
Absolute Fit Index
Value Type Item Satorra-Bentler
2
df SRMR RMSEA IFI
Achievement 6 3,635.99*** 540 0.20 0.18 0.05
1,007.13*** 420 0.19 0.088 0.67
1,213.36*** 120 0.00 0.23 0.62
Benevolence 9 4,217.44*** 1,080 0.13 0.13 0.39
1,546.99*** 900 0.11 0.063 0.78
1,491.44*** 180 0.00 0.27 0.74
Conformity 4 1,742.36*** 280 0.16 0.17 0.11
409.75*** 200 0.15 0.076 0.81
1,135.17*** 80 0.00 0.27 0.39
Hedonism 3 2,420.64*** 180 0.21 0.26 0.38
491.47*** 120 0.20 0.13 0.59
967.88*** 60 0.00 0.29 0.30
Power 5 2,409.43*** 400 0.11 0.17 0.36
728.15*** 300 0.13 0.089 0.83
1,217.60*** 100 0.00 0.25 0.60
Security 7 3,758.66*** 700 0.14 0.16 0.08
963.48*** 560 0.17 0.063 0.71
1566.02*** 140 0.00 0.24 0.46
Self-Direction 6 3,292.49*** 540 0.16 0.17 0.44
1,165.45*** 420 0.16 0.099 0.44
1,131.79*** 120 0.00 0.22 0.45
Stimulation 3 1,724.54*** 180 0.10 0.22 0.31
487.99*** 120 0.17 0.14 0.77
654.76*** 60 0.00 0.24 0.65
Tradition 5 4,290.59*** 400 0.14 0.23 1.37
632.25*** 300 0.15 0.079 0.63
1,513.41*** 100 0.00 0.28 0.29
Universalism 9 4,217.44*** 1,080 0.13 0.13 0.39
1,546.99*** 900 0.11 0.063 0.78
1,491.44*** 180 0.00 0.20 0.74
***p < .001.
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
Four models were evaluated for Security values. The hypothesis of configural equiva-
lence could be accepted for all the models tested on the basis of SRMRand RMSEAindices.
The IFI also indicated a good fit of the four-item model, whereas the values for the other
models were relatively lower. Thus, all four models were considered for the evaluation of
metric equivalence.
The complete Self-Direction value type contains six items, thus three models were evalu-
ated. However, the five-itemand four-itemmodels included a Heywood case (negative error
Spini / MEASUREMENT EQUIVALENCE OF VALUES 13
TABLE 4
Absolute Fit Indices for the Basic Models (H
base
)
Value Type
Overall Fit Index
Hypothesis Satorra-Bentler
2
df SRMR RMSEA RMSEA; 90% CI IFI
Achievement
H
base
6 207.49 189 0.032 0.023 0.000; 0.041 0.94
H
base
5 151.63** 105 0.030 0.050 0.031; 0.061 0.94
H
base
4 41.75 42 0.025 0.000 0.000; 0.050 0.98
Benevolence
H
base
9 730.92*** 567 0.082 0.040 0.031; 0.048 0.86
H
base
8 505.99** 420 0.059 0.039 0.021; 0.044 0.88
H
base
7 390.41*** 294 0.045 0.043 0.031; 0.054 0.87
H
base
6 261.53*** 189 0.041 0.046 0.032; 0.059 0.87
H
base
5 174.87*** 105 0.026 0.061 0.045; 0.077 0.87
H
base
4 54.42 42 0.020 0.041 0.000; 0.069 0.94
Conformity
H
base
4 47.91 42 0.036 0.028 0.000; 0.060 0.98
Power
H
base
5 237.27*** 105 0.051 0.084 0.070; 0.098 0.93
H
base
4 80.50*** 42 0.030 0.071 0.047; 0.095 0.97
Security
H
base
7 372.37** 294 0.071 0.039 0.025; 0.050 0.90
H
base
6 233.29* 189 0.078 0.036 0.017; 0.051 0.91
H
base
5 134.15* 105 0.065 0.039 0.014; 0.058 0.93
H
base
4 58.79* 42 0.026 0.047 0.008; 0.074 0.96
Self-Direction
H
base
6 268.99*** 189 0.043 0.049 0.035; 0.061 0.86
H
base
5 115.11 100 0.028 0.029 0.000; 0.013 0.93
H
base
4 29.36 40 0.024 0.000 0.000; 0.024 0.98
Tradition
H
base
5 107.00 100 0.067 0.020 0.000; 0.045 0.95
H
base
4 25.55 42 0.003 0.000 0.000; 0.000 1.00
Universalism
H
base
9 934.57*** 567 0.082 0.060 0.053; 0.067 0.82
H
base
8 636.96*** 420 0.073 0.054 0.045; 0.062 0.83
H
base
7 504.63*** 294 0.080 0.063 0.054; 0.072 0.83
H
base
6 270.37*** 189 0.081 0.049 0.035; 0.062 0.88
H
base
5a 137.96* 105 0.056 0.042 0.019; 0.060 0.89
H
base
5b 132.60* 105 0.056 0.038 0.011; 0.057 0.90
H
base
4 58.10* 42 0.058 0.046 0.000; 0.073 0.94
NOTE: RMSEA = root mean square error of approximation; CI = confidence interval; IFI = Incremental Fit
Index. =computed on 20 samples, without French sample. =computed on 20 samples, without Ugandan sample.

2
reports differences in
2
for nested models. Model 5a includes the item social justice. Model 5b includes the
item wisdom.
*p < .05. **p < .01. ***p < .001.
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
variance for the itemcreativity) in the French sample. It is always difficult to understand why
improper solutions appear, as they may be the consequence of different factors such as sam-
pling fluctuations, of outliers, of a fundamental fault in the specification of the model, or sim-
ply due to sampling (Bollen, 1989b). The proposed solution was to evaluate the overall fit of
the models excluding the French sample in which the Heywood case appeared. The IFI and
RMSEA indicated that the four- and five-item models had a relatively better fit than the six-
item model. But taking into account the SRMR and RMSEA, the three models were
accepted.
The next value type concerned Tradition values. AHeywood case (negative error variance
on the itemhumble) appeared in the Ugandan sample for the five-itemmodel. This problem
was again resolved by discarding this sample fromthe analysis. The results for the four- and
five-item models indicated an excellent fit of both models.
Finally, for Universalism values, seven models were tested. Of these, two models were
tested with five items each. This is due to the fact that the fifth itemsocial justiceand the
sixth itemwisdomwere equally correctly situated in the Universalism value type in
Schwartzs analyses (see Table 1). Concerning the fit of the models for Universalism, the pic-
ture is somewhat fuzzy. First, on the basis of the SRMR, the nine-, seven-, and six-itemmod-
els had values equal to or just above the indicated threshold of 0.80. Taken together, the
RMSEAand IFI also led us to reject the nine- and seven-itemmodels. We had then the choice
between the eight-item, the two five-item, and the four-item models. Considering the fact
that the seven- and six-item models did not reach acceptable standards, it was decided to
reject the eight-itemmodel, and thus to consider that the two five-itemmodels contained the
maximum number of items that were defined by a single dimension of values.
Metric invariance. To test the metric invariance of the value types across the country sam-
ples, a constraint was added to the models used for the configural invariance. It concerned the
factor loadings, which were fixed at a level equal across the samples (H

). As these models
were nested in the previous models (H
base
), it was possible, in addition to the absolute fit indi-
ces, to report comparative indices of fit (
2
and RMSEA) in order to assess whether the
constraints on the factor loadings significantly deteriorated the specified models or not (see
Table 5).
For Achievement values three models were considered, with six, five, and four items,
respectively. Considering the RMSEA and IFI overall fit indices, the five-item was associ-
ated with unacceptable values. This led us to also reject the six-item, which contained the
five-item model, and to accept the four-item model as metrically invariant. This conclusion
also considered the RMSEAindex, which indicated that this model was not different from
the basic model.
Five models were tested for Benevolence values comprising eight to four items. Con-
sidering the overall fit indices, and in particular the SRMRand RMSEA, only the eight-item
model was rejected for metric equivalence on the basis of SRMR and IFI.
Both overall and comparative fit indices for Conformity values indicated that the four-
item model was metrically equivalent across the national groups. As for the configural
hypothesis, this value dimension appeared to have excellent measurement properties for
cross-cultural comparisons, here of difference scores.
Hedonismwas tested for the hypothesis of metric equivalence using the three-itemform.
This implied that no comparative test was available, as the basic model has no degree of free-
dom. Surprisingly, taking into account that a small number of items usually favored the
acceptance of the hypotheses of equivalence across our results, here both RMSEA and IFI
14 JOURNAL OF CROSS-CULTURAL PSYCHOLOGY
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
indices were just above the defined thresholds. Therefore, the hypothesis of metric equiva-
lence was rejected for Hedonism.
The Power four-item H

model was rejected on the basis of both RMSEA and IFI. The
Power values included the items social power, authority, and wealth as best indicators (see
Table 1), which together constituted a fairly consistent set of power values. This was the
Spini / MEASUREMENT EQUIVALENCE OF VALUES 15
TABLE 5
Absolute and Comparative Fit Indices for the Equal Factor Loadings Models (H

)
and the Change in Fit From the Basic Models (H
base
)
Absolute Fit Index Comparative Fit Index
Value Type Satorra- RMSEA; RMSEA;
Hypothesis Bentler
2
df SRMR RMSEA 90% CI IFI
2
RMSEA 90% CI
Achievement
H

6 408.21*** 289 0.064 0.048 0.037; 0.058 0.87 200.72*** 0.016 0.013; 0.020
H

5 329.01*** 185 0.063 0.066 0.054; 0.077 0.87 177.38*** 0.018 0.014; 0.021
H

4 139.54** 102 0.076 0.045 0.024; 0.063 0.93 97.79** 0.013 0.008; 0.017
Benevolence
H

8 755.43*** 560 0.081 0.044 0.036; 0.052 0.82 294.44*** 0.017 0.014; 0.020
H

7 562.56*** 414 0.062 0.045 0.035; 0.054 0.82 172.15** 0.011 0.007; 0.014
H

6 407.83*** 289 0.050 0.048 0.037; 0.058 0.81 146.30** 0.011 0.007; 0.015
H

5 273.41*** 185 0.038 0.052 0.038; 0.064 0.82 98.54 0.008 0.000; 0.013
H

4 122.31 102 0.031 0.033 0.000; 0.053 0.88 67.89 0.006 0.000; 0.012
Conformity
H

4 102.65 102 0.058 0.006 0.000; 0.040 0.97 54.74 0.000 0.000; 0.008
Hedonism
H

3 68.42** 40 0.055 0.063 0.036; 0.088 0.94 NA NA NA

Power
H

4 195.72*** 102 0.047 0.072 0.056; 0.087 0.94 115.22*** 0.016 0.011; 0.020
H

3 48.13 40 0.045 0.034 0.000; 0.064 0.98 NA NA NA

Security
H

7 516.31*** 414 0.083 0.037 0.026; 0.047 0.87 143.94 0.000 0.000; 0.006
H

6 355.63** 289 0.091 0.036 0.021; 0.048 0.88 122.34 0.008 0.000; 0.012
H

5 237.04** 185 0.100 0.040 0.022; 0.054 0.90 102.89* 0.009 0.002; 0.013
H

4 135.23* 102 0.120 0.043 0.020; 0.061 0.92 76.44 0.009 0.000; 0.014
H

3 52.10 40 0.071 0.041 0.000; 0.070 0.95 NA NA NA

Self-Direction
H

6 400.17*** 289 0.062 0.046 0.035; 0.057 0.82 131.18 0.009 0.004; 0.013
H

5 228.65* 185 0.052 0.036 0.017; 0.051 0.86 NA NA NA

H

4 98.75 102 0.050 0.000 0.000; 0.037 0.92 NA NA NA

Stimulation
H

3 63.05 40 0.064 0.057 0.027; 0.082 0.97 NA NA NA

Tradition
H

5 244.81** 185 0.130 0.042 0.026; 0.056 0.87 NA NA NA

H

4 79.99 102 0.075 0.000 0.000; 0.046 0.97 42.34 0.000 0.000; 0.000
Universalism
H

5a 230.52* 185 0.056 0.037 0.018; 0.052 0.85 92.56 0.006 0.000; 0.012
H

5b 231.17* 185 0.056 0.037 0.019; 0.052 0.84 98.57 0.008 0.000; 0.013
H

4 142.82** 102 0.067 0.047 0.027; 0.065 0.87 84.72* 0.010 0.004; 0.015
NOTE: SRMR = standardized root mean squared residual; RMSEA = root mean square error of approximation;
CI = confidence interval; IFI = Incremental Fit Index; NA = not available.
2
reports differences in
2
for nested
models. Model 5a includes the itemsocial justice; Model 5b includes the itemwisdom. Degrees of freedomfor
2
are respectively 120, 100, 80, and 60 for 7, 6, 5, and 4 items.
*p < .05. **p < .01. ***p < .001.
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
reason why the three-item model was also considered and tested, even if the rule of four
items or more per factor was generally followed. The result was that this three-item model
had a relatively good fit as indicated by all the absolute fit indices and may, in consequence,
be considered as metrically equivalent across the samples.
The four Security models including fromseven to four items had to be rejected as they all
were associated with high SRMR values. Only the three-item model, including the items
clean, national security, and social order, reached an acceptable SRMR value, and the other
overall fit indices confirmed that this reduced model, including the items clean, national
security, and social order, may be metrically equivalent across national samples.
Three models for Self-Direction were evaluated for metric equivalence with six, five, and
four items. The overall fit indices indicated an acceptable overall fit of all three models, espe-
cially taking into account the SRMR and RMSEA. Moreover, even if the comparative fit
indices were not all available, those concerning the six-item model indicated that this con-
strained model was equivalent to the basic model.
For the three-itemStimulation model, SRMRand IFI indicated a good fit even though the
value of the RMSEAupper confidence limit was a little bit too high. The hypothesis of metric
equivalence was accepted for the three-item model.
For Tradition values, the results were clear. Both overall and comparative fit indices con-
firmed that the four-item model had an excellent fit, better than the five-item model (for
which the SRMRand IFI indicated a lack of fit). The four-itemmodel was accepted as metri-
cally equivalent.
Finally, the three models of Universalism showed comparable results with acceptable
SRMR and RMSEA values for the absolute fit indices and relatively low values for the IFI.
At the exception of the significant difference in chi-square for the four-item models testing
configural and metric equivalence, all other comparative evaluations indicated that the most
constrained models were equivalent to the basic models and should thus be accepted as indi-
cating metric equivalence for Universalism values measured with the two five-item models
across the samples.
Scalar invariance. The hypothesis tackled here concerns equivalence in means or scalar
invariance (H

). The models that were tested, as specified before, were nested in the H

models and thus were statistically compared to them. Here, as one might expect on the basis
of the results presented in Table 3, all the models tested showed a very bad fit and the results
are therefore not presented in detail. The Conformity four-item model (a good fitting model
in the previous analyses) can be taken as an example: Satorra-Bentler
2
(182) = 1236.69, p <
.001; SRMR = 0.058; RMSEA = 0.18 [0.17; 0.19]; IFI = 0.30;
2
(80) = 1134.04, p < .001;
RMSEA = 0.059 [0.056; 0.062]. In consequence, the hypothesis of equivalent means
across the samples had to be rejected for all value types.
Factor variance invariance. Another step in the analysis of levels of equivalence involves
the test of factor variance invariance across the samples (H

, see Table 6). These models are

nested in the H

models.
Values of the SRMR for the four-item model of Achievement indicated that this model
had to be rejected. This was also true for the three-item model (not reported here) that had
also been tested and that also showed a rather bad fit (SRMR = 0.14).
Four models were tested for Benevolence values. The result was somewhat ambiguous.
On one hand, all RMSEA values indicated a good fit, contrary to the IFI values that all indi-
cated a relatively bad fit. On the other hand, the SRMR values were really close to the
16 JOURNAL OF CROSS-CULTURAL PSYCHOLOGY
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
threshold value of 0.08, but only the seven- and four-itemmodels had acceptable SRMRval-
ues. Comparative fit indices showed that all models were equivalent to the H

models. As all
these models were somewhat equivalent, and as the seven-itemand four-itemmodels should
be accepted on the basis of the rules we are following, we suggest rejecting the seven-item
model that included the six- and five-itemrejected models and accepting the four-itemmodel
as having a unique and equal variance across the samples.
The Conformity four-item model was accepted as indicated by the overall and compara-
tive fit indices. Thus, the hypothesis of equal factor variance across the 21 countries was con-
firmed for this value type.
Both the IFI and RMSEAindicated a lack of overall fit of the three-itemPower model for
which the hypothesis of factor variance equivalence across samples was rejected. All the
models for the Security values, including fromseven to four items, were rejected on the basis
Spini / MEASUREMENT EQUIVALENCE OF VALUES 17
TABLE 6
Fit Indices for the Variance Equivalence Models (H

) and
the Change in Fit From the Equal Factor Loadings Models (H

)
Overall Fit Index Comparative Fit Index
Value Type Satorra- RMSEA; RMSEA;
Hypothesis Bentler
2
df SRMR RMSEA 90% CI IFI
2
RMSEA 90% CI
Achievement
H

4 183.86*** 122 0.12 0.053 0.037; 0.068 0.90 44.32** 0.018 0.011; 0.025
Benevolence
H

7 597.11*** 434 0.078 0.046 0.036; 0.055 0.81 34.55* 0.014 0.005; 0.021
H

6 439.49*** 309 0.084 0.049 0.038; 0.059 0.80 31.28 0.012 0.000; 0.020
H

5 303.04*** 205 0.088 0.052 0.039; 0.064 0.80 29.63 0.011 0.000; 0.019
H

4 156.98** 122 0.073 0.040 0.018; 0.057 0.83 34.67* 0.014 0.005; 0.022
Conformity
H

4 145.27** 122 0.057 0.033 0.000; 0.051 0.95 42.62** 0.017 0.010; 0.025
Power
H

3 99.85*** 60 0.049 0.061 0.039; 0.081 0.93 51.72*** 0.021 0.014; 0.027
Security
H

7 552.75*** 434 0.120 0.039 0.028; 0.049 0.86 36.44* 0.015 0.007; 0.022
H

6 392.93*** 309 0.140 0.039 0.026; 0.050 0.87 37.30* 0.015 0.007; 0.023
H

5 275.52*** 205 0.150 0.044 0.029; 0.057 0.88 46.87*** 0.019 0.012; 0.029
H

4 174.90** 122 0.140 0.049 0.032; 0.065 0.89 39.67** 0.016 0.009; 0.023
Self-Direction
H

6 427.59*** 309 0.066 0.046 0.035; 0.057 0.81 27.42 0.010 0.010; 0011
H

5 256.33** 205 0.059 0.037 0.020; 0.051 0.85 27.68 0.010 0.000; 0.018
H

4 122.72 122 0.051 0.006 0.000; 0.038 0.90 23.97 0.007 0.000; 0.016
Stimulation
H

3 100.67*** 60 0.120 0.061 0.040; 0.082 0.96 37.62** 0.015 0.007; 0.023
Tradition
H

4 116.75 122 0.120 0.000 0.000; 0.033 0.93 36.76* 0.015 0.007; 0.022
Universalism
H

5a 246.86* 205 0.068 0.034 0.014; 0.048 0.84 16.34 0.000 0.000; 0.011
H

5b 247.38* 205 0.067 0.034 0.014; 0.048 0.84 16.21 0.000 0.000; 0.011
H

4 159.41* 122 0.070 0.041 0.020; 0.058 0.87 16.59 0.000 0.000; 0.011
NOTE: SRMR = standardized root mean squared residual; RMSEA = root mean square error of approximation;
CI = confidence interval; IFI = Incremental Fit Index. Model 5a includes the itemsocial justice; Model 5b includes
the item wisdom. Degrees of freedom for
2
= 20.
*p < .05. **p < .01. ***p < .001.
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
of RMSEA and IFI. A complementary analysis with the three-item model gave the same
results with RMSEA = 0.060 [0.038; 0.081]; IFI = 0.90.
The SRMRand RMSEAand the two comparative fit indices confirmed that the hypothe-
sis of equal variance could be accepted for the four-itemSelf-Direction values. This is not the
case for the three-itemmodel for the Stimulation values, which had to be rejected on the basis
of both SRMR and RMSEA.
Concerning Tradition values, the hypothesis of equal variance across the samples was
rejected for the four-item model on the basis of SRMR and IFI. The three-item model (not
reported here) had also been tested but revealed a lack of fit, with almost equal SRMRand IFI
values.
Finally, the three models for Universalism had acceptable overall and comparative fit
indices. These indicated that the hypothesis of equal factor variance across the samples could
be accepted for both five-itemmodels, including the itemsocial justice or the itemwisdom.
Error variance invariance. The last step in the analysis of equivalence concerns the
hypothesis of invariance of the items error variances (H

). Results are not reported, as all

model evaluations indicated a poor fit. The fit indices for the four-item Conformity model
may again serve as an example of result: Satorra-Bentler
2
(202) = 429.66, p < .001,
SRMR = 0.15; RMSEA = 0.079 [0.069; 0.090]; IFI = 0.79,
2
(80) = 284.39, p < .001;
RMSEA = 0.026 [0.023; 0.029]. Thus, results clearly indicated a lack of support for the
hypothesis of equivalent error variance and the related hypothesis of measurement reliability
across the samples for all value types irrespective of the number of items included.
DISCUSSION
The confirmatory factor models confirmed that configural and metric equivalence, as
well as factor variance invariance, held for most value types. The only exception is Hedo-
nism, for which metric equivalence is rejected and configural equivalence impossible to
evaluate. Thus, this article generally confirms Schwartzs (1994; Schwartz & Sagiv, 1995)
conclusions on the legitimacy of using the Schwartz Value Survey as an instrument for cross-
cultural research. But the results also reveal with more precision the level of equivalences
across cultures that can be achieved with the diverse value types.
Results presented in Table 7 indicate the maximum number of items that can be consid-
ered equivalent for each value dimension by levels of equivalence following our results and
Schwartzs indications based on the 75%rule (see Table 1). The first conclusion emerging
from this synthesis is that almost all value types are equivalent at the configural and metric
levels, and four are equivalent at the level of factor variance. However, the indices for mea-
suring the 10 value types are not reliable, as indicated by the rejection of the error variance
models, and not scalar equivalent, which hampers comparison of means across the samples.
This result concerning factor variances is somewhat at odds with the Schwartz and Sagie
(2000) study which showed, using teacher and representative samples across 42 nations, that
the variability of all value types correlate with the level of democratization or the national
socioeconomic status. Following our results, one could conclude that for four value types,
Benevolence, Conformity, Self-Direction, and Universalism, there is no difference in factor
variance across the samples, which could be in discordance with the idea that there should be
major and explainable differences in factor variance across the samples included in this study
as they surely are not equivalent in democratization and socioeconomic status. However, it is
18 JOURNAL OF CROSS-CULTURAL PSYCHOLOGY
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
not possible to drawfirmconclusions as, on one hand, the student samples used in this study
may be too homogeneous and impede existing variance to appear, and, on the other hand,
Schwartz and Sagie (2000) have not directly tested the equality of variance across their sam-
ples. At least, the results indicate that variance in the value types should not be taken as
granted.
Comparing these results with Schwartzs concerning the number of items valid for the
measurement of the value types across cultures, the conclusion is that there is a good conver-
gence between the results derived from two methods, namely, structural equation models
and the 75%criterion used in the similarity structure analyses by Schwartz (1992). Neverthe-
less, some differences also become apparent. Concerning configural equivalence, there is a
strict convergence for only two value types, Conformity and Tradition. The analyses also
indicate that the number of items could be greater than the number previously indicated for
Achievement (six instead of four), Benevolence (eight instead of five), Power (four instead
of three), and Security (seven instead of five). The opposite conclusion can be drawn for Uni-
versalism (five instead of eight).
At the level of metric invariance four value types, Achievement, Conformity, Power, and
Stimulation, are equivalent, with the same number of items in our study at both metric and
factor variance levels and in Schwartzs studies. For two other value types, Benevolence and
Self-Direction, results indicate that more items can be used cross-culturally assuming metric
equivalence. For the other value types (Security, Tradition, and Universalism), a smaller
number of items should be used. Overall, there is a remarkable convergence of results
between Schwartzs and those presented here, which should encourage cross-cultural
researchers to use the Schwartz Value Survey.
CONCLUSION
Confirmatory factor analyses have been used in the past in cross-cultural psychology
(see, e.g., De Groot, Koot, &Verhulst, 1994; Windle, Iwawaki, &Lerner, 1988), but usually
with a much more limited number of samples than the analyses presented here. Structural
equation models have various advantages over other multidimensional techniques. The first
Spini / MEASUREMENT EQUIVALENCE OF VALUES 19
TABLE 7
Number of Items for Each Value Type by Level of Equivalence
Level of Equivalence
Value Type Schwartzs Indication Configural Metric Factor Variance
Achievement 4 6 4 X
Benevolence 5 8 7 4
Conformity 4 4 4 4
Hedonism 3 n.t. X X
Power 3 4 3 X
Security 5 7 3 X
Self-Direction 5 6 6 4
Stimulation 3 n.t. 3 X
Tradition 5 5 4 X
Universalism 8 5a, b 5a, b 5a, b
NOTE: X=rejected models; n.t. =not tested;. 5a =model including the itemsocial justice; 5b =model including the
item wisdom.
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
is that they provide a means of statistically testing the hypothesis of an equivalent structure of
dimensions across a large number of samples. The value of different parameters composing
the models can be fixed, with the result that a progressive test of equivalence across different
samples can be performed (see Steenkamp &Baumgartner, 1998). The second advantage of
this statistical technique is that it provides a means of evaluating the equivalence of the com-
ponents of the model and, in particular, of obtaining precise information on the links between
the underlying dimensions of values and the items composing them. In this regard, it should
be noted that the parameters and their standard errors are not reported in this study, but an
examination of these parameters indicates that they are almost all in the expected direction
and significant at least at the usual statistical p < .05 level.
However, this statistical technique does not solve all the problems of testing equivalence
across different cultures. Difficulties in using it arose during the process of getting the results
and interpreting them. For example, practical difficulties appeared with improper solutions,
which impeded the test of some models across all the samples.
One has also to acknowledge that most models tested in this study were rejected. For the
moment, the different sources explaining this lack of fit can only be briefly mentioned:
(a) The large number of samples and participants taken into account in this study certainly favors
the rejection of models. This is the main reason we decided to take into account either the asso-
ciation of SRMRand RMSEAor SRMRand IFI in the evaluation of fit. Otherwise, most mod-
els should have been rejected if the three indices had been evaluated together. Even if consistent
progress was made, future developments should define more precisely howto evaluate fit indi-
ces as a function of a given number of groups and participants.
(b) Some items may not be the best indicators of the latent variables. The tests performed here
relied exclusively on the content of the Schwartz Value Survey. For each value type, better indi-
cators or better ways of measuring them across cultures will no doubt be tested in the future
(see, e.g., Schwartz, Lehmann, Melech, Burgess, & Harris, 2001). In this regard, a good bal-
ance between reliability of indicators and construct validity should be carefully determined,
using, for example, multitrait-multimethod designs (see Schmitt et al., 1995). But relying
exclusively on structural equation modeling, in our experience, tends to privilege simple mod-
els with fewindicators having very similar meaning. For example, the value type Security, with
only three metrically equivalent items (clean, national security, and social order), appears to
include only items related to society security and to exclude the items concerned with security
in the community or in more restricted groups (especially family security and reciprocation of
favors). Another example lies in the fact that for Universalism, the two five-item models were
accepted at the configural and metric levels, but not the six-itemmodel. This may mean that the
items social justice and wisdom belong to two different dimensions, and that Universalism
should be conceived as supraordinate dimension covering different aspects. Thus, only
narrowly defined constructs may be good candidates for measurement equivalence across
cultures.
(c) The lack of fit is also surely caused by real differences in the value types structure and content.
In this regard, less restricted models seeking partial equivalence (freeing some estimates in spe-
cific groups that have the strongest impact on the reduction of residuals) may lead to other con-
clusions showing real differences across countries, which have then to be interpreted
(Steenkamp & Baumgartner, 1998).
It is hard at this point to say if the glass is half empty or half full. The answer probably
depends on the epistemological (or methodological) standpoint of the reader. If the aim of
research is to set forth a definitive theoretical model of the cultural differences on the basis of
universal and shared values, the glass is probably half empty, as the path leading to confirma-
tion (or nonrejection) of a whole structure of values with structural equation models across
many different cultures is full of obstacles. But if we take the view, as proposed by Mulaik
20 JOURNAL OF CROSS-CULTURAL PSYCHOLOGY
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
(1987) or Bollen (1989b), that research gradually approaches reality, then the present
research represents more than a half-full glass. It demonstrates that it is possible for separate
value types to be equivalent at different levels across a large number of samples rooted in cul-
tural diversity and provides hints for future developments.
NOTES
1. Attempts were made to test the entire structure of values (the structural part of the model) with structural equa-
tion models. Up to now, the models have not reached satisfactory levels of fit.
2. I would like to thank Giovanna Guerzoni, Monica Herrera, Isabell Kempf, Fatima Savory, and Tomaso Solari
for their help in this task.
3. The matrices of correlations, as well as the means and standard deviations, could not be reported here as they
would represent 21 matrices of 57 57 variables. On request to the author, a full account of all correlations, means,
and standard deviations is available.
4. One reviewer suggested here to use other comparative measures such as the Aikake Information Criterion
(Aikake, 1987), the Consistent version of the AIC (Bozdogan, 1987), or the Bayes Information Criterion (Raftery,
1993; G. Schwarz, 1978). These fit indices are not reported here as they invariably indicated that parsimonious mod-
els with fewer items were to be preferred over more inclusive measures.
REFERENCES
Aikake, H. (1987). Factor analysis and AIC. Psychometrika, 52, 317-332.
Bentler, P. M. (1990). Comparative fit indices in structural models. Psychological Bulletin, 107, 238-246.
Bentler, P. M. (1995). EQS structural equations program manual. Encino, CA: Multivariate Software.
Bollen, K. A. (1989a). A new Incremental Fit Index for general structural equation models. Sociological Methods
and Research, 17, 303-316.
Bollen, K. A. (1989b). Structural equations with latent variables. New York: John Wiley.
Bond, M. H. (1988). Finding universal dimensions of individual variation in multicultural studies of values: The
Rokeach and Chinese Value Surveys. Journal of Personality and Social Psychology, 55(6), 1009-1015.
Borg, I., & Lingoes, J. C. (1987). Multidimensional similarity structure analysis. New York: Springer Verlag.
Bozdogan, H. (1987). Model selection and Aikakes information criterion (AIC): The general theory and its analyti-
cal extensions. Psychometrika, 52, 345-370.
Braithwaite, V. A., &Law, H. G. (1985). Structure of human values: Testing the adequacy of the Rokeach Value Sur-
vey. Journal of Personality and Social Psychology, 49, 250-263.
Brislin, R. W., Lonner, W., &Thorndike, R. M. (1973). Cross-cultural research methods. NewYork: John Wiley.
Browne, M. W. (1992). FITMOD: Point and interval estimates of measures of fit of a model [computer software].
Browne, M. W., &Cudeck, R. (1993). Alternative ways of assessing model fit. In K. A. Bollen &J. S. Long (Eds.),
Testing structural equation models (pp. 136-162). Newbury Park, CA: Sage.
Browne, M. W., & Dutoit, S. H. C. (1992). Automated fitting of nonstandard models. Multivariate Behavioral
Research, 27, 269-300.
Caprara, G. V., Barbaranelli, C., Bermdez, J, Maslach, C., & Ruch, W. (2000). Multivariate methods for the com-
parison of factor structures in cross-cultural research: An illustration with the Big Five Questionnaire. Journal
of Cross-Cultural Psychology, 31(4), 437-464.
Cheung, G. W., & Rensvold, R. B. (2000). Assessing extreme and acquiescence response sets in cross-cultural
research using structural equation modeling. Journal of Cross-Cultural Psychology, 31(2), 187-212.
Chinese Culture Connection. (1987). Chinese values and the search for culture-free dimensions of culture. Journal
of Cross-Cultural Psychology, 18(2), 143-164.
De Groot, A., Koot, H. M., &Verhulst, F. C. (1994). Cross-cultural generalizability of the Child Behavior Checklist
cross-informant syndromes. Psychological Assessment, 6, 225-230.
Devos, T., Spini, D., & Schwartz, S. H. (in press). Structure and conflicts of human values: the role of religiosity,
political orientation, patriotism, and trust in institutions. British Journal of Social Psychology.
Feather, N. T. (1997). Values, national identification and favouritismtowards the in-group. British Journal of Social
Psychology, 33(4), 467-476.
Guttman, L. A. (1968). Ageneral nonmetric technique for finding the smallest coordinate space for a configuration
of points. Psychometrika, 33, 469-506.
Spini / MEASUREMENT EQUIVALENCE OF VALUES 21
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
Hofstede, G. (1980). Cultures consequences: International differences in work-related values. Beverly Hills, CA:
Sage.
Hofstede, G. (1982). Dimensions of national cultures. In R. Rath, H. S. Asthana, D. Sinha, & J. B. H. Sinha (Eds.),
Diversity and unity in cross-cultural psychology (pp. 173-187). Lisse, Netherlands: Swets & Zeitlinger.
Hofstede, G. (1983). Dimensions of national cultures in fifty countries and three regions. In J. B. Deregowski, S.
Dzivrawiec, &R. C. Annis (Eds.), Expiscations in cross-cultural psychology (pp. 335-355). Lisse, Netherlands:
Swets & Zeitlinger.
Hofstede, G., & Bond, M. H. (1984). Hofstedes culture dimensions: An independent validation using Rokeachs
Value Survey. Journal of Cross-Cultural Psychology, 15(4), 417-433.
Hu, L., &Bentler, P. M. (1999). Cutoff criteria for fit indices in covariance structure analysis: Conventional criteria
versus new alternatives. Structural Equation Modeling, 6(1), 1-55.
Jreskog, K. G. (1971). Simultaneous factor analysis in several populations. Psychometrika, 36, 409-426.
Jreskog, K. G., & Srbom, D. (1999). Lisrel 8.30 and Prelis 2.30. Chicago: Scientific Software International.
Kruskal, J. B., & Wish, M. (1990). Multidimensional scaling (15th ed.). Newbury Park, CA: Sage.
Little, T. D. (1997). Mean and covariance structures (MACS) analyses of cross-cultural data: Practical and theoreti-
cal issues. Multivariate Behavioural Research, 32(1), 53-76.
McDonald, R. P. (1989). An index of goodness-of-fit based on noncentrality. Journal of Classification, 6, 97-103.
McDonald, R. P., & Marsh, H. W. (1990). Choosing a multivariate model: Noncentrality and goodness of fit. Psy-
chological Bulletin, 107, 247-255.
Mulaik, S. A. (1987). Toward a conception of causality applicable to experimentation and causal modeling. Child
Development, 58, 18-32.
Mulaik, S. A., & Millsap, R. E. (2000). Doing the four-step right. Structural Equation Modeling, 7(1), 36-73.
Ng, S. H., Akhtar-Hosain, A. B. M., Ball, P., Bond, M. H., Hayashi, K., Lim, S. P., et al. (1982). Values in nine coun-
tries. In R. Rath, H. S. Asthana, & J. B. H. Sinha (Eds.), Diversity and unity in cross-cultural psychology (pp.
196-205). Lisse, Netherlands: Swets & Zeitlinger.
Raftery, A. E. (1993). Bayesian structural model selection in structural equation models. In K. A. Bollen & J. S.
Long (Eds.), Testing structural equation models (pp. 163-180). Newbury Park, CA: Sage.
Rokeach, M. (1967). Value survey. Sunnyvale, CA: Halgren Tests.
Rokeach, M. (1973). The nature of human values. New York: Free Press
Sagiv, L., &Schwartz, S. H. (1995). Value priorities and readiness for out-group social contact. Journal of Personal-
ity and Social Psychology, 69(3), 437-448.
Satorra, A., &Bentler, P. M. (1994). Corrections to test and statistic and standard errors in covariance structure anal-
ysis. In A. Von Eye &C. C. Clogg (Eds.), Analysis of latent variables in developmental research (pp. 399-419).
Thousand Oaks, CA: Sage.
Schmitt, M. J., Schwartz, S. H., Steyer, R., & Schmitt, T. (1995). Measurement models for the Schwartz Values
Inventory. European Journal of Psychological Assessment, 9(2), 107-121.
Schwarz, G. (1978). Estimating the dimension of a model. Annals of Statistics, 6, 461-464.
Schwartz, S. H. (1992). Universals in the content and structure of values: Theoretical advances and empirical tests in
20 countries. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 25, pp. 1-65). San Diego,
CA: Academic Press.
Schwartz, S. H. (1994). Are there universal aspects in the structure and contents of human values? Journal of Social
Issues, 4, 19-45.
Schwartz, S. H., &Bilsky, W. (1987). Toward a psychological structure of human values. Journal of Personality and
Social Psychology, 53, 550-562.
Schwartz, S. H., & Bilsky, W. (1990). Toward a theory of the universal content and structure of values: Extensions
and cross-cultural replications. Journal of Personality and Social Psychology, 53, 550-562.
Schwartz, S. H., Lehmann, A., Melech, G., Burgess, S., &Harris, M. (2001). Validation of a theory of basic human
values with a new instrument in new populations. Journal of Cross-Cultural Psychology, 32(5), 519-542.
Schwartz, S. H., & Sagie, G. (2000). Value consensus and importance: A cross-national study. Journal of Cross-
Cultural Psychology, 31(4), 465-497.
Schwartz, S. H., & Sagiv, L. (1995). Identifying culture-specifics in the content and structure of values. Journal of
Cross-Cultural Psychology, 26(1), 92-116.
Spini, D. (1997). Valeurs et reprsentations sociales des droits de lhomme: Une approche structurale [Values and
social representations of human rights: Astructural approach]. Unpublished doctoral dissertation, University of
Geneva, No 243, Switzerland.
Spini, D., & Doise, W. (1998). Organizing principles of involvement in human rights and their social anchoring in
value priorities. European Journal of Social Psychology, 28(4), 603-622.
Steenkamp, J.-B. E. M., &Baumgartner, H. (1998). Assessing measurement invariance in cross-national consumer
research. Journal of Consumer Research, 25, 78-90.
Steiger, J. H., &Lind, J. C. (1980, May). Statistically based tests for the number of common factors. Paper presented
at the annual meeting of the Psychometric Society, Iowa City, IA.
Tucker, L. R., &Lewis, C. (1973). Areliability coefficient for maximumlikelihood factor analysis. Psychometrika,
38, 1-10.
22 JOURNAL OF CROSS-CULTURAL PSYCHOLOGY
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from
Van de Vijver, F., &Leung, K. (1997a). Methods and data analysis for cross-cultural research. Thousand Oaks, CA:
Sage.
Van de Vijver, F., & Leung, K. (1997b). Methods and data analysis of comparative research. In J. W. Berry, Y. H.
Poortinga, &J. Pandey (Eds.), Handbook of Cross-Cultural Psychology (Vol. 1, pp. 257-300). Boston: Allyn &
Bacon.
West, S. G., Finch, J. F., &Curran, P. J. (1995). Structural equation models with nonnormal variables: Problems and
remedies. In R. H. Hoyle (Ed.), Structural equation modeling: Concepts, issues, and applications (pp. 56-75).
Thousand Oaks, CA: Sage.
Windle, M., Iwawaki, S., & Lerner, R. M. (1988). Cross-cultural comparability of temperament among Japanese
and American preschool children. International Journal of Psychology, 23, 547-567.
Dario Spini is assistant professor at the Institute for Life Course and Life Style Studies at the Universities of
Lausanne and Geneva. He also works at the Centre for Interdisciplinary Gerontology at the University of
Geneva, where he participates in a longitudinal study on the oldest old. He received his Ph.D. at the Univer-
sity of Geneva and was trained as a social psychologist. His research interests are the relationship between
values and social behaviour, social representations of human rights and social norms, and in the develop-
ment of a social psychological perspective in the orientation of life course research.
Spini / MEASUREMENT EQUIVALENCE OF VALUES 23
2003 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
at Bibliotheques de l'Universite Lumiere Lyon 2 on May 2, 2008 http://jcc.sagepub.com Downloaded from