FIND A SOLUTION AT Academic Writers Bay

Peter B Smith; Shaun Dugan
Journal of Cross-Cultural Psychology, Mar96, Vol. 27 Issue 2, p231
A Dimensional Analysis Across 43 Nations
The values of 8,841 managers and organization employees from 43 countries were
surveyed. The range of nations included paralleled many of those surveyed by Hofstede
(1980) but added also substantial samples from ex-communist nations. Questionnaire
items focused primarily on measures of universalism-particularism, achievementascription, and individualism-collectivism Multidimensional scaling of country means
revealed three interpretable dimensions. The relation of these dimensions to the results of
earlier large-scale surveys and to a variety of demographic indexes is explored. It is
found that there are continuing substantial differences in modal cultural values of
organization employees and that these are largely consistent with differences reported by
others. The present results suggest that the dimensions defined by Hofstede as
individualism-collectivism and power distance may be better defined as representing
varying orientations toward continuity of group membership (loyal
involvement/utilitarian involvement) and varying orientations toward the obligations of
social relationship (conservatism/egalitarian commitment).
Most psychological research has been undertaken in Western nations, particularly in North
America (Lonner, 1989; Sampson, 1985; Smith & Bond, 1993). Smith and Bond (1993) note the
frequent failures to replicate social psychological findings from American studies in other
national cultures, thus throwing some doubt on the cross-cultural validity of Western theories of
such phenomena as leadership, conformity, group decision making, attribution theory, and
intergroup relations.
Reflection on the culture-bound nature of much empirical and theoretical psychology underlines
the need for studies in which the conceptualization and measurement of culture is given a central
role. A major goal of cross-cultural psychology has been the identification of dimensions of
cultural variation (e.g., Leung & Bond, 1989; Schwartz, 1994). This is an important goal because
it opens the way to more adequate operationalizations of the concept of culture. The identification
of reliable dimensions of cultural variation should help create a nomological framework that is
both capable of integrating diverse attitudinal and behavioral empirical phenomena and of
providing a basis for hypothesis generation. Such a framework would help explain, for example,
why some replications of North American studies are successful whereas others are not, as a
function of cultural background. It should also enable researchers to select cultural groups for
study on an a priori basis, according to their positioning on relevant dimensions (Bond, 1988;
Leung, Bond, Carment, Krishnan, & Liebrand, 1990). Informed choices of this kind can enable
certain features of culture to be held constant while the effect of varying others is examined.
To identify such dimensions, it is desirable for studies to include as many and as wide a range of
cultures as possible. Most extant cross-cultural work has been confined to a small number of
cultures, though there are notable exceptions (Buss et al., 1990; Williams & Best, 1990; as well as
others now to be discussed). Three pioneering research projects have aimed directly at identifying
cultural dimensions of values, namely the projects of Hofstede (1980, 1983, 1991), Bond (1988;
Chinese Culture Connection, 1987), and Schwartz (1990, 1994; Schwartz & Bilsky, 1987, 1990).
The purpose of the present article is to examine the replicability of their conclusions and to
examine whether clarifications are required of the dimensions defined by these authors. Three
issues are of particular interest. First, a questionnaire designed without direct reference to the
results of existing large-scale cross-national studies is used to check the replicability of
dimensions identified earlier. Second, the data bank on which this study is based makes it
possible to determine whether demographic attributes earlier associated with nations scoring high
on, for instance, individualism or collectivism continue to do so. Third, we can examine whether
the scores obtained by specific nations cluster similarly to those obtained earlier.
We shall first examine the prior studies before outlining the basis of the present survey.
Hofstede’s Value Dimensions
Hofstede’s (1980) well-known study was based on the responses of 117,000 personnel from a
large American-owned multinational company in the period between 1967 and 1973 to a
questionnaire containing items predominantly tapping work-related values. On the basis of an
ecological factor analysis of mean responses from 40 nations on 14 items concerning the
importance of different work goals, Hofstede identified two factors that he labeled individualism
and masculinity. A further two dimensions of national culture labeled power distance and
uncertainty avoidance emerged from a so-called eclectic analysis, combining items largely on the
basis of theoretical expectations. Hofstede’s analysis of his data bank was later expanded to 53
cultures (Hofstede, 1983).
Hofstede’s (1980) power distance dimension is defined in terms of the prevailing norms of
inequality within a culture. Individualism-collectivism refers to the extent to which the identity of
members of a given culture is shaped primarily by personal choices and achievements or by the
groups to which they belong. Masculinity-femininity corresponds to a “tough-tender” dimension.
In masculine cultures, values such as competition, success, and performance are relatively more
prevalent than in feminine cultures, where there is relatively more emphasis on values such as
warm social relationships, quality of life, and care of the weak. The fourth dimension, uncertainty
avoidance, alludes to the degree to which members of a culture are uncomfortable with
uncertainties in life. Societies high on this dimension prefer structured rather than unstructured
situations, where there are clear guidelines for behavior.
Validation of the Hofstede (1980) constructs was achieved through establishing the significance
of their correlations with geographic, economic, and social indicators. The association between
individualism and economic development was particularly strong, with collectivist nations
tending to have lower per capita gross domestic product.
Hofstede (1980) reports a high correlation between the power distance and individualism
dimensions. Indeed, his ecological factor analysis based on 32 value items resulted in the
extraction of three rather than four factors, corresponding to the masculinity and uncertainty
avoidance dimensions and a combination of power distance and individualism. Thus, although
Hofstede preferred to separate these latter two dimensions conceptually, it appears that
empirically speaking they may be manifestations of the same underlying dimension.
Hofstede’s (1980) dimensions have been extensively invoked by researchers to help explain
cross-cultural differences (Kagitcibasi & Berry, 1989; Smith & Bond, 1993), despite which his
study has not escaped criticism. The range of countries sampled can theoretically affect the
dimensions emerging from studies of this type, and Hofstede notes the lack of samples from
(former) communist nations. Nevertheless, his study remains to this day the most comprehensive
survey in terms of the number of national cultures sampled. Another criticism is that the values
sampled were not comprehensive, so that the dimensions identified may not be exhaustive.
Hofstede (1991) adds a fifth dimension relating to time perspective.
Hofstede’s study (1980) has also been criticized on the basis of the type of samples drawn from
the national cultures, because all respondents shared a common corporate culture, which
distinguished them from the broader national populations from which they were drawn. Hofstede
regarded his matching strategy as a strength, though doubts can be raised as to whether those
employed in servicing and marketing in an industrialized nation are necessarily equivalent to
those in similar roles in a Third World nation. The latter will tend to differ from the former in that
they may have higher levels of wealth and social status relative to the rest of the national
population. However, in the absence of random population samples, a sample matching strategy
is a necessity (Schwartz, 1994). Evidence for the durability of core cultural dimensions can then
be shown by the extent that similar patterns of findings emerge from different types of samples,
different time periods, and measures from different domains of social behavior.
Other Studies of Cultural Dimensions of Valuing
These results leave open the possibility that the measures Hofstede (1980) used may have in
some way reflected the Western values of those who designed them. To test for this possibility,
Michael Bond in Hong Kong sampled a different domain of values, derived from Chinese culture.
He assembled a group of researchers named the Chinese Culture Connection (CCC; 1987), who
presented these to student samples from 22 countries. Respondents were required to rate the
importance of each value. After standardization within each culture, an ecological factor analysis
yielded four factors, labelled integration, Confucian work dynamism, human-heartedness, and
moral discipline.
Twenty countries were common to this study and Hofstede’s 1983 study, enabling correlations
between the Hofstede and CCC factors to be made. Moderate correlations (.50 to .65) were
reported between the Hofstede power distance and individualism dimensions and the CCC
integration and moral discipline factors. These four factors loaded together in a second order
factor analysis, suggesting that they represent aspects of the same underlying dimension of
individualism-collectivism. The CCC also reported a correlation of .67 between Hofstede’s scores
on masculinity and the CCC human-heartedness factor.
The convergence in findings from these two studies is impressive, given that they used measures
derived from two very different cultures and that they used different types of samples. However,
cultural variation tapped by Hofstede’s (1980) uncertainty-avoidance dimension was not
approximated by the CCC factors, whereas the Confucian work dynamism factor was unique to
the CCC study. This latter factor, which emphasizes Confucian work ethics such as thrift and
persistence, was found to relate highly and positively to recent data for national economic growth.
Bond (1988) reanalyzed the CCC data bank at the level of individuals rather than cultures. Using
factor analysis, he extracted two factors. The first of these corresponded very closely in terms of
factor structure to the CCC integration dimension, with one pole consisting primarily of items
emphasizing cohesion with others in general and the other pole consisting of items denoting
loyalty to more narrowly defined in-groups and their customs. Bond concludes that the
“convergence between the cultural and individual levels of analysis indicates the presence of a
strong universal” (p. 1012).
Schwartz (1992, 1994) surveyed value preferences of individuals in 25 countries, most of which
contributed two samples, secondary school teachers and students. The 56 values included were
selected on the basis of theoretical reasoning about value types and previous theoretical and
empirical studies from both Western and non-Western sources.
Schwartz (1992) reported parallel analyses from each of his samples, using smallest space
analysis to represent the patterns of intercorrelations between the rated values. The analyses
revealed a remarkable degree of similarity in the value structures of the diverse samples, as
indicated by similarity in the ordering of the values in two-dimensional multidimensional scaling
solutions. For the vast majority of the samples, the solutions could be partitioned into regions,
such that within each region the values items corresponded to 1 of 10 previously hypothesized
value types, namely, universalism, benevolence, tradition, conformity, security, power,
achievement, hedonism, stimulation, and self-direction. Moreover, these regions themselves
displayed nearly identical patterns of association in each sample, which could be summarized in
terms of two bipolar dimensions. The first of these was labeled openness to change versus
conservation. Openness to change embraced the self-direction and stimulation value types,
whereas conservation included the security, conformity, and tradition value types. The other
dimension was labeled self-enhancement versus self-transcendence, with the former including the
hedonism, power, and achievement value types, and the latter consisting of the universalism and
benevolence value types.
Schwartz (1994) further analyzed his data at the cultural level, using those 45 values previously
shown to have cross-culturally equivalent meanings for individuals, in an attempt to identify
“derived-etic” dimensions of cultural variation (Berry, 1989). As hypothesized, the same two
fundamental dimensions of variation were found to structure the value items.
The two dimensions reported by Schwartz (1992, 1994) were further divided into seven value
types. Three value types broadly corresponded to the openness to change versus conservatism
dimension. Conservatism values (e.g., obedience, family security, respect for tradition)
emphasized maintenance of the status quo, propriety, and avoidance of actions or inclination of
individuals that might disturb the traditional order. Intellectual autonomy (e.g., creativity, broadmindedness) and affective autonomy (e.g., pleasure, exciting life) reflected “a more intellectual
emphasis on self direction and a more affective emphasis on stimulation and hedonism” (1994, p.
102), respectively. Schwartz distinguishes these clusters of values from the definitions of
individualism and collectivism used by others, though they clearly have some overlap.
Four value types were identified corresponding to the self-enhancement versus self-transcendence
dimension. Subsumed by the self-enhancement categorization, hierarchy values (e.g., influential,
social power) are viewed as “emphasizing the legitimacy of hierarchical role and resource
allocation,” whereas mastery values (e:g:, dating; capable, ambitious) “emphasize active mastery
of the social environment through self assertion” (Schwartz, 1994, p. 103). The selftranscendence pole embraces two value types, egalitarian commitment (e.g., freedom, equality,
social justice), “values that express concern for the welfare of other people” (p. 104), and
harmony (e.g., protect environment, world of beauty), consisting of values “emphasizing
harmony with nature” (p. 105). Whereas the two underlying dimensions are (presumably)
orthogonal, some of the value types subsumed under these dimensions are not. For example,
egalitarian commitment values, which form part of the self-transcendence dimension, are more
closely related to the openness to change than to the conservatism dimension. This is a
consequence of delineating a seven-fold typology of values in a two-dimensional space.
Schwartz (1994) correlated his value types with country scores on Hofstede’s (1980) dimensions.
The most important concordance was between the Hofstede power distance and individualismcollectivism dimensions and the Schwartz value types representing the openness to change versus
conservatism dimension, with these associations strongest for Schwartz’s student samples.
All of the studies reviewed have yielded a factor related to individualism-collectivism. Parallel
individual-level concepts have also been shown to have predictive validity within cultures
(Triandis, 1990). It is probably safe to infer that this dimension is the most important yield of
cross-cultural psychology to date, though conceptual dichotomies paralleling individualismcollectivism predate the emergence of the discipline and can be traced back at least to founders of
the social sciences at the turn of the century such as Tonnies (1887/1963) and Durkheim
(1893/1960). Smith and Bond (1993) explore ways in which the individualism-collectivism
construct can be used to integrate a wide range of cross-cultural social psychological findings,
although due attention is also needed to obvious limitations on the predictive validity of any
single construct in accounting for the existing global diversity of social behavior (e.g., Schwartz,
Trompenaars (1985) devised a questionnaire based largely upon the delineation of cultural and
personal “pattern variables” or value dilemmas identified by Parsons and Shils (1951). Some of
the questionnaire items elicit values by asking respondents how they would act in a series of
briefly described imaginary situations. Others provide a forced choice between two value
statements referring to aspects of organizational behavior or more general issues. The
questionnaire was originally designed to tap the values of employees from energy and hosiery
companies in nine countries. It has subsequently been used in a very much wider range of
countries (Trompenaars, 1993). Most cross-cultural studies concerning organizations and their
employees focus on a small group of nations, in particular the United States, Japan, and a group
of Western European nations (Smith, 1992). The Trompenaars database provides a rare
opportunity to examine the values espoused by organization members across a range of national
cultures as broad as that sampled by Hofstede (1980).
The Trompenaars (1993) questionnaire includes items addressing seven hypothesized dimensions
of cultural valuing. The first five of these are derived directly from Parsons and Shils (1951),
namely universalism-particularism, achievement-ascription, individualism-collectivism,
affectivity-neutrality, and specificity-diffuseness. Choice of the remaining dimensions was
influenced by the value survey of Kluckhohn and Strodtbeck (1961), but the actual measure of
internal-external control that was used is that of Rotter (1966), and the measure of time
perspective is derived from Cottle (1968). Additional items in the questionnaire refer to aspects of
management and organizational structuring. This last group of items is not seen as tapping a
separate domain of values but as illustrating the manner in which more general values are applied
to day-to-day organizational problems.
The present study concerns primarily the first three of these proposed dimensions, together with
the management and organization items. Data derived from the Rotter (1966) locus of control
scale are examined in a separate publication (Smith, Trompenaars, & Dugan, 1995). Data from
the remaining items do not lend themselves readily to the type of data analyses used in this
The concept of universalism-particularism is derived from the work of Parsons and Shils (1951).
They distinguish two types of value standards that may guide behavior of persons or of whole
cultures. Persons relying on particularistic value standards will emphasize relationships to
particular people to a greater extent than persons with universalistic value standards, who will be
guided relatively more often by standards independent of specific social relationships. “The
particularistic actor predominantly values interpersonal ties, while the universalistic actor values
abstract societal expectations” (Trompenaars, 1985, p. 84). Parsons and Shils saw individualismcollectivism (which they termed self vs. collectivity) and universalism-particularism as separate
dimensions. Given the previous discussion that emphasized the importance of in-groups in
collectivist societies, it is more plausible to suppose that particularist societies will also be
collectivist. There is less reason to expect that a universalist society would necessarily be
individualistic, because this would only be the case where the universal principles to be followed
endorsed individualist values. These considerations suggest some association between the two
dimensions but not a total overlap. Indeed, this prediction is supported by the results of Zurcher,
Medow, and Zurcher (1965) who found Mexican bank employees much more particularistic than
U.S. bank employees, using the questionnaire devised by Stouffer and Toby (1951), on which the
present measure is also based.
Parsons and Shils’s (1951) concept of achievement-ascription refers to the characteristics of
persons that determine their status. Achieved statuses can be “filled” through ability, effort, and
competition, so that social mobility is possible. For example, the status of an athlete will be
achieved. Ascribed statuses, on the other hand, are largely predicated on who a person is.
Excellent examples are provided by the British monarchy and the Indian caste system.
The third domain of values in Trompenaars’s (1985) questionnaire was individualismcollectivism, a concept that we have already discussed. The definitions provided here make it
likely that there is also some linkage of individualism-collectivism and achievement-ascription.
Striving for achieved status is much more likely to occur in individualist societies than in
collectivist societies. The findings of Williams and Best (1990) that the relative rigidity of sex
roles is higher in more collectivist cultures also fits into this conception.
We can now state our hypotheses as to likely structures and interrelationships among the
available measures, drawing upon the results of the various earlier surveys and the preceding
discussion of Trompenaars’s (1985) intentions in designing his questionnaire. The first issue at
stake is whether or not the dimensions found can be predicted on the basis of earlier work:
Hypothesis 1a: A dimension of cultural variation will be found that is significantly
related to individualism-collectivism/power distance.
Hypothesis 1b: A second dimension, related to masculinity-femininity, humanheartedness, or self-enhancement/self-transcendence, can also be expected.
Because the subjects of this survey are organizational employees, we may test the face validity of
the predicted individualism-collectivism dimension by examining preferences for different types
of organizational behavior in countries located at different points along the dimension:
Hypothesis 2: The collectivism pole of an individualism-collectivism/power distance
dimension will be associated with items concerning preference for closer bosssubordinate relations, higher company involvement in employees’ lives, preferences for
ascription over achievement values, and a relatively particularistic perspective.
To establish the external validity of emergent dimensions, tests of their relationship to
independent country-level measures are required.
Hypothesis 3: The ordering of nations on dimensions related to individualismcollectivism will correlate with indexes of modernity such as per capita gross national
product, literacy rates, and life expectancy.
The 8,841 respondents used in this study were drawn from 43 nations. They were selected from a
larger data bank comprising 10,993 employees of business organizations from 58 nations. The
data were subject to several screening processes. A total of 9,920 employees replied to a complete
or near complete form of the questionnaire described below, with the remainder taking a
substantially shortened version. Respondents taking the shortened version were excluded from the
study. The data from 12 nations were also excluded because there were fewer than 25 cases for
each of these countries. Three further countries were excluded, namely Nepal, Egypt, and
Uruguay. A back-translation revealed substantial inaccuracies in the Nepalese translation that had
been used, and there was reason to doubt whether the Egyptian data collection had been
competently executed. The 36 Uruguayan responses were found to be extreme outliers on the
distribution of many variables. They were excluded on the basis of statistical tests, to be reported
in the Results section, which indicate that the extremity of the scores make it probable that
Uruguay was not accurately represented in the sample. Four hundred thirty-seven individual
respondents with more than four data points missing on the 39 input variables, and 29 whose
coding for gender was missing, were also excluded.
Of the remaining respondents, 2,269 (24.2%) were categorized as having lower socioeconomic
status, such as manual and clerical workers, and 5,082 (54.2%) were categorized as managerial or
professional workers (predominantly the former). A further 2,025 (21.6%) respondents could not
be categorized at this point for occupational status, most often because the version of the
questionnaire that they had received did not request this information.
A breakdown of the remaining sample sizes for each country by gender is shown in Table 1. The
sample overrepresents European nations (n = 24), with 72.4% of all respondents being
Europeans. African, Asian, and South American countries are underrepresented, constituting
3.2% (n = 3), 17.2% (n = 11), and 4.7% (n = 3) of the sample, respectively.
The overall sample is also biased by gender, with males constituting 68.9% of the total.
Aggregating data by country and giving unit weighting to each country, the overall proportion of
males is .68, with a standard deviation of .14. Sixteen country samples include 75% or more of
males. Countries with relatively high proportions of females (40% or more) are Argentina,
Bulgaria, former Czechoslovakia, former East Germany, Finland, Hungary, Indonesia, Mexico,
the Philippines, Rumania, the former Union of Soviet Socialist Republics (U.S.S.R.), and the
former Yugoslavia. Thus the gender distribution is much more equal in the former communist
bloc countries than in most of the other countries.
Mean age by country ranges from a low of 26.5 years in the former Yugoslavia to 43.2 years in
the Norwegian sample. Aggregating the data by country, mean age is 36.1, with a standard
deviation of 4.2. The sample is therefore relatively homogeneous in terms of age distribution.
Universalistic Versus Particularistic Obligations
The universalism-particularism construct was measured using modified versions of four items
devised by Stouffer and Toby (1951). They used stories describing life situations in which
individuals are asked to state their preference between their obligations of friendship to an
individual (particularistic obligations) and their obligations to society (universalistic obligations).
For example, the first item (Unpa 1) describes a situation where the respondent is required to
imagine he or she is in a car being driven by a close friend (but not a relative). The friend hits a
pedestrian while exceeding the 20-mile-per-hour speed limit, and there are no witnesses. The
friend’s lawyer says that if the respondent testifies that the friend was within the legal speed limit,
it may save the friend from serious consequences. The respondent is required to report what he or
she would probably do in this situation in view of his or her obligations to the friend and the
obligations of a sworn witness. Would or would not the respondent testify that the friend was
exceeding the speed limit? The respondent is also required to state whether the friend in the given
situation has a “definite right,” “some right,” or “no right” to expect the respondent to testify in
their favor. Thus the items tap both behavioral intentions and perceptions of norms.
The other three dilemmas have a similar structure. One (Unpa2) concerns a situation where the
friend in this case runs a restaurant and the respondent is asked to imagine that he or she is a
journalist who reviews restaurants. The food is poor. Should the respondent report the truth of the
matter or not? In another (Unpa3), the respondent is asked to imagine that he or she is a doctor for
an insurance company. The friend in this instance is being examined for eligibility for more
insurance. The examination reveals some health problems. Should the doctor ignore these for the
sake of his or her friend’s needs? In the final situation (Unpa4), the subject is an insider dealing in
financial markets, where a friend will be financially mined unless tipped off by the respondent
about the outcome of a confidential meeting.
Achievement Versus Ascription Orientation
The six items for measuring achievement-ascription were designed by Trompenaars (1985) and
are freely adapted from measures used by Karl (1965) and by Kluckhohn and Strodtbeck (1961).
These items were presented in the form of five-point Likert scales with strongly agree and
strongly disagree anchoring the endpoints. The statements were as follows: “The most important
thing in life is to think and act in the ways that best suit the way you really are, even if you don’t
get things done” (Achasc 1); “The respect a person gets is highly dependent on the family out of
which they come” (Achasc2); “When someone is born, the success they are going to have is
already in the cards, so they might as well accept it and not fight against it” (Achasc3); “A child
should be taught from infancy to be more gentle with women than with men” (Achasc4); “It is
important for managers to be older than most of their subordinates” (Achasc5); “Older people
should be more respected than younger people” (Achasc6).
Individualism-Collectivism and Conceptions of Organizational Structure
A further 25 questions were devised to measure facets of individualism-collectivism and various
ways of conceiving of the organizational structure of business organizations. Several of the
individualism-collectivism items were heavily modified versions of those used by Kluckhohn and
Strodtbeck (1961). All these items were phrased in the form of dilemmas, with a dichotomous
forced-choice response format.
Examples of these dilemmas are given where they prove relevant in the Results section of the
article. In general terms, the topics covered include the extent of company involvement in the life
of individual employees; for example, in providing housing or organizing social functions, the
working relationships of subordinates with their superiors, optimal modes of departmental
organization and job assignment, valued characteristics of managers, ways of dealing with
unsatisfactory work of employees, and the nature of contractual obligations.
Other Measures
The questionnaire included an additional 40 items referring to the other four domains of cultural
valuing postulated by Trompenaars (1985). Data from each of these will be included in the
present analysis: mean scores on the Rotter (1966) locus of control scale, time perspective,
affectivity-neutrality, and preference for specific versus diffuse social relationships. The time
perspective score used here is derived from a procedure used by Cottle (1968). The respondent is
asked to list his or her 10 most important life experiences and to indicate whether they fall in the
past, the present, or the future. Scores for affectivity-neutrality and for preference for specific
versus diffuse relationships were both derived from different responses to the same four
questions, which asked directly about preferred type of relationships with others. Scores on these
dimensions are thus not independent of one another. Respondents also reported their sex, age,
employer, job function, and educational level.
All items were translated into the required languages. The method of back-translation (e.g.,
Brislin, Lonner, & Thorndike, 1973) was employed for most of the translations. In the few cases
where it was not, the English and translated versions of the questionnaire were examined and
where necessary amended by bilingual company representatives. The questionnaires were
administered to respondents at varying dates between 1982 and 1993, using the language versions
indicated in Table 1. In most countries, subjects were participants in brief training courses
concerning management in cross-cultural contexts. Questionnaires were completed prior to
participation in the training.
Responses to the 39 relevant questionnaire items were used as the primary basis of the analysis.
Although within-subjects data standardization is an increasingly favored procedure in this type of
study, this was not done in the present case. There were two reasons for this. First, the forcedchoice response format of almost all items eliminates the risk of cultural variations in the use of
more or less extreme response categories, which is likely where Liken-type-scale responses are
used. Second, the range of questionnaire items included spanned a variety of scales constructed
on the basis of hypothesized differences grounded in theory. To standardize across scales would
be very likely to eliminate variance that is substantive rather than artifactual.
The scores for each item were first aggregated by nation, yielding a 39 (items) by 43 (nations)
data matrix. Each item was scaled so that all had a potential range of between 0 and 100. The
items with the most variability should therefore contribute most to the differences in dissimilarity
coefficients. The SPSS proximities program was used to compute dissimilarity measures using
the City Black metric (Coxon, 1982) between each pair of nations on the basis of their scores on
the 39 items. One feature of this metric is that equal dissimilarity coefficients are assigned to two
objects that are (a) two units apart on each of two variables, or (b) one unit apart on one variable
and three units apart on the other (Everitt, 1980). A further feature is that multidimenstional
scaling (MDS) solutions based on this metric do not remain invariant under rigid rotations. The
resulting dimensions account for more variation in the proximities data than any variation
resulting from rotation of the solution. This is not true of Euclidean metrics (Coxon, 1982).
The symmetric proximities matrix provided the input to the SPSS Alscal MDS package, which
was programmed to produce nonmetric scaling with the primary approach to ties. The resulting
MDS configuration was then interpreted using linear multiple regression (Kruskal & Wish, 1976).
This procedure enabled us to examine the association between the configuration and the original
variables from which the proximities data were calculated. This method also facilitates validation
of the resulting configuration by relating it to other measures not used in obtaining the
dissimilarities, such as dimensions of cultural variation revealed from other studies. In this
procedure, the various measures are regressed onto the coordinates of the MDS configuration.
Each of the original 39 measures was used as a separate criterion variable. In addition, a number
of other variables included in the questionnaire were entered, such as proportion of women and
mean age of sample. Hofstede’s (1983) country scores for his four dimensions of cultural
variation were also entered, as were Schwartz’s (1994) country scores on seven dimensions of
cultural values, and scores for the dimension of “Confucian work dynamism” identified by the
Chinese Culture Connection (1987). Finally, a number of economic and demographic
characteristics of nations were used as criterion measures, including indexes of average income,
literacy rates, life expectancy, proportion of economically active women, proportion of
population engaged in agricultural production, and proportion of Christians in the population
(Economist book of vital world statistics, 1990). Christianity was used as an index of religious
affiliation because the sample contains many more Christians than adherents of other major
For a variable to facilitate a reasonable interpretation of an MDS dimension, it is desirable that
multiple correlations should exceed the .01 level of significance and preferably exceed .70 in
magnitude (Kruskal & Wish, 1976). A large number of measures in this study do exceed these
criteria. A further condition is that variables used to interpret dimensions should have high
regression weights for those dimensions. We have predominantly used measures that have
direction cosines greater than .80 (direction cosines are regression weights normalized, so that
their sums of squares equal unity). However, this condition was not fulfilled in relation to the
third dimension of the solution described below.
As an additional check, the results of the MDS analysis were compared with the results from a
factor analysis using the same input variables.
MDS solutions of dimensionality one through four were computed. The Kruskal stress formula
one and R[sup 2] values, respectively, were .097 and .942 for the four-dimensionalsolution, .132
and .910 for the three-dimensional solution, .201 and .841 for the two-dimensional solution, and
.325 and .714 for the one-dimensional solution. Examination of a plot of stress against
dimensionality revealed no discernible “elbow,” a feature that has sometimes been used to
indicate the correct number of dimensions to extract (e.g., Kruskal & Wish, 1976). In choosing
which dimensionality gives the most satisfactory solution, it was clear that the stress value for the
one-dimensional representation was too high. The four-dimensional solution was rejected because
the fourth dimension of the four-dimensional solution was not readily interpretable either
intuitively or through the more objective criteria discussed earlier. The choice then was between
the two- and three-dimensional solutions. The three-dimensional solution revealed some
interpretable structure that was not present in the two-dimensional solution and was therefore
retained despite the fact that the extra dimension only accounts for a further 7% of variation in the
proximities data. It can be noted that nothing in the two-dimensional solution is lost by opting for
the three-dimensional solution, because the former is largely a perpendicular projection of the
first two dimensions of the latter. A comparison of the MDS solution with the factor analytic
solution showed that where two principal components were extracted and rotated using the
varimax method, and factor scores calculated, the product-moment correlation between country
scores on the first factor and the first dimension of the MDS solution was +.98, and between the
corresponding scores for the second dimension, -.93. Where three factors were extracted and
rotated, the convergence between the solutions was less impressive. The product-moment
correlations between the corresponding dimensions were calculated as +.65, -.68, and +.68 for the
first, second, and third dimensions, respectively.
The sample used to generate this three-dimensional solution is a relatively heterogeneous one in
terms of variables such as seniority, age, and gender. To check whether or not demographic
heterogeneity has contributed to the definition of the emerging dimensions, additional analyses
were conducted upon demographically defined subsamples, and these will be reported at relevant
points below.
The naming of dimensions is as much an art as a science. It may help the reader to follow our
presentation if we state at this point that our ultimate conclusion will be that Dimension 1
represents conservatism versus egalitarian commitment (terms that are taken from Schwartz,
1994) and that Dimension 2 is best represented as utilitarian involvement versus loyal
A plot of the first two dimensions against each other is shown in Figure 1. The figure shows some
clear geographical and sociopolitical grouping that can be observed without recourse to more
objective interpretive criteria. A striking feature is the clustering of the former communist nations
of Eastern Europe in or adjacent to the lower left quadrant. China (still formally a communist
nation) also appears in this quadrant. The lower right quadrant shows a clear clustering of
Northern European nations. Toward the positive pole of Dimension 1, and located around the
midpoint of the second dimension, there is a cluster of English-speaking and Latin nations. A
further cluster of nations are situated above 0.4 on the second dimension and around the 0.0 point
of the first dimension. These are predominantly Asian and African nations, though two of the
more collectivist European nations, Greece and Turkey, are also located within this cluster. The
relation of Dimensions 1 and 2 with Dimension 3 is not shown, but the most notable features are
the raising of Japan, South Korea, and Hong Kong toward the positive pole, and the lowering of
Indonesia, the United Arab Emirates, and Rumania toward the negative pole.
Table 2 shows the regression weights based on the three-dimensional solution, together with
direction cosines and multiple correlations. These indicate the extent to which the various
measures can be predicted by the three dimensions singly and in combination. The measures in
the table are those that were included as input to the SPSS proximities program. In examining
these items we may see which of them best define the dimensions of the MDS solution.
The items that are associated with the first dimension of the MDS solution are quite numerous.
The items measuring achievement-ascription orientation are very strongly associated with this
firm dimension. A particularly good predictor is Achasc4, the item stating that “a child should be
taught from infancy to be more gentle with women than with men.” However, the composite of
the six items (ACHASC) is an even better predictor of Dimension 1 scores, with a multiple
correlation of .91 and direction cosine of .92. Thus most of the variation of country samples on
the first dimension can be explained by the scores on these six items. The samples situated toward
the positive pole of the dimension tend to value achieved status over ascribed status, whereas
those toward the negative pole favor ascribed over achieved status. The country-level alpha
coefficient for the achievement-ascription scale, a simple summation of country means for the six
items, was calculated as 0.83.
The items that were designed to measure the concept of universalism-particularism also load
highly on this first dimension, though there are also smaller correlations with the other two
dimensions, particularly the third. Country samples situated toward the positive pole of the
dimension tend to be more universalistic, whereas those toward the negative pole are more
particularistic. It should be recalled that these items measured both behavioral intentions and
perceptions of norms. In the case of each of the four scenarios, the perceptions of whether the
friend has a right to expect favorable treatment are associated more closely with the dimension
than the behavioral intention. The index labeled RIGHTS in Table 2 is a weighted summation of
each of the four items pertaining to the friend’s right to preferential treatment. The country-level
alpha coefficient for this scale is 0.91.
Next, there are a number of perhaps less coherent forced-choice items that also relate strongly to
the first dimension. Question 18 has a very high association, as can be seen from the multiple
correlation and direction cosines. This item asks respondents whether or not they believe that a
company should take some responsibility for the housing of employees or whether this task
should rest solely on the employee. Countries toward the positive pole tend to believe that
housing is the sole responsibility of the employee. There may be no directly logical reason why
provision of company housing should covary with ascribed status and high particularism, but the
pattern of further responses presented below indicates a preference for a strong and long-term
involvement by the employer in many aspects of the employee’s life.
The response pattern to Question 18 is mirrored somewhat in that of Questions 16 and 19, though
the relationships are considerably weaker. Question 19 concerns the ways in which companies
should determine the income of their employees. In one view, the company should take account
of the size of the employee’s family, whereas the other view suggests that the employee “should
be paid entirely according to the work he is doing for the company.” This latter view is more
consistent with an emphasis on achieved status. Question 16 concerns the impending marriage of
a colleague. In one view, the company should be involved in throwing a party to celebrate the
marriage, whereas the alternative view emphasizes that marriage is a “family affair” and that the
company should not be responsible for organizing a party. The measure in Table 2 labeled
PATERNALISM is a summation of the four questionnaire items (11, 16, 18, and 19) that refer to
aspects of company paternalism in the life of the individual. It is highly associated with this first
Question 23 asked how employees should be compensated for working overtime. One response,
favored by country samples at the negative pole of the dimension, suggests that working overtime
belongs to the job and that the boss’s appreciation is reward enough, whereas countries at the
positive pole believe that overtime should be financially rewarded and regulated contractually.
This appears conceptually consistent with the meaning of universal-ism-particularism and
Question 21, which also loads highly on the first dimension, refers to the organizational structure
of work environments. One view, endorsed more often by samples at the negative pole, suggests
that “the main reason for having an organizational structure is so that everyone knows who has
authority over whom,” whereas the opposing response is that “the main reason for having an
organizational structure is so that everyone knows how functions are allocated and coordinated.”
The emphasis on formalized hierarchy is characteristic of societies where ascribed status is more
Question 13 asks what makes a good manager. In one view, endorsed more often in nations at the
positive pole, the manager facilitates the execution of tasks but intervenes only when necessary.
The other view endorses the idea of the manager as a paternalistic figure, “a kind of father” who
“guides his subordinates continuously,” gets subordinates working well together, and knows the
answers to most problems that crop up. Question 10 follows a similar pattern. Here respondents
are asked to imagine a situation where they are given an order by a boss that they believe to be
wrong. Samples toward the positive pole of the dimension think that the individual should be able
to point out the mistake to the boss, whereas at the negative pole, the view is that “in most cases
the boss is right. Even if he is wrong, one should do what he tells you to do, otherwise there will
be problems.” Responses to Question 10 also correlate moderately with the second dimension.
Thus, if we consider these items for the moment as referring to power distance, then the highest
power distance will be found at the upper left quadrant of the projection of the first two
dimensions, whereas the lowest will be found at the lower fight quadrant. The direction of fit to
the solution is the same as that of the country scores for Hofstede’s (1980) dimensions of
individualism-collectivism and power distance, which are discussed a little later.
A weighted combination of the measures labeled ACHASC, RIGHTS, and PATERNALISM
obtained a multiple correlation with the first dimension of .98 in a multiple regression analysis,
thus explaining virtually all of the variation upon it.
The items that are associated with Dimension 2 are primarily items that Trompenaars (1985)
intended to measure the concept of individualism-collectivism. For greater clarity, they are here
identified as contrasting involvement in a group or organization on the basis of loyalty or on the
basis of utilitarian considerations. The measures most closely associated with the second
dimension are Questions 39 and 8, as evidenced by the high multiple correlations and the high
direction cosines. Question 39 poses the dilemma of what a family should do with a father’s
business when he has died. Should the children keep the business and work together to improve it
(loyal response), or should they sell their individual shares to set up on their own (utilitarian
response). Those country samples scoring high on the second dimension were more likely to opt
for the loyal response, whereas the converse was true for the samples located toward the negative
pole of the dimension:
Question 8 invited respondents to consider whether the responsibility for a fault in an installation
caused by negligence of a team member should be apportioned to that individual alone, or
whether the responsibility should be carried by the group (“this time one member has made a
mistake, the next time someone else will make one”). Question 7 asked the sample what would be
most prominent in their thoughts if they were to be promoted, either the “new group of people
with whom you will work,” or “the increased work responsibility and the higher income.”
Another item associated closely with the second dimension of the solution is Question 36. This
asked respondents whether they would prefer to work in a job where “everybody works together
and where you don’t get individual credit” (loyal response), or in a job “where you are part of a
company that allows everybody to work individually and where individual credit can be received”
(utilitarian response).
A further item loading on this dimension is Question 20, which asked the respondent whether
they thought the best representation of the goals of a company was, on one hand, that “the only
real goal of a company is making profit,” or, on the other hand, that a company “besides making
profit, has a goal of attaining the well-being of various stakeholders such as its employees,
customers etc.”
For Question 17, respondents were asked to consider the situation in which the job performance
of a long-standing employee of 15 years has dropped to an unsatisfactory level- for the past year,
and where “there are no reasons to believe that this situation will improve.” The respondent has to
select from two options. In the first, job performance is the sole criterion for dismissal, whereas
the second suggests that the employee’s past performance and previous years of service should be
considered, because “one has to take into account the company’s responsibility for his life.”
Finally, Question 37 refers to a situation in which local government comes up with a project to
close the street on which you live to all traffic. The street members are to send a delegate to a
meeting where a decision on the project will be made. How should the delegate be chosen? The
response alternatives are that “it is better that all the people in the street meet and discuss things
until almost everyone agrees on the same person,” versus “it is better that all people in the street
meet, names be put up, a vote taken then send a man who gets the majority of the votes even if
there are several people who are still against this person.”
The relationship of these items to the conceptions of individualism-collectivism discussed earlier
is self-evident, but because this is true also for some items loading on Dimension 1, alternative
terms have been preferred here. In Table 2, the measure labeled INVOLVEMENT is a summation
of Questions 8, 17, 20, 36, 37, and 39. The correlation of this measure with the second dimension
is -.85.
The third dimension is not so easily defined. There are fewer variables that correlate significantly
with it, and those that do are associated at a much lower level than is the case for either of the first
two dimensions. Furthermore, most of the variables that load highest on the third dimension are
also associated, more strongly in some cases, with one or the other of the first two dimensions.
According to Kruskal and Wish’s (1976) guidelines for the interpretation of MDS dimensions,
attempts to define the third dimension are therefore problematic. The difficulties of a dimensional
interpretation are unsurprising, given the distribution of cases on the third dimension, which
shows most cases clustered around the origin. However, we shall describe the measures that are
most closely associated with this third dimension.
From Table 2, we can see that there is a small tendency for country samples with high positive
scores on Dimension 3 to give more particularistic responses. Respondents from these countries
also tend to think that a department’s best work is done when members agree on objectives, and
then the individual is left to decide how these are implemented, rather than the opposing view that
the manager should set objectives and constantly direct employees in their work (Question 15).
Another item with a moderate loading is Question 36, which has already been described in
connection with Dimension 2. High scoring samples on Dimension 3 state that they would prefer
to work in a job where people can work individually and where individual credit can be received,
rather than a job where everyone works together and individual credit is not given for one’s work.
Finally, we can observe a small tendency (though not entirely consistent) for samples higher on
Dimension 3 to be oriented more toward achievement than ascription. This is most evident on
item Achasc6, which states that “older people should be more respected than younger people.”
Discussion of what these items have in common is best deferred until we have examined further
correlates of Dimension 3.
Having established the nature of the dimensions that emerged from the data, we are now in a
position to examine how these dimensions relate to characterizations of the relevant nations and
their distinctive cultural values that are available from other sources. These relationships are
portrayed in Table 3.
Possible Effects of Demographic Qualities of the Sample
We must consider first whether the solution found has been in any way influenced by the
nonrandomness with which respondents were selected. Two characteristics of the actual country
sampler mean age and proportion of women, are associated primarily with the first dimension.
Samples tend to be older toward the positive pole and to contain fewer women. Proportion of
women is also significantly negatively correlated with the second dimension. The correlation with
gender can be attributed to the considerably larger proportion of women in samples from the
former communist nations of Eastern Europe, which as we have seen cluster together toward the
negative pole of the first dimension. This finding suggests the hypothesis that these countries are
contiguous in the multidimensional space because of their different gender distributions.
To address this possibility, a further MDS analysis was computed using country samples with 25
or more cases and based only on males (n = 39 countries). The resulting configuration is virtually
identical to that presented here, with the product-moment correlations between the corresponding
dimensions being .96, -.89, and -.85 for the first, second, and third dimensions, respectively. As a
consequence, we should reject the hypothesis that gender differences in the values measured here
have strongly influenced the dimensions that emerged. It appears more likely that the association
of gender with Dimensions 1 and 2 is a consequence of differences between countries in the
sexual division of labor.
It is necessary to clarify in a similar way the reasons for the association between age and
Dimension 1. An additional MDS analysis was conducted using 42 countries within which the
oldest and youngest cases were deleted until a mean was achieved between ages 33 and 36. This
was impossible within the Norwegian sample. The resulting solution yielded three dimensions
correlating .96, .87, and .88 with the solution based on the total sample. This indicates that the
association of Dimension 1 is not because those who are older endorse egalitarian commitment
rather than conservatism; indeed, it would be surprising if it were so. The explanation is likely to
be that the samples from countries endorsing conservatism were younger, no doubt because many
developing countries have a younger population than the European and North American countries
where egalitarian commitment is more endorsed.
Respondents’ seniority was not associated with any of the three dimensions, but to test further
whether the demographic heterogeneity of the sample has affected the dimensions found, another
MDS analysis was conducted using only the 5,981 respondents who were of managerial status,
omitting the more junior respondents. The emergent dimensions correlated .96, .92, and .90 with
the solution from the full sample. There is therefore no evidence that diversity of country sample
in age, gender, or status has influenced the dimensions found.
Indexes of Modernity and Religious Affiliation
A number of variables that can be taken as indexes of “modernity” are moderately associated with
the dimensions of the MDS solution. These include life expectancy, literacy rates, proportion
engaged in agriculture, national income per capita, proportion of the population of higher
socioeconomic status, and the proportion of the population economically active. In general, the
strongest relationships are with Dimension 2, with which modernity is negatively related.
Modernity is also positively associated, though less strongly, with Dimension 1. The associations
of these measures with Dimension 3 are less clear, though life expectancy, literacy, and economic
activity rates are positively associated with it, and the proportion engaged in agriculture is
negatively related.
All of the measures of modernity are markedly nonnormal in distribution, and transformations do
little to improve the observed associations, with the exception of a logarithmic transformation of
per capita income, which strengthens relationships with the dimensions somewhat. There are a
number of quite marked outliers, which also tends to attenuate the observed associations. For
example, the United Arab Emirates and Singapore score highly on several indexes of modernity,
but they also score highly on collectivist values. Among European nations, the same can be said
of the location in the MDS space of Austria. Conversely, India and Mexico score less highly on
collectivist values than would be expected on the basis of their modernity scores. Thus we can
conclude that there is only a moderate relationship between modernity and the scaling of the
values observed in the present sample.
Proportion of Christians in the population is also moderately associated with the solution.
Countries toward the positive pole of Dimension 1 tend to have a higher proportion of Christians.
However, this measure is also quite strongly and negatively related to the second dimension.
Therefore, the lower right-hand quadrant of the projection of Dimensions 1 and 2 has the highest
proportion of Christians, and the upper left quadrant has the fewest. Cases that markedly diverge
from this pattern are Japan, China, India, Thailand, and Turkey, among predominantly nonChristian nations, and the former Yugoslavia, Rumania, Greece, and the Philippines, among
countries with a higher proportion of Christians. These countries in particular would be located
differently in the multidimensional space if there was a closer relationship between adherence to
Christianity and the values measured here.
Turning now to dimensions that have been found in other prominent studies of values, there is a
high association between Hofstede’s (1983) dimension of individualism-collectivism and both of
the first two dimensions of the MDS solution. As with the indexes of modernity described earlier,
the strongest relationship is with the second dimension, though there is also a moderate
association with the first. The countries toward the positive pole of the second dimension tend to
have a higher Hofstede collectivism score, as do countries toward the negative pole of the first
dimension. It should be remembered, however, that Hofstede’s sample did not include any
Eastern European nations, whereas these proved crucial in defining both the first and second
dimensions of the present results.
The Hofstede (1980) scores for power distance are also related to both of the first two
dimensions, but less strongly so than for his individualism index. This study provides no direct
confirmation of Hofstede’s other two dimensions, namely, masculinity and uncertainty avoidance,
despite the inclusion of some items that do seem theoretically coherent with his conception of the
meaning of these constructs.
It was noted earlier that the small data sample from Uruguay was dropped from the analyses
reported above. In fact, a parallel series of 44-nation analyses was also run, including the
Uruguay sample. When this was done, the correlations with the Hofstede (1983) dimensions are
substantially weakened. The multiple correlation with individualism becomes .76, yielding a
reduction in R[sup 2] of .08, proportionately a very large decrease caused by the inclusion of just
37 cases within the sample of 8,878.
Examination of residuals statistics for the regression of individualism on the three dimensions of
the MDS solution clearly show Uruguay to be an outlying case. The value of Cook’s distance
(Glantz & Slinker, 1990) for Uruguay is 1.81, compared with the next largest value of 0.25 for
Indonesia, with the mean Cook’s distance for all cases 0.08. According to Glantz and Slinker, a
Cook’s distance of 1 indicates a case that is worth investigating as an outlying case. On this
criterion, Uruguay is clearly an outlier. Until such time as further Uruguayan responses are
available to check whether the present small sample is in any way representative, it was judged
appropriate to discount the Uruguayan data.
Sixteen countries are common to this study and those sampled by Schwartz (1994). Only one of
Schwartz’s value dimensions is significantly associated with the dimensions found in this study,
namely, egalitarian commitment, which lines up primarily with the first dimension of the MDS
solution presented here, whereas conservatism falls just short of significance.
Only 18 countries are common to this study and that by the Chinese Culture Connection (1987).
The country scores for Confucian work dynamism abstracted from their study can be predicted
from scores on the third of the dimensions presented here. The presence of a significant
correlation with the third dimension is unsurprising when we remember that South Korea, Japan,
and Hong Kong are the most extreme positive scorers on this dimension and that they are
countries identified by the Chinese Culture Connection as scoring high on the values that they
label as Confucian work dynamism. Of the other CCC dimensions, the integration factor scores
can be predicted from a weighted composite of scores on Dimensions 1 and 3. As with Hofstede’s
(1980) masculinity factor, their was no significant association between our configuration
and the CCC human-heartedness dimension.
Other Measures
Rotter’s (1966) locus of control scale country mean scores are primarily associated with the first
dimension of the solution, with countries toward the positive pole tending to provide more
internal responses. The measure of time perspective included in the Trompenaars (1985)
questionnaire is significantly associated with the third dimension. As described earlier,
respondents were invited to recall or look forward to the 10 most important experiences in their
life and to categorize these according to whether they happened or will happen in the distant past,
recent past, present, near future, or distant future. Country samples scoring highly and positively
on the third dimension reported more experiences in the near future. The associations are not very
strong, but they are consistent with the findings of the CCC (1987), which identified a strong
future orientation as one facet of the cluster of values they defined as Confucian work dynamism.
Our results provide substantial encouragement for the view that there is considerable replicability
in the results emerging from value surveys sampling relatively large numbers of nations. The
present sample has the weakness that country samples are demographically diverse, but checks
that have been made through analyses of subsamples indicate that this is unlikely to have
substantially distorted the dimensions found. The three dimensions that emerged are significantly
related to nation scores collected at widely differing times, and among quite different types of
samples, by earlier researchers. From this perspective, the methodological weaknesses inherent in
each particular study are a source of collective strength: Dimensions that emerge consistently
despite such variations are plainly robust.
The dimensions found to show some relationship with those found in this study have been
individualism-collectivism, power distance, egalitarian commitment, integration, and Confucian
work dynamism. Hypothesis la is therefore supported. The linkage of these dimensions with
preference for specific organizational behaviors and country-level demographic indicators also
give support to Hypotheses 2 and 3.
The remaining dimensions identified by Hofstede (1980), namely uncertainty avoidance and
masculinity-femininity, are not readily apparent. Hypothesis 1b must be rejected. Three possible
reasons for this can be examined. First, it could be the case that the sample of nations included in
the present survey brings other aspects of cultural variability to the fore and leads to the loss of
uncertainty avoidance and masculinity-femininity as salient dimensions. Second, the present
questionnaire was not designed to tap these dimensions directly, and it would be unsurprising if
they were therefore less salient in the data. Third, the dimensions might have become less
important over time.
The first of these explanations has substantial plausibility. The present sample and the Hofstede
(1980) sample differ principally by the inclusion in the present study of nine former communist
countries. As Figure 1 makes clear, these nine nations are crucial in defining the poles of both of
the first two dimensions found in the MDS solution obtained. It is interesting that if these
countries are covered up in the figure, then the basic structure goes from the lower right to the
upper left quadrant, which is the best fit of our solution to the Hofstede individualism and power
distance dimensions. Eight of the nine ex-communist samples show a pattern of values that is
individualist, ascriptive, and particularist. Pursuing this intriguing result further, we find that if
the communist and ex-communist nations are deleted from the MDS analysis, then the scores on
the first dimension of the new solution correlate -.80 (p < .0001) with Hofstede’s individualism
scores and .68 with power distance (p < .0001).
It thus appears highly likely that at least some of the differences between our results and those of
Hofstede (1980) are due to divergences in the countries sampled. The emergence of achievementascription and universalism-particularism as the principal definers of the first dimension of the
MDS solution necessarily positions specific countries somewhat differently from those derived
from Hofstede’s results. The weightings of achievement-ascription and universalism-particularism
on the first dimension are both high but not identical. Although each of the two scales has good
reliability, it could nonetheless be the case that elements of unreliability in each attenuates their
relationship and that they are essentially measuring the same thing. This first dimension is most
readily interpretable in terms of Schwartz’s (1994) value domain of egalitarian commitment and
its polar opposite domain, conservatism. Schwartz has argued that these domains should not be
confounded with individualism-collectivism, and the present results support, this view. These
values concern personal rights and achievements versus social obligations, including the
obligation to respect hierarchy. However, the countries located at the positive pole of this
dimension are those that were characterized as most individualist by the Hofstede measures.
The items defining the second dimension of the present result are all concerned with the basis of
group membership. However, even if we discount the East European nations that did not appear
in Hofstede’s (1980) sample, the array of most individualist nations is no closer to his results than
is the array on Dimension 1. The nations of Northern (and Eastern) Europe, rather than the United
States or Australia, are the ones that consistently score toward the negative pole.
The Hofstede (1980) project yielded two dimensions that he found to be closely interrelated at the
country level of analysis, individualism and power distance. The present analysis also finds two
dimensions that both relate significantly to each of these Hofstede dimensions. We must assume
that this is because the manner in which each project defines these dimensions is in some way
interwoven. However, the pattern of results suggests that we can now obtain some clarification of
how best to define each dimension. The first dimension has to do with the nature of one’s
obligations to groups and organizations. Hofstede’s concept of power distance captures one
aspect, but the more general terms of conservatism, ascription, and particularism cover not just
hierarchy but the overall basis of one’s obligations toward all other in-group members. Schwartz’s
terms of conservatism and egalitarian commitment provide satisfactory labels for it. If we take the
second dimension as representing the basis of one’s commitment to a group, it may best be
defined in terms of a loyal and continuing commitment versus a utilitarian commitment that is
much more readily terminated.
The increasing popularity of individualism-collectivism as an explanatory concept in crosscultural psychology has been attended by a growing diversity in the manner in which the
dimension has been defined. Whereas some authors define the term primarily in terms of
continuity of group membership (e.g., Triandis, 1990), others focus more upon the values
governing one’s relations with others (e.g., Schwartz, 1992). Hofstede’s (1980) own
operationalization relied upon the opposition of the individualistic work goals of personal time,
freedom, and challenge to the more contextualized goals of training opportunities, good physical
environment, and chance to use one’s skills. The Hofstede operationalization does separate
different priorities in one’s interpretation of work-role obligations, but these priorities are open to
a variety of interpretations. It now appears insufficiently precise to guide the wider divergence of
definitions that has followed his pioneering study. The operationalization used by Hofstede may
explain why the present results show that his ordering of nations in terms of individualism
collectivism follows Dimension 1 (r = .70) equally as much as it follows Dimension 2 (r = -.74).
It is to be hoped that the differentiation of these two dimensions can help to reduce the degree to
which the two conceptualizations are confounded with one another and with the related concepts
now available. Both conservatism-egalitarian commitment and loyal involvement-utilitarian
involvement have great potential in clarifying cultural variations in social and organizational
behavior. The advantages of distinguishing these dimensions is further underlined by their
similarity to the concepts of”vertical” and “horizontal” individualism-collectivism proposed by
Triandis (1995). Whereas the present study provides some evidence of the utility of these
concepts in culture-level analyses, Singelis, Triandis, Bhawuk, and Gelfand (1994) present
evidence for their validity in individual-level studies.
The findings of this project are necessarily limited by the initial choice of questionnaire items.
Each of the three Parsonian pattern variables that we chose to examine did prove to relate
meaningfully to the results of other large-scale cross-national projects. In focussing our
discussion particularly on the first two dimensions obtained, we do not wish to propose that other
dimensions of cultural variation are not equally important or worthy of study. Our results have
suggested that despite the theoretical basis from which we started, there is substantial but
subsidiary variance within our data bank along other dimensions, such as those related to
Confucian work dynamism and uncertainty avoidance, and we intend to explore these more fully
in later publications.
AUTHORS’ NOTE: This article was accepted for publication during the editorship of John
Williams. We gratefully acknowledge the support of the Economic and Social Research Council,
Grant R 000 23 4336, and the helpful comments of Michael Bond, Geert Hofstede, and Shalom
Schwartz in the preparation of this article, Requests for reprints should be sent to the first author
at the School of Social Sciences, University of Sussex, Falmer, Brighton BNI 9QN, UK.
Details of the Country Samples
Legend for Chart:
A – Country
B – Number of Subjects, Female
C – Number of Subjects, Male
D – Number of Subjects, Total
E – Mean Age
F – Language Version
36 42 78 35.0 Spanish
12 27 39 40.8 English
22 39 61 37.0 German
90 290 380 37.6 French/Dutch
64 108 172 36.4 Portuguese
85 97 182 30.5 Bulgarian
Burkina Faso
31 100 131 33.9 French
108 175 283 34.4 Chinese
154 156 310 37.2 Czech
9 42 51 39.3 English
Ex-East Germany
106 134 240 42.1 German
7 57 64 42.7 English
54 58 112 37.2 English
37 258 295 38.4 French
43 76 119 32.6 Greek
Hong Kong
14 30 44 34.6 English
125 68 193 37.8 Hungarian
67 249 316 38.6 English
45 53 98 33.5 Bahasu
20 50 70 38.1 English
26 93 119 38.6 Italian
46 92 138 30.7 Japanese
80 119 199 33.7 Spanish
175 1,026 1,201 35.9 Dutch
16 67 83 37.6 English
4 25 29 43.2 Norwegian
8 72 80 42.6 English
26 30 56 34.8 English
89 142 231 39.3 Polish
23 40 63 34.5 Portuguese
143 133 276 34.6 Romanian
56 168 224 29.7 English
South Korea
22 79 101 29.6 Korean
23 92 115 39.2 Spanish
85 262 347 41.5 Swedish
15 33 48 36.2 English
37 101 138 35.2 Turkish
United Arab Emirates
11 23 34 28.7 Arabic
400 812 1,212 38.8 English
60 149 209 41.4 English
191 188 379 32.6 Russian
Ex-West Germany
22 160 182 40.2 German
51 41 92 26.5 Serbo-Croat
Total sample
2,747 6,094 8,841 — —
Regression Weights for Three-Dimensional Multidimensional Scaling Solution
Legend for Chart:
A – Item
B – Beta (Direction Cosines), Dimension 1
C – Beta (Direction Cosines), Dimension 2
D – Beta (Direction Cosines), Dimension 3
E – Multiple R
Achasc1 .67 (.99) .05 (.07) -.09 (-.13) .68[d]
Achasc2 .50 (.72) -.22 (-.31) .43 (.62) .69[d]
Achasc3 .68 (.83) .03 (.03) .45 (.55) .81[d]
Achasc4 .77 (.89) .31 (.36) .23 (.27) .86[d]
Achasc5 .60 (.89) -.26 (-.38) -.17 (-.25) .68[d]
Achasc6 .50 (.63) -.20 (-.25) .59 (.73) .79[d]
ACHASC .84 (.92) -.02 (.03) .35 (.38) .91[d]
Unpabehav1 .63 (.80) -.37(-.47) -.29 (.37) .79[d]
Unpabehav2 .08
Unpabehav3 — — — .42
Unpabehav4 — — — .42
UNPABEHAV .47 (.89) -.09 (.16) -.22 (.42) .53[b]
Unparight1 .75 (.81) -.29 (-.32) -.46 (-.50) .93[d]
Unparight2 .71 (.91) -.09 (-.12) -.31 (-.40) .78[d]
Unparight3 .77 (.89) -.04 (-.05) -.38 (-.44) .86[d]
Unparight4 .86 (.96) .12 (.13) -.23 (-.25) .89[d]
RIGHTS .86 (.91) -.10 (-.10) -.39 (-.41) .95[d]
UNPASCALE .82 (.90) -.10 (.11) -.37 (-.41) .91[d]
Question 7 .18 (.33) .36 (.67) -.36 (-.66) .53[b]
Question 8 -.17 (-.24) -.70 (-.94) .17 (.23) .74[d]
Question 9 .27 (.56) .24 (.50) -.31 (-.66) .48[a]
Question 10 .54 (.83) -.36 (-.56) -.02 (-.03) .65[d]
Question 11 .43 (.64) -.47 (-.71) .20 (.30) .66[d]
Question 12 .43 (.71) -.42 (-.71) -.02 (-.03) .60[c]
Question 13 .78 (.95) -.24 (-.29) .12 (.15) .83[d]
Question 14 — — — .22
Question 15 .46 (.70) .07 (.11) .46 (.70) .65[d]
Question 16 .51 (.89) -.16 (-.28) -.22 (-.37) .58[c]
Question 17 .28 (.58) -.40 (-.81) -.05 (-.10) .49[a]
Question 18 .82 (.92) .36 (.40) .01 (.01) .90[d]
Question 19 .55 (.85) -.18(-.28) -.29(-.45) .65[d]
Question 20 -.14 (-.27) -.51 (-.96) -.03 (-.06) .53[b]
Question 21 .64 (.94) .15 (.22) .18 (.27) .68[d]
Question 22 -.34 (-.48) -.46 (-.64) -.43 (-.61) .72[d]
Question 23 .62 (.93) .23 (.35) .04 (.07) .67[d]
Question 24 .42 (.78) .05 (.09) -.33 (-.61) .54[b]
Question 25 .38 (64) -.45 (-.77) .00 (.01) .59[c]
Question 35 — — — .33
Question 36 .15 (.15) -.57 (-.76) .48 (.64) .75[d]
Question 37 -.31 (-.41) -.51 (-.67) .46 (.62) .75[d]
Question 38 -.55 (-.79) -.41 (-.59) -.14 (-.19) .70[d]
Question 39 .32 (.41) -.70 (-.90) -.14 (-.18) .79[d]
Question 40 — — — .31
INVOLVEMENT -.03 (-.04) -.85 (-.97) .22 (.24) .88[d]
PATERNALISM .89 (.99) -.07 (-.08) -.08 (-.09) .90[d]
NOTE: Achasc 1 through Achasc6 measure achievement-ascription; ACHASC is a composite of
these six measures. Unpabehav 1 through Unpabehav4 measure behavioral intentions in relation
to universalism-particularism; UNPABEHAV is a composite of these four measures. Unparight 1
through Unparight4 measure perceptions of the rights of individuals in relation to universalismparticularism; RIGHTS is a composite of these four measures. UNPASCALE is a composite of
UNPABEHAV and RIGHTS. INVOLVEMENT is the sum of Questions 8, 17, 20, 36, 37, and
39. PATERNALISM is the sum of Questions 11, 18, 19, and 16. See Research Instruments
section in text for full description of these measures.
a p < .05; b p < .01; c p < .001; d p < .0001.
Loading of Other Measures on the Three-Dimensional Multidimensional Scaling Solution
Legend for Chart:
A – Beta (Direction Cosines), Dimension 1
B – Beta (Direction Cosines), Dimension 2
C – Beta (Direction Cosines), Dimension 3
D – Multiple R
(male) .25 (.46) -.37 (.68) .31 (.57) .54[j]
Life expectancy
(female) .25 (.43) -.42 (-.72) .31 (.54) .58[j]
Per capita
income (log) .32 (.64) -.34 (-.70) .19 (.38) .51[j]
Literacy rate .22 (.36) -.50 (-.80) .30 (.48) .62[k]
% population
in agriculture -.34 (-.68) .27 (.53) -.26 (-.50) .51[j]
% high
status in
population[a] .34 (.58) -.46 (-.80) .02 (.03) .64[j]
% economically
active[b] .11 (.21) -.34 (-.62) -.41 (-.75) .55[i]
% Christians
in nation .53 (.78) -.41 (-.61) .08 (.11) .66[l]
Mean age of
sample .73 (.92) -.26 (-.32) -.18 (-.23) .80[k]
% men in
sample .55 (.82) .36 (.54) -.11 (-.17) .60[l]
conservatism[c] -.58 (-.92) -.08 (-.12) -.23 (-.37) .59
commitment[c] .48 (.79) .33 (.56) -.14 (-.24) .69[i]
individualism[d] .36 (.58) -.51 (-.82) .02 (.03) .81[l]
Hofstede power
distance[d] -.30 (-.56) .44 (.82) -.03 (-.06) .69[l]
integration[e] .92 (.79) -.11 (-.10) .70 (.60) .84[k]
Connection moral
discipline[e] -.39 (-.94) -.09 (-.22) -.11 (-.25) .49
Confucian work
dynamism[e] -.58 (-.82) -.22 (-.31) .35 (.49) .79[j]
Locus of
control .43 (.92) .19 (.39) .02 (.05) .47[i]
Specificdiffuse[f] .22 (.54) -.32 (-.78) .13 (.32) .41
Neutralaffective[g] .39 (.98) -.03 (-.08) .07 (.18) .40
Present .15 (.87) .03 (.21) .06 (.37) .16
Near future[h] -.14 (.27) -.32 (-.61) .39 (.74) .52[j]a. n = 30.
b. n = 34.
c. Country means for conservatism and egalitarian commitment (n = 16) are from Schwartz
d. n = 33; power distance and individualism-collectivism scores for the United Arab Emirates,
Nigeria, and Burkina Faso are from Hofstede’s (1983) means for the Arab states and West Africa.
e. n = 17; country means are from Chinese Culture Connection (1987).
f. t for Dimension 2 = -2.21, p < .05.
g. t for Dimension 1 = 2.68, p = .01.
h. t for Dimension 3 = 2.80, p < .01.
i p < .05; j p < .01; k p < .001; l p < .0001.
GRAPH: Figure 1: Location of Countries on Dimensions 1 and 2 of Multi-Dimensional Scaling
Berry, J. (1989). Imposed etics–emics–derived etics: The operationalisation of a compelling
idea. International Journal of Psychology, 24, 721-735.
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, 1009-1015.
Brislin, R. W., Lonner, W. J., & Thorndike, R. M. (1973). Cross-cultural research methods. New
York: John Wiley.
Buss, D. M., et al. (1990). International differences in selecting mates: A study of 37 cultures.
Journal of Cross-Cultural Psychology, 21, 5-47.
Chinese Culture Connection. (1987). Chinese values and the search for culture-free dimensions of
culture. Journal of Cross-Cultural Psychology, 18, 143-164.
Cottle, T. J. (1968). The location of experience: A manifest time orientation. Acta Psychologica,
28, 129-149.
Coxon, A.P.M. (1982). The user’s guide to multidimensional scaling. London: Heinemann.
Durkheim, E. (1960). The division of labour in society. New York: Free Press. (Original work
published 1893)
The Economist book of vital world statistics: A complete guide to the world in figures. (1990).
London: Hutchinson.
Everitt, B. S. (1980). Cluster analysis. London: Heinemann.
Glantz, S. A., & Slinker, B. K. (1990). Primer of applied regression and analysis of variance.
New York: McGraw-Hill.
Hofstede, G. (1980). Culture’s consequences: International differences in work-related values.
Beverly Hills, CA: Sage.
Hofstede, G. (1983). Dimensions of national cultures in fifty countries and three regions. In J.
Deregowski, S. Dzuirawiec, and R. Annis (Eds.), Expiscations in cross-cultural psychology (pp.
335-355). Lisse, The Netherlands: Swets & Zeitlinger.
Hofstede, G. (1991). Cultures and organizations: Software of the mind. London: McGraw-Hill.
Kagitcibasi, C., & Berry, J. W. (1989). Cross-cultural psychology: Current research and trends.
Annual Review of Psychology, 40, 493-531.
Kahl, J. A. (1965). Some measurements of achievement orientation. American Journal of
Sociology, 70, 669-681.
Kluckhohn, F., & Strodtbeck, F. L. (1961). Variations in value orientations. Evanston, IL: Row,
Kruskal, J. B., & Wish, M. (1976). Multidimensional scaling. Beverly Hills, CA: Sage.
Leung, K., & Bond, M. H. (1989). On the empirical identification of dimensions for crosscultural comparisons. Journal of Cross-Cultural Psychology, 20, 133-151.
Leung, K., Bond, M. H., Carment, D. W., Krishnan, L., & Liebrand, W.B.G. (1990). Effects of
cultural femininity on preference for methods of conflict processing: A cross-cultural study.
Journal of Experimental Social Psychology, 26, 373-388. [Correction to this article found in
(1991) Journal of Experimental Social Psychology, 27, 201-202]
Lonner, W. (1989). The introductory psychology text: Beyond Ekman, Whorf and biased IQ tests.
In D. M. Keats, D. Munro, and L. Mann (Eds.), Heterogeneity in cross-cultural psychology (pp,
4-22). Amsterdam: Swets & Zeitlinger.
Parsons, T, & Shils, E. A. (Eds.). (1951). Toward a general theory of action. Cambridge, MA:
Harvard University Press.
Rotter, J. B. (1966). Generalized expectancies for internal versus external control of
reinforcement, Psychological Monographs, 80(609).
Sampson, E. E. (1985). The decentralization of identity: Toward a revised concept of personal
and social order. American Psychologist, 40, 1203-1211.
Schwartz, S. H. (1990). Individualism-collectivism: Critique and proposed refinements. Journal
of Cross-Cultural Psychology, 21, 139-157.
Schwartz, S. H. (1992). The universal content and structure of values: Theoretical advances and
empirical tests in 20 countries. In M. Zanna (Ed.), Advances in experimental social psychology
(Vol. 25, pp. 1-65), New York: Academic Press.
Schwartz, S. H. (1994). Cultural dimensions of values: Towards an understanding of national
differences. In U. Kim, H. C. Triandis, C. Kagitcibasi, S.C. Choi, & G. Yoon (Eds.),
Individualism and collectivism: Theoretical and methodological issues (pp. 85-119), Thousand
Oaks, CA: Sage.
Schwartz, S. H., & Bilsky, W. (1987). Towards 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,
58, 878-891.
Singelis, T., Triandis, H. C., Bhawuk, D. S., & Gelfand, M (1994). Horizontal and vertical
dimensions of individualism and collectivism: A theoretical and measurement refinement.
Manuscript submitted for publication.
Smith, P. B. (1992). Organisational behaviour and national cultures. British Journal of
Management, 3, 39-51.
Smith, P. B., & Bond, M. H. (1993). Social psychology across cultures: Analysis and
perspectives. Needham, MA: Allyn & Bacon.
Smith, P. B., Trompenaars, E, & Dugan, S. (1995). The Rotter locus of control scale in 43
countries: A test of cultural relativity. International Journal of Psychology, 30, 377-400.
Stouffer, S. A., & Toby, J. (1951). Role conflict and personality. American Journal of Sociology,
56, 395-406.
Tonnies, F. (1963). Community and society. New York: Harper, Row. (Original work published
Triandis, H. C. (1990). Cross-cultural studies of individualism and collectivism. In J. J. Berman
(Ed.), Nebraska symposium on motivation, 1989: Vol. 37. Cross-cultural perspectives. Lincoln:
University of Nebraska Press.
Triandis, H. C. (1995). Individualism and collectivism. Boulder, CO: Westview.
Trompenaars, F. (1985). The organisation of meaning and the meaning of organisation: A
comparative study on the conceptions of organisational structure in different cultures.
Unpublished doctoral dissertation, University of Pennsylvania.
Trompenaars, F. (1993). Riding the waves of culture. London: Brealey.
Williams, J., & Best, D. (1990, July). Psychological factors associated with cross-cultural
differences in individualism-collectivism. Paper presented at conference on individualism and
collectivism, Seoul, Korea.
Zurcher, L. A., Medow, A., & Zurcher, S. L. (1965). Value orientation, role conflict and
alienation from work: A cross-cultural study. American Sociological Review, 30, 539-548.
By FONS TROMPENAARS, Centre for International Business Studies
By PETER B. SMITH and SHAUN DUGAN, University of Sussex
Peter B. Smith received his Ph.D. from the University of Cambridge. He is a professor of social
psychology at the University of Sussex and director of the Centre for Research Into CrossCultural Organization and Management (CRICCOM) at Roffey Park Management Institute,
United Kingdom He is coauthor (with M. E Peterson) of Leadership, Organisations and Culture
(1988) and (with M. H. Bond) of Social Psychology Across Cultures (1993). He is editor of this
Shaun Dugan, B.A., is a graduate student and researcher at the University of Sussex. He has
assisted the first author on a number of recent cross-cultural studies
Fons Trompenaars learned about cultural differences firsthand: at home, where he grew up
speaking French and Dutch, and in his career within the multinational Royal Dutch Shell. After
receiving his Ph.D. from the Wharton School of Business in 1982, he established the Centre for
International Business Studies, a consultancy and training organization based in Amsterdam. He
is author of Riding the Waves of Culture (1993).
Copyright of Journal of Cross-Cultural Psychology is the property of Western Washington
University and its content may not be copied or emailed to multiple sites or posted to a listserv
without the copyright holder’s express written permission. However, users may print, download,
or email articles for individual use.
Source: Journal of Cross-Cultural Psychology, Mar96, Vol. 27 Issue 2, p231, 34p, 3 charts, 1
Item Number: 9603194867

YOU MAY ALSO READ ...  MN621 Advanced Network DesignPHYS20009 Individual Assignmentlogical with excellent levelBUS5PB – Principles of Business AnalyticsACC304 – Accounting CapstoneHI5020 Corporate AccountingCorporate AccountingBusiness Research Proposal and Literature Review
Order from Academic Writers Bay
Best Custom Essay Writing Services