Are cultural values the same as the values of individuals? An examination of similarities in personal,social and cultural value structures

Abstract

Value dimensions are used widely to make distinctions between countries and also serve to explain behaviour of individuals within countries. There is a paradox: leading researchers (Hofstede, 1980; Schwartz, 1994) have been adamant that country-level value structures differ from those at individual level. A recent analysis has suggested that dimensional structures at the two levels show substantial overlap. The current multi-level study extends previous research by examining to what extent ratings of the values of others can serve as an intermediate structure to help determine the degree of isomorphism between individual and culture-level structures. The findings indicate that a single value structure across levels is most parsimonious. Implications for international business research are discussed.

Keywords

Culture isomorphism multi-level social values Schwartz Value Survey

Values and value dimensions are used frequently to explain cultural or national differences in organizational behaviour. Hofstede’s (1980) four well-known cultural value dimensions were obtained from factor analysis on country-level item scores in 60 national samples of an international company (IBM). These country-level scores had been derived through aggregation of individual self-report answers on the items by taking item mean scores.

An alternative mapping of values was developed by Schwartz (e.g. 1992). He also identified the underlying structure of personal values from individual self-reports. Asking participants about the importance of various values in their personal life, he suggested that they can be arranged into 10 motivationally distinct value types ordered along two major dimensions. These value types were distinguished consistently in sets of samples drawn from all major regions of the world (Fontaine et al., 2008). Next to this structure at the individual level, Schwartz (1994, 2006) proposed a different value structure when the same data were aggregated to the country level. At this level Schwartz distinguished between seven value regions ordered along three distinct dimensions. This structure was thought to reflect societal or culture-level processes that are different from individual-level processes.

By proposing two different dimensional structures at the individual and the culture level, Schwartz followed a tradition established by Hofstede (1980) who insisted that the use of country-level dimensions at individual level amounted to an ‘ecological fallacy’ (see Robinson, 1950). Thus, both Hofstede and Schwartz have called for a separate interpretation of individual and aggregate level data. Over recent decades there has been a proliferation of organizational research at both individual and country level with little integration of findings (e.g. Kirkman et al., 2006; Tsui et al., 2007). The existence and legitimacy of the differentiation between levels has not been challenged, but it was largely ignored in culture-comparative research. The suggestion by Triandis et al. (1985) of using the terms individualismand collectivismto refer to the country-level dimension and the terms idiocentrismand allocentrismto refer to the corresponding concepts at individual level has rarely been followed up.

Since Hofstede (1980, 2001) did not provide individual-level structures, subsequent researchers could only use country-level scores. These country-level scores have been used at individual, group and country levels (see Kirkman et al., 2006, for a review). However, with Schwartz value dimensions being available at each of the two levels there is a choice and the question is whether the differences between levels do matter. In a recent reanalysis of data from 69 countries using a multi-level analysis technique, Fischer et al. (2010) found that the structures at the two levels were more similar than previously stated by Schwartz (1994, 2006); the level of similarity fell only somewhat short of strict isomorphism (identity of structures across levels, see also Fischer, 2011).

In this article we pursue the question of isomorphism by examining what happens if people are asked to rate the value priorities of others within their society. Tsui et al.’s (2007) first recommendation for future management research was to consider the group property of the culture concept. Here we follow this call and examine ratings of social or cultural descriptive norms (Fischer, 2006) in addition to self-ratings. If the individual and culture-level structures are different as claimed by Hofstede and others, would the resulting social value structures be more similar to the individual-level structure or would they better resemble the structure at the country level? Examining value structures in this way should help to throw light on the validity of the widespread practice of using the same value dimensions across levels and to further clarify the degree of isomorphism between culture level and individual level.

Our analysis further contributes to ongoing discussions in the literature. We apply variants of methods successfully used to examine level of isomorphism in personality (McCrae et al., 2005a) and social axioms (Cheung et al., 2006). Whereas personality was shown to have an isomorphic structure across levels, social axioms showed different structures at individual and country level. The question of whether values show similar or different structures at individual and country level is valid and important in so far as failure to address this question will lead to inappropriate conclusions in comparative research (see Dansereau et al., 1984; Fischer, 2011).

We first present a brief overview of recent value research in cross-cultural management and business research, outline arguments calling for a separation of the value theories at the different levels, and describe the individual-level and culture-level value structure developed by Schwartz (1992, 1994). Then we introduce the concept of social values (Rohan, 2000). Thereafter we put the discussion into a larger framework of multi-level isomorphism and formally state the questions we seek to answer.

Value research in cross-cultural management and business research

Values have emerged as a core construct in cross-cultural research. Values are used both as main effects and moderators in research. At the country level, main effects in the form of cultural distance research have been dominant (Kirkman et al., 2006). Recent findings indicate that greater cultural distance is associated with both pre-acquisition communication and understandability, and increased acquisition performance through increasing learning opportunities between diverse partners in international acquisitions (Reus and Lamont, 2009). Similarly, greater distance in values was found to be related to decentralization in multinational corporations (Williams and van Triest, 2009), entry-mode choices in multinationals (Drogendijk and Slangen, 2006), and supply concentration in offshore outsourcing (Griffith et al., 2009).

At the individual level, value orientations of individuals are typically measured and then related to some attitudes or behaviour. One important innovation for business research has been to unpackage cultural differences using individual-level values (Leung and van de Vijver, 2008; Matsumoto and Yoo, 2006; Poortinga and Van de Vijver, 1987). In a study of Romanian, German and Chinese employees, collectivism could account for country differences in organizational commitment (Felfe et al., 2008).

Studies with value dimensions as moderator variables have been less frequent (Kirkman et al., 2006; Tsui et al., 2007) and are more typical at the individual level. For example, Fischer and Smith (2006) used the Schwartz value survey and found that more individualistic employees (openness to change values) were more strongly affected by the absence of justice in their levels of commitment and citizenship behaviour.

Cross-level studies are needed to understand cultural processes in business contexts (Kirkman et al., 2006; Tsui et al., 2007). Few studies address multiple levels simultaneously. Fischer and Mansell (2009) examined the impact of cultural values on commitment and the correlation between commitment and turnover intentions. They found that values at the country level had significant main and moderating effects. Ralston et al. (2009) in a rare study using values at both individual and country level demonstrated that individualism-collectivism relates to influence tactics at both levels, but independently of the other level. Similarly, Fu et al. (2004) reported that both individual-level beliefs and cultural values independently contributed to influencing strategies of managers at the individual level. In contrast, Smith and Fischer (2008) found that individual and culture-level values interact on response tendencies of managers.

This brief review highlights that values are important for examining and understanding organizational phenomena. However, it is unclear whether the same constructs are measured at individual and country level and what the meaning is at the each level. Values are broad psychological beliefs of individuals about desirable modes of conduct developed through a person’s socialization history. Hence only individuals can have values in the narrow sense; organizations and countries strictly speaking do not have values (Smith et al., 2008).

Separating individual- and culture-level value theories

Hofstede (1980) argued that the ‘collective programming of the mind’ ascribed to culture can be examined through examining the factors emerging from data where individual-level responses are aggregated to the country level. At the individual level, some items were negatively correlated, but emerged on the same factor when aggregated. For example, there was a negative correlation between an item measuring stress and responses to how long people would like to continue working in the company. The more stressed you are, the less likely you are planning to stay. However, both items correlated positively and formed the uncertainty avoidance index at the country level, together with a third item (‘company rules should not be broken’). The explanation for this counter-intuitive finding was that all three items are related to how societies (compared to individuals) deal with stress. Emphasizing rule adherence and longer tenure are mechanisms for reducing uncertainty within society. Similarly, items related to work goals formed a six-factorial structure resembling Maslow’s (1954) need hierarchy at the individual level, whereas at the country level, the two factors later labelled individualism-collectivism and masculinity-femininity emerged. The argument is that, through aggregation, researchers can gain an understanding of how latent cultural processes unfold (e.g. Schwartz, 2011).

Schwartz (1992) developed his two-dimensional model of personal values at the individual level using smallest space analysis and a configurational verification approach. He argued two points. First, across cultures values cluster consistently in a two-dimensional space (i.e. a plane defined by two orthogonal dimensions). Secondly, the clusters are separable into ten distinct motivational types ordered in the space along two dimensions based on their mutual compatibilities and conflicts. In at least 75 percent of the samples 45 (later 46) values emerged in their predicted motivational type (Schwartz, 1992). The naming of the two dimensions depends on their rotation in the plane. Schwartz (1992) labelled the two individual-level dimensions ‘openness to change versus conservation’ and ‘self-enhancement versus self-transcendence’.

Aggregating the same individual-level data on which the individual-level distinctions were based, Schwartz (1994, 2006) argued that the mean country scores reflect latent culture-level processes. Individual differences are cancelled out through aggregation and the means reflect ‘assumptions about the desirables that are built into the institutions of the society and are passed on through intentional and unintentional socialization’ (1994: 92). It is important to note that individual- and country-level structures are based on independent information and are statistically independent (Dansereau et al., 1984). Therefore, it is theoretically plausible to encounter a state of affairs in which individual-level structures cannot be replicated at the country level, even when based on the same data (i.e. individual-level ratings). In emphasizing the distinction between levels, Hofstede (1980) set a precedent that was followed not only by Schwartz but also by others (e.g. Chinese Culture Connection, 1987; Smith et al., 1996; Triandis, 1995). Schwartz (1994) also found a two-dimensional space at the culture level, but here he postulated three (non-orthogonal) dimensions in the space, each dealing with one of three basic societal problems that cultures have to address. By doing so Schwartz emphasized a different structure of values at the country level.

Social values as a third value structure?

What happens if, instead of asking participants about their personal value priorities, we are asking what most other people in one’s country value? In the management literature researchers have argued that questions need to be aligned with the focus of investigation (see Fischer, 2009). For example, Glick (1985) stated that accuracy and construct validity of instruments measuring higher-level processes can be improved if people are asked descriptive questions about the higher level directly rather than questions about personal experiences. Following this logic, asking about the typical member or most members of a society should lead to more accurate descriptions of the society. Available research to date indeed suggests that asking about the society or typical members of a culture leads to different inferences compared to when using self-ratings (Fischer, 2006; House et al., 2004; Terraciano et al., 2005). However, these studies typically have not addressed whether these differences in means are also accompanied by differences in structure. In fact, these studies assume isomorphism. Could it be that these differences in position between personal versus other-focused ratings are also associated with structural differences? Extending the arguments by Glick (1985), one could expect that, if there is a different value structure at the country level, the social value structure would be more similar to Schwartz’s (1994) country-level structure than to the individual-level structure.

In contrast, Rohan (2000) viewed social values as perceptions of value priorities of others that are organized according to what individuals think about the judgement of others in terms of best possible living or functioning. She argued that principles of parsimony require that the perceptions of others’ values be organized in the same way as the individual-level structure. According to this view, the structure of social values should be more similar to the individual-level structure found by Schwartz (1992) than to the country-level structure.

Integration within an isomorphism framework

Questions about the similarity of value structures across levels in the cross-cultural literature focus on isomorphism. In multi-level analysis the degree of isomorphism can be divided in three ranges (Fischer et al., 2010). First, strict isomorphismindicates that structures are identical at individual and country level, or so similar that differences are negligible. The selection of a cut-off point for strict isomorphism is somewhat arbitrary; there is no firm rule. A tradition seems to be growing to follow the literature on structural equivalence of data across cultures (Fischer et al., 2010; Mylonas et al., 2008; van de Vijver and Poortinga, 2002). There an acceptable lower bound for Tucker’s phi, a commonly used statistic for equivalence, ranges from φ (phi) = .80 (Barrett, 1986) to φ ≥ .90 (Van de Vijver and Poortinga, 1994). Second, non-isomorphism indicates that the structures are not overlapping and there is no direct relationship between the levels (Van de Vijver et al., 2008). The upper bound for non-isomorphism can be estimated by comparing the overlap between random configurations with overlap in real data. If the observed levels of isomorphism fall within the confidence interval of similarity found with random structures, non-isomorphism has to be concluded. Finally, partial isomorphism is found when there is a degree of similarity of structures, which is larger than predicted by chance, but does not meet the adopted cut-off point for strict isomorphism.

We can now address our questions in terms of isomorphism. Schwartz (1992) found a stable dimensional structure at the individual level, but subsequently proposed a different structure at the country level (Schwartz, 1994). Although there appeared to be overlap between the two structures, empirically this was not evaluated until recently. Using the large data set collected with the Schwartz Value Scale, Fischer et al. (2010) compared the structural similarity of 55 values across the two levels using both student and teacher samples from a total of 69 countries. They found evidence of partial isomorphism; the similarity indices were much larger than expected by chance, but mostly fell short of the accepted level of .90 for strict isomorphism.

There are four research questions in the present study. The first question is which degree of isomorphism is found with another data set. For the purpose of cross-validation, in addition to the data collected by Schwartz (1992, 2005) we use data examining personal values across student samples in 22 countries reported by Spini (2003).1

Second, the management literature would suggest that the individual-level structure derived from social value ratings will be more similar to the country-level value structure derived from self-ratings (Schwartz, 1994) than to the individual-level value structure derived from self-ratings (Schwartz, 1992). In this scenario, the social values structure would provide a validation of the country-level structure. Thus, the question is whether social value dimensions (at the individual level) are more similar to culture-level self-ratings than to the individual-level self-ratings. For this analysis we use data collected by Fischer (2006).

Third, Rohan (2000) proposes that social values are intrapsychic and, therefore, should be structured like personal values. Consequently, ratings of social values and individual self-ratings should show similar structures. To answer this question we also use the data collected by Fischer (2006).

Fourth, it can be examined to what extent social value structures are similar at individual and, after aggregation, at country level. This entails a direct comparison of the social value structure across the two levels. Since these values were measured with data asking for values of others rather than self, this allows for an alternative test of isomorphism.

We focus on the 44 values (out of the 55 administered) that Schwartz (1992) originally recommended for cross-cultural analyses; these were shared across all three data sets. We will rely on recent innovations in the multi-level literature to examine the similarity of structures (see Van de Vijver et al., 2008). Unfortunately, as discussed before, there are no direct criteria that we can rely on when deciding on degree of isomorphism and no significance tests of similarity are available. Therefore, we also report some secondary analyses on the similarity matrices to judge overall isomorphism.

Method

Participants

For self-ratings of values, we used archival data sets collected by Schwartz (1992, 2005) and Spini (2003). We included samples of students larger than the number of value items in the scale (n > 44). Data from 66 countries (n = 26,024) were available for Schwartz and data from 22 countries for Spini (n = 3467). These data sets include highly diverse geographic, cultural, linguistic and religious groups.

For ratings of social values (what is important to most people in my country), we used data from Fischer (2006). Inclusion criteria were the same as those already mentioned. A data set including 10 countries (n = 2066) was available (Argentina, Brazil, Germany, India, Lebanon, New Zealand, Peru, Taiwan, Saudi Arabia and United States).

Measures

The Schwartz Value Survey consists of 56 or 57 items. As mentioned, in our analysis we used the 44 items that showed stable positions across cultures (Schwartz, 1992). Respondents rated the importance of each value as a guiding principle in their life on a nine-point scale ranging from -1 (opposed to my values), 0 (not important), 3 (important), 6 (very important), to 7 (of supreme importance). Schwartz (1992) offers further explanation of this response format.

The social values scale included the same 44 value items (Fischer, 2006). Responses were coded on an eight-point scale from ‘not at all important’ to ‘of supreme importance’ (the ‘opposed to my values’ option was omitted). Respondents were asked to rate the importance of each value ‘for most people in your country of birth’.

Data analysis

The individual-level structure for each data set was derived from the pooled within-country matrix (Muthén, 1994). This is the averaged correlation matrix across the separate correlation matrices derived for each sample. This matrix was analysed with the Proxscal routine in SPSS 16, using the correlations as similarity indices and specifying interval data. A two-dimensional MDS structure made most sense for all three data sets, in line with earlier findings (Fontaine et al., 2008; Schwartz, 1992).

The country-level structure for each data set was based on the aggregated item means per country. Euclidean distances were derived based on z-transformed mean scores. Using Proxscal in SPSS 16, an interval-level MDS was run.

The co-ordinates of the two MDS configurations cannot be directly compared in a multi-level analysis. This is because rotation, reflection, translation and dilatation of the co-ordinate system can affect the co-ordinates, even when they have no effect on the relative distances between points. Hence, the co-ordinates may appear to differ even when the value structures are in fact much the same. To derive comparable co-ordinates, we subjected all configurations to Generalized Procrustes Analysis (GPA, Borg and Groenen, 1997; Commandeur, 1991). GPA calculates the configurations in such a way that they correspond as closely as possible, without affecting the relative distances between points within each configuration. Following GPA, the spatial co-ordinates of individual value items are directly comparable between the various data sets. After GPA the correlations between the co-ordinates form an index of structural similarity of value structures across data sets and levels.

Results

Table 1 reports the correlations between the item co-ordinates for the various analyses after GPA rotation. Examining the similarity between the three data sets at the individual level, the averaged similarity for dimension 1 was .92 and for dimension 2 it was .64. At the country level, the averaged similarity across all three data sets for dimension 1 was .64 and for dimension 2 it was .71. Key entries in Table 1 are the indices for isomorphism. The average of the two indices for the Schwartz and Spini data was .79 for dimension 1 and .73 for dimension 2. For the social ratings of Fischer these values were .90 and .83 respectively.

Table 1.

Similarity between GPA-rotated MDS configurations

	Dim 1						Dim 2
	1	2	3	4	5	6	7	8	9	10	11
Dim 1
1. SVS-Ind
2. SVS-Cultural	0.86^*
3. Spini-Ind	0.96^*	0.84^*
4. Spini-Cultural	0.72^*	0.76^*	0.72^*
5. Social-Ind	0.91^*	0.77^*	0.89^*	0.70^*
6. Social-Cultural	0.76^*	0.60^*	0.71^*	0.58^*	0.90^*
Dim 2
7. SVS-Ind	0.08	0.18	0.07	0.17	−0.11	−0.13
8. SVS-Cultural	0.11	0.19	0.05	0.23	−0.01	0.05	0.78^*
9. Spini-Ind	0.04	0.15	0.00	0.24	−0.15	−0.15	0.85^*	0.74^*
10. Spini-Cultural	0.08	0.21	0.04	0.11	−0.06	0.00	0.58^*	0.80^*	0.68^*
11. Social-Ind	−0.10	−0.05	−0.04	−0.21	−0.26	−0.19	0.63^*	0.48^*	0.43^*	0.52^*
12. Social-Cultural	−0.04	−0.02	−0.03	−0.14	−0.20	−0.16	0.71^*	0.69^*	0.55^*	0.63^*	0.83^*

Note: ^* p < .01

Examining the average similarity of individual-level social value ratings with individual and culture-level value self-ratings, we found for the individual level .90 (dimension 1), and .53 (dimension 2). With the culture-level dimensions we found .74 (dimension 1), and .50 (dimension 2). The corresponding average similarities of culture level social value scores with individual level ratings were .74 and .63, and with culture-level value ratings .59 and .66, respectively.

For the first individual-level dimension in Schwartz’s SVS we found higher similarities than for the second dimension. This is in line with earlier observations of lower stability for the second dimension (Fontaine et al., 2008). It may also be noted that on average the culture-level associations between the data sets were somewhat lower than the associations across the two levels. This suggests that the reliability of the culture-level indices may be a limiting factor in estimates of isomorphism. Overall, the relevant values for isomorphism were lower than required for claiming identity (> .90), but much higher than would be expected using random configurations (.23 and .10 for dimension 1 and 2, respectively; see Fischer et al., 2010).

To make sense of the fairly large number of associations in the three data sets, we conducted a second order analysis, i.e. a principal component analysis on the co-ordinates with varimax rotation of components. Such an analysis could demonstrate meaningful differentiations between the rotated GPA solutions (e.g. should we or should we not treat the structure of individual or culture-level ratings or self-ratings versus social values as empirically distinct?). A clear two-dimensional structure was found (see Table 2). The first component accounted for 41.46 percent of the variance and the second component for 36.54 percent, in total explaining 78.00 percent of the variance. Eigenvalues for the first five factors were 4.98, 4.38, 0.91, 0.57, 0.38, strongly suggesting a two-dimensional structure. The first component clearly captures dimension 1 of the MDS, whereas the other component captures dimension 2. The loadings were all consistently high and did not suggest any pattern allowing a reliable differentiation between either individual versus culture-level solutions or a separation of personal, social or cultural value structures.

Table 2.

Principal Component Analysis of the Similarity Matrix

	Component 1	Component 2
SVS-Ind Dim 1	.96
Social-Ind Dim 1	.95
Spini-Ind Dim 1	.94
SVS-Culture Dim 1	.89
Social-Culture Dim 1	.84
Spini-Culture Dim 1	.83
SVS-Ind Dim 2		.90
SVS-Culture Dim 2		.89
Social-Culture Dim 2		.86
Spini-Ind Dim 2		.85
Spini-Culture Dim 2		.83
Social-Ind Dim 2		.75

Note: Ind = Individual-level matrix; Culture = Culture-level matrix

Discussion

In this article we sought an answer to four questions. The first question concerned the question of isomorphism between individual-level and country level structure of value dimensions in the SVS values scale (Schwartz, 1992). With data from Spini (2003) we replicated results of Fischer et al. (2010) showing high correlations between levels, but failing to find full support for strict isomorphism. We noted that the reliability of the aggregated country scores is likely to be a limiting factor on the estimates of isomorphism. Therefore, for self-ratings, we found partial isomorphism. The fourth question asked about the social value structure across the two levels. Here the standard for strict isomorphism was met for the first dimension, though not for the second dimension. Therefore, for social value ratings we found strict isomorphism for the first dimension and partial isomorphism for the second dimension. Most importantly, the second and third questions asked about the similarity of the social value structure with individual-level and country-level structures found with the SVS data. There is no clear pattern discernible in Table 1. In light of the answers to questions one and four and the final principal component analysis, we can comfortably claim at least partial isomorphism. However, the high partial isomorphism of the individual-level and country-level structures in the two sets of SVS data makes it difficult to answer the questions whether the social value scores structurally are more similar to the country-level or to the individual-level structures.

Our analysis suggests that a single configuration provides a good representation of personal, social and cultural value structures when using 44 values of the Schwartz Values Survey (SVS). We found no justification to differentiate between two different representations of value structures at the individual and country level. Since the structures at individual and country level are not differentiated it is not surprising that ratings of social values show a structure that is similar to both. Therefore, the results support the speculation put forward by Rohan (2000) that principles of parsimony would lead to a single structure. If individuals were to develop value structures for other entities (groups, organizations, cultures, etc.), this potentially would lead to large numbers of intrapsychic value structures. It is hard to imagine that individuals have the cognitive capacities to develop such different structures for different entities. The finding of a single value space that organizes values makes intuitive sense from an information-processing perspective.

Similarly, aggregated and individual-level value structures showed high correspondence across the three different data sets. Within each data set, the levels of isomorphism were reasonably high and approached the suggested cut-off point for strict isomorphism. For dimension 1, we found on average a similarity of .82 and for dimension 2 (which appears to be less stable) a similarity of .76. This falls short of the strict criterion of .90 for isomorphism. Furthermore, these indices of isomorphism within the data sets were lower than those found for similarity between structures at the individual level, but higher than the associations between culture-level value structures.

Further validating previous findings, average value structures at the individual level are robust. In contrast, culture-level value structures fluctuate more. The similarity of country-level structures was relatively low (averaging .64 and .71 for dimension 1 and 2, respectively). One reason for this could be the number of data points at the culture level; the number of countries varied dramatically (from 10 to 66 countries for which data were available). Fischer et al. (2010) found a systematic effect of country-level sample size on the stability of culture-level structures. The current pattern gives further support for the hypothesis that sampling fluctuations at the country level play a major role in the divergence between individual- and culture-level structures.

All in all, the current findings are clearly at variance with traditional thinking. Hofstede (1980) argued for non-isomorphism between the individual and the cultural level. Schwartz (1994, 2006) saw reason to distinguish three dimensions in his aggregated data set at the culture level as opposed to two dimensions at the individual level. However, neither Hofstede nor Schwartz, nor any of the other authors mentioned before (Chinese Culture Connection, 1987; Smith et al., 1996; Triandis, 1995) actually compared the structures of the two levels in a single multi-level analysis. Obviously, our analysis cannot shed any light on these other studies and a reanalysis of the relevant data would be necessary to resolve any issues about dimensionality (e.g. see a previous study by Bond, 1988).

Equally obvious, we cannot categorically rule out the possibility that differences between values structure across levels could emerge with a more extensive set of value items or a broader range of cultural samples (all the present data came from students). Moreover, it is quite possible that a few items for some specific culture (or cultural region) occupy a deviant position in the structure. However, the findings provide retrospective justification for the widespread practice of not distinguishing conceptually between value dimensions at individual and country level.

Our study also fits into a wider attempt in the psychological literature to understand the dimensionality of constructs at both the individual and aggregated societal level. McCrae et al. (2005a, 2005b) examined the structure of observer ratings of personality at both the individual and societal level and found high similarity. In contrast, Cheung et al. (2006) examined the structure of social axioms at the two levels and found different structures. Leung and Bond (2008) then speculated about meanings of such emerging differences. Given the relative conceptual similarity of personality and values, the emerging similarity in structure at the two levels may not be that surprising in hindsight. At the same time, the findings for more context-specific social axioms show that a simple assumption of isomorphism is not warranted and needs to be tested empirically.

Implications for international management and business research

The issue of shifts in meaning of variables across levels has been drawing increasing attention in organizational research, where individual employees are nested in teams, with the teams nested in divisions and/or subsidiaries, and ultimately countries (e.g. Bliese et al., 2007; Fischer, 2009).

For values, an important domain in cross-cultural research, it appears that a single value structure is most parsimonious. We have found no evidence to suggest that individual and cultural value dimensions should be treated as distinct. Ratings of social values have high overlap with both. This suggests that future research could use individual-level value types or clusters (e.g. Schwartz’s, 1992, 10 value types or the two higher-order value dimensions) at both individual and aggregate levels. This greatly simplifies research and allows for comparison of effects across levels. Values can be directly measured at the individual level and comparison across individuals and cultures is possible. In a sense, this is not surprising. After all, individual persons enact opinions, decisions and actions (and not cultural entities).

Such a treatment simplifies business research without leading to a substantive loss of information. For example, current work on entry mode choice, international acquisitions and outsourcing typically rely on cultural distance of countries. Nevertheless, the active agents in these decisions are managers. It is well-known that individual differences in values are larger than cultural differences. For example, Poortinga and van Hemert (2001) reported that between 6 and 21 percent of the variability in value scores across 34 countries was due to country. More recently, Fischer and Schwartz (2011) found that across three different value studies including representative samples from 62 countries, country differences did not exceed 30 percent for any value and that on average country explained less than 12 percent of the variability. Gerhart (2009) reviewed other culture-level studies and found only small country differences in a number of widely cited cross-national studies. Values are psychological constructs and individual differences account for the major part of the variability in responses. To understand managerial behaviour, researchers might be advised to focus more on this larger share of variability than on the smaller amount that is due to cultural differences.

The possibility of using identical dimensions at both levels should also facilitate cross-level research in organizations. This has been repeatedly identified as one of the major prospects for future organizations research (e.g. Gerhart, 2009; Kirkman et al., 2006; Tsui et al., 2007; van de Vijver and Fischer, 2009). With the same value dimensions across levels, the independence and interdependence of individual and cultural processes on business processes can be more directly studied (see Smith and Fischer, 2008, for an example).

Hofstede (1980), whose four value dimensions included only a limited number of items, did not provide individual-level scores or an individual-level value structure. His data were collected more than four decades ago. It may be time to move on and use other value dimensions that better satisfy psychometric demands. Today, more recent and better validated value instruments are available. Our study demonstrates that the Schwartz value dimensions are convergent between levels. To the extent that this is the case, individual-level value dimensions may be assigned to countries and used in country-level analysis. However, for the time being we recommend that isomorphism should be demonstrated rather than assumed, definitely when other instruments than the SVS are used.

Conclusions

We have extended previous research on the similarity of individual- and culture-level values structures by comparing both with social-level data. Across different data sets, we found substantive similarity of value structures that do not justify treating the two levels of individual and country as separate structures. It appears the same dimensions underlie value structures at each of the two levels.

Footnotes

Funding

This study was partially supported by a Marsden Fast Start grant from the Royal Society of New Zealand – Te Putea Rangahau a Marsden and grants by the School of Psychology at Victoria University Wellington, New Zealand.

Acknowledgements

SVS data collected by Shalom Schwartz are distributed by the Israel Social Science Data Centre at Hebrew University. We would like to thank to Shalom Schwartz and Dario Spini for allowing us to use their data, as well as invaluable discussions and comments. We would also like to express our thanks to the editor and two anonymous reviewers.

Notes

References

Barrett

(1986) Factor comparison: an examination of three methods. Personality and Individual Differences 7: 327–40.

Bliese

Chan

Ployhart

(2007) Multilevel models. Organizational Research Methods 10: 551–63.

Bond

(1988) Finding universal dimensions of individual variation in multi-cultural studies of value. Journal of Personality and Social Psychology 55: 1009–15.

Borg

Groenen

(1997) Modern Multidimensional Scaling: Theory and Applications. New York: Springer.

Cheung

MWL

Leung

(2006) Evaluating multilevel models in cross-cultural research: an illustration with Social axioms. Journal of Cross-Cultural Psychology 37: 522–41.

Chinese Culture Connection (1987) Chinese values and the search for culture-free dimensions of culture. Journal of Cross-Cultural Psychology 18: 143–64.

Commandeur

JJF

(1991) Matching Configurations. Leiden: DSWO Press.

Dansereau

Alutto

Yammarino

(1984) Theory-Testing in Organizational Behavior: The ‘Variant’ Approach. Englewood Cliffs, NJ: Prentice Hall.

Drogendijk

Slangen

. (2006) Hofstede, Schwartz or managerial perceptions? The effects of different cultural distance measures on establishment mode choice by multinational enterprises. International Business Review 15: 361–89.

10.

Felfe

Yan

Six

(2008) The impact of individual collectivism on commitment and its influence on organizational citizenship behaviour and turnover in three countries. International Journal of Cross-Cultural Management 8: 211–37.

11.

Fischer

(2006) Congruence and functions of personal and cultural values: do my values reflect my culture’s values? Personality and Social Psychology Bulletin 32: 1419–31.

12.

Fischer

(2009) Where is culture in cross-cultural research? An outline of a multi-level research process for measuring culture as a shared meaning system. International Journal of Cross-Cultural Management 9: 25–49.

13.

Fischer

(2011) Value isomorphism in the European Social Survey: exploration of meaning shifts in values across levels. Journal of Cross-Cultural Psychology. Published online before print July 10, 2011, doi:10.1177/0022022111413276

14.

Fischer

Mansell

(2009) Commitment across cultures: a meta-analytical approach. Journal of International Business Studies 40: 1339–58.

15.

Fischer

Smith

(2006) Justice criteria in the work place: the importance of values for perceptions of justice. Applied Psychology: International Review 55: 541–62.

16.

Fischer

Schwartz

(2011) Whence differences in value priorities? Individual, cultural, or artifactual sources. Journal of Cross-Cultural Psychology. Electronically pre-published on 9 Sept. 2010. doi:10.1177/0022022110381429.

17.

Fischer

Vauclair

Fontaine

Schwartz

(2010) Are individual- and culture-level value structures different? Testing Hofstede’s legacy with the Schwartz data. Journal of Cross-Cultural Psychology 41: 135–51.

18.

Fontaine

JRJ

Poortinga

Delbeke

Schwartz

(2008) Structural equivalence of the values domain across cultures: distinguishing sampling fluctuations from meaningful variation. Journal of Cross-Cultural Psychology 39: 345–65.

19.

Kennedy

Tata

Yukl

Bond

et al. (2004) The impact of societal cultural values and individual social beliefs on the perceived effectiveness of managerial influence strategies: a meso approach. Journal of International Business Studies 35: 284–305.

20.

Gerhart

(2009) Does national culture constrain organization culture and human resource strategy? The role of individual mechanisms and implications for employee selection. Research in Personnel and Human Resources Management 28: 1–48.

21.

Glick

(1985) Conceptualising and measuring organizational and psychological climate: pitfalls in multilevel research. Academy of Management Review 10: 601–16.

22.

Griffith

Harmancioglu

Droge

(2009) Governance decisions for the offshore outsourcing of new product development in technology intensive markets. Journal of World Business 44: 217–24.

23.

Hofstede

(1980) Culture’s Consequences: International Differences in Work-Related Values. Beverly Hills, CA: SAGE.

24.

Hofstede

(2000) Culture’s Consequences: Comparing Values, Behaviors, Institutions and Organizationd across Nations, 2nd edn. Thousand Oaks, CA: SAGE.

25.

House

Hanges

Javidan

Dorfman

Gupta

(2004) Culture, Leadership and Organizations: The GLOBE Study of 62 Societies. Thousand Oaks, CA: SAGE.

26.

Kirkman

Lowe

Gibson

(2006) A quarter century of culture’s consequences: a review of empirical research incorporating Hofstede’s cultural values framework. Journal of International Business Studies 37: 285–320.

27.

Leung

Bond

(2008) Psycho-logic and eco-logic: insights from social axiom dimensions. In: van de Vijver

van Hemert

Poortinga

(eds) Individuals and Cultures in Multilevel Analysis. Mahwah, NJ: Lawrence Erlbaum Associates, 199–221.

28.

Leung

van de Vijver

FJR

(2008) Strategies for strengthening causal inferences in cross cultural research: the consilience approach. International Journal of Cross-Cultural Management 8: 145–69.

29.

McCrae

Terracciano

, and members of the Personality Profiles of Cultures Project (2005a) Universal features of personality traits from the observer’s perspective: data from 50 cultures. Journal of Personality and Social Psychology 88: 547–61.

30.

McCrae

Terracciano

, and members of the Personality Profiles of Cultures Project (2005b) Personality profiles of cultures: Aggregate personality traits. Journal of Personality and Social Psychology 89: 407–25.

31.

Maslow

(1954) Motivation and Personality. New York: Harper & Row.

32.

Matsumoto

Yoo

(2006) Toward a new generation of cross-cultural research. Perspectives on Psychological Science, 1: 234–50.

33.

Muthén

(1994) Multilevel covariance structure analysis. Sociological Methods and Research 22: 376–98.

34.

Mylonas

Pavlopoulos

Georgas

(2008) Multilevel structure analysis for family-related constructs. In: Van de Vijver

FJR

Van Hemert

Poortinga

(eds) Individuals and Cultures in Multilevel Analysis. Mahwah, NJ: Erlbaum, 345–77.

35.

Poortinga

Van de Vijver

FJR

(1987) Explaining cross-cultural differences: bias analysis and beyond. Journal of Cross-Cultural Psychology 18: 259–82.

36.

Poortinga

van Hemert

(2001) Personality and culture: demarcating between the common and the unique. Journal of Personality 69: 1033–60.

37.

Ralston

Egri

de la Garza Carranza

Ramburuth

Terpstra-Tong

Pekerti

AA, et al.

(2009) Ethical preferences for influencing superiors: a 41-society study. Journal of International Business Studies 40: 1022–45.

38.

Reus

Lamont

(2009) The double-edged sword of cultural distance in international acquisitions. Journal of International Business Studies 40: 1298–1316.

39.

Robinson

(1950) Ecological correlations and the behavior of individuals. American Sociological Review 15: 351–7.

40.

Rohan

(2000) A rose by any name? The values construct. Personality and Social Psychology Review 4: 255–77.

41.

Schwartz

(1992) Universals in the content and structure of values: theoretical advances and empirical tests in 20 countries. Advances in Experimental Social Psychology 25: 1–65.

42.

Schwartz

(1994) Beyond individualism-collectivism: new cultural dimensions of values. In: Kim

Triandis

Kagitcibasi

Choi

Yoon

(eds) Individualism and Collectivism: Theory, Method and Applications. Thousand Oaks, CA: SAGE, 85–119.

43.

Schwartz

(2005) Basic human values: their content and structure across countries. In: Tamayo

Porto

(eds) Valores e comportamento nas organizações (Values and behaviour in organizations). Petrópolis, Brazil: Vozes, 21–55.

44.

Schwartz

(2006) A theory of cultural value orientations: explication and applications. Comparative Sociology 5: 136–82.

45.

Schwartz

(2011) Values: cultural and individual. In Breugelmans

Chasiotis

van de Vijver

FJR

(eds) Fundamental Questions in Cross-Cultural Psychology. Cambridge: Cambridge University Press, 463–93.

46.

Smith

Fischer

(2008) Acquiesence, extreme response bias and levels of cross-cultural analysis. In: van de Vijver

FJR

van Hemert

Poortinga

(eds) Individuals and Cultures in Multi-Level Analysis. Mahwah, NJ: Lawrence Erlbaum Associates, 283–311.

47.

Smith

Peterson

Thomas

(eds) (2008) Handbook of Cross-Cultural Management. Los Angeles: SAGE.

48.

Smith

Dugan

Trompenaars

(1996) National culture and the values of organisational employees: a dimensional; analysis across 43 nations. Journal of Cross-Cultural Psychology 27: 231–64.

49.

Spini

(2003) Measurement equivalence of 10 value types from the Schwartz value survey across 21 countries. Journal of Cross-Cultural Psychology 34: 3–23.

50.

Terracciano

Abdel-Khalak

Ádám

N, et al.

(2005) National character does not reflect mean personality trait levels in 49 cultures. Science 310: 96–100.

51.

Triandis

(1995) Individualism and Collectivism. Boulder, CO: Westview Press.

52.

Triandis

Leung

Villareal

Clark

(1985) Allocentric versus idiocentric tendencies: convergent and discriminant validation. Journal of Research in Personality 19: 395–415.

53.

Tsui

Nifadkar

(2007) Cross-national, cross-cultural organizational behavior research: advance, gaps and recommendations. Journal of Management 33: 426–78.

54.

Van de Vijver

Fischer

(2009) Improving methodological robustness in cross-cultural organizational research. In: Bhagat

Steers

(eds) Handbook of Culture, Organizations, and Work. Cambridge: Cambridge University Press, 491–517.

55.

Van de Vijver

FJR

Poortinga

(1994) Methodological issues in cross-cultural studies on parental rearing behavior and psychopathology. In: Perris

Arrindell

Eisemann

(eds) Parental Rearing and Psychopathology (pp. 173–97). Chichester: Wiley.

56.

Van de Vijver

FJR

Poortinga

(2002) Structural equivalence in multilevel research. Journal of Cross-Cultural Psychology 33: 141–56.

57.

Van de Vijver

FJR

Van Hemert

Poortinga

(2008) Conceptual issues in multilevel models. In: Van de Vijver

FJR

Van Hemert

Poortinga

(eds) Individuals and Cultures in Multilevel Analysis. Mahwah, NJ: Erlbaum, 3–26.

58.

Williams

van Triest

(2009) The impact of corporate and national cultures on decentralization in multinational corporations. International Business Review 18: 156–67.