Culture as a Configuration of Values: An Archetypal Perspective

Experimental Economics and Culture

ISBN: 978-1-78743-820-0, eISBN: 978-1-78743-819-4

ISSN: 0193-2306

Publication date: 14 December 2018


The aim of this chapter is to: (1) model culture as a configuration of multiple values, (2) identify different culture archetypes across the globe, and (3) empirically demonstrate heterogeneity in culture archetypes within and across 52 countries. We use Schwartz values from the World Values Survey (WVS) and the archetypal analysis (AA) method to identify diverse culture archetypes within and across countries. We find significant heterogeneity in culture values archetypes within countries and homogeneity across countries, calling into question the assumption of uniform national culture values in economics and other fields. We show how the heterogeneity in culture values across the globe can be represented with a small number of distinctive archetypes. The study could be extended to include a larger set of countries, and/or cover a broader range of theoretically grounded values than those available in the Schwartz values model in the WVS. Research and practice often assume cultural homogeneity within nations and cultural diversity across nations. Our finding of different culture archetypes within countries and similar archetypes across countries demonstrates the important role of culture sharing and exchange as a source of reducing cultural conflicts between nations and enhancing creativity and innovation through interaction and integration in novel ways. We examine culture as a configuration of multiple values, and use a novel AA method to capture heterogeneity in culture values within and across countries.



Midgley, D., Venaik, S. and Christopoulos, D. (2018), "Culture as a Configuration of Values: An Archetypal Perspective", Experimental Economics and Culture (Research in Experimental Economics, Vol. 20), Emerald Publishing Limited, pp. 63-88.

Download as .RIS



Emerald Publishing Limited

Copyright © 2019 Emerald Publishing Limited

1. Introduction

Until recent times, the focus of economists has largely been on individual rationality as an explanation for human behavior. However, it is now generally acknowledged that such explanations alone may not fully capture people’s decision-making processes and outcomes, resulting in growing interest in examining the role of culture and institutions in understanding societies (North, 2005). The culture-related–economics literature encompasses multiple and widely varying perspectives, ranging from those where culture is subsumed within economics under the rubric of “cultural materialism” (e.g., Harris, 1979), to those where economics itself is regarded as part of culture (e.g., Geertz, 1983). There is also an intermediate viewpoint (and which we follow in this chapter) that regards culture as providing an useful additional and sometimes alternative plausible explanation for individual and societal beliefs, behaviors, and outcomes (e.g., Inglehart, 2006; North, 2005).

There is also diversity in how culture is viewed as an emic (unique) meaning system by anthropologists, versus an etic (universal) system by economists (Beugelsdijk & Maseland, 2011). Within the etic perspective that is common in economics and other disciplines, the concept of culture is often represented simply as values dimensions, and diversity in culture across countries or regions is captured as differences in scores on these values dimensions. For example, Hofstede (2001) introduced five dimensions of national culture to differentiate nations and societies. Similarly, the GLOBE study (House, Hanges, Javidan, Dorfman, & Vipin, 2004) examined culture differences among countries based on their 9 dimensions of culture values and practices, and Schwartz (1992) represented culture values on 10 dimensions. According to Beugelsdijk and Maseland (2011, p. 142), “Schwartz and GLOBE are not used as extensively as Hofstede in economics, but are considered by some scholars to be superior to Hofstede.” The common theme across these culture models is that each country is represented with the average national score on each of the different national culture dimensions in isolation. These dimension scores are then compared between countries to identify national cultural similarities or differences. For example, Guiso, Sapienza, and Zingales (2006) review shows that culture-based approaches to understand economic phenomena use different aspects of culture such as religion, trust, etc. individually to explain national economic outcomes.

Notwithstanding the considerable progress in examining the role of culture in economics, there are several challenges that need to be overcome to enhance our description and understanding of culture and its role in human economic behavior. First, we need a more “theory driven approach to develop quantitative scales of culture” (Beugelsdijk & Maseland, 2011, p. 143). Second, the popular unidimensional approach to the application of culture dimensions wherein each culture dimension is examined in isolation (even in multidimensional culture models such as those of Hofstede and GLOBE), is at odds with the more realistic perspective of culture as a configuration of values where individuals hold multiple values as a bundle rather than separately (Tsui, Nifadkar, & Ou, 2007). Third, notwithstanding the common, albeit erroneous, assumption of culture as homogeneous within countries and heterogeneous across countries (Brewer & Venaik, 2014; Taras, Steel, & Kirkman, 2016), “[…] there is no study in which this question (of intranational cultural heterogeneity) is explicitly addressed, most probably because of the empirical difficulties associated with measuring value diversity in countries” (Beugelsdijk & Maseland, 2011, p. 142). Fourth, there is a longstanding view in the culture literature, as noted by Kroeber and Kluckhohn (1952, p. 162) that “[…]the main unresolved problems of culture will never be resolved by statistical techniques precisely because cultural behavior is patterned and never randomly distributed. Mathematical help may come from matrix algebra or some form of topological mathematics” (italics added).

In this chapter, we adopt the configurational perspective on culture values, and present our theory of culture archetypes as a useful way to represent such configurations in a comprehensive and parsimonious manner. In doing so, we overcome some of the challenges in understanding culture and culture heterogeneity across and within nations. We define a culture archetype as “a configuration of the fundamental values shared by a group of people and represented by a hypothetical individual who perfectly embodies these values” (Venaik & Midgley, 2015, p. 1055) (italics original). We believe culture archetypes provide a more comprehensive understanding of culture than the unidimensional perspective commonly used in economics and other fields. Furthermore, inherent in the archetypal approach is the ability to delineate culture groups, that is, groups of individuals who share similar values configurations, allowing us to describe culture heterogeneity in a parsimonious and easily understandable manner.

Here, we also outline the archetypal analysis (AA) approach that we employ to identify these culture archetypes from empirical data, providing an illustration of both culture archetypes and their practical use. For this, we use the Schwartz values data from the World Values Survey (WVS) Wave Five, as the Schwartz values model, which derives from the work of Rokeach (1973), is one of the most theoretically grounded models of culture. AA draws more on ideas from topology and matrix algebra than from the typical statistical methods in the literature. AA thus enables us to identify both intranational cultural heterogeneity as well as transnational cultural homogeneity, neither of which is empirically explored in the current culture literature. Our AA also obviates the need to create culture groups a priori based on presumed cultural differences across national, regional, or religious boundaries. Instead, archetypes, and the groups of individuals sharing them, are identified empirically from values data and presumed cultural differences tested subsequently. Finally, we extend our previous work (Venaik & Midgley, 2015) from 4 countries to 52, a broader sample of nations that allows us to introduce the concept of global archetypes. That is, a small set of archetypes that represent the diversity of culture values across and within the nations of the world.

The chapter is organized as follows. First, we provide a brief overview of the concept of culture. Next, we introduce our theory of culture archetypes as diverse configuration of values. We then discuss the AA method that we use to identify archetypes, and illustrate the AA approach with the ten Schwartz values in the WVS Wave Five data. We conclude the chapter with a discussion of our AA results and the usefulness of AA in parsimoniously representing culture heterogeneity with a small set of archetypes.

2. What is Culture?

The concept of culture is of interest to scholars across a wide range of disciplines including anthropology (e.g., Tylor, 1871), social psychology (e.g., Fischer & Schwartz, 2011), and economics (Guiso, Sapienza, & Zingales, 2016). “Culture” has many definitions that encompass both subjective values, norms and beliefs, and objective artifacts (Kroeber & Kluckhohn, 1952). Notwithstanding the differences, the two main characteristics underlying most culture definitions are: (1) culture includes values, beliefs and meaning systems (Rokeach, 1973, p. 5), (2) which are shared by groups of people (Hall, 1977, p. 16). It is important to note that our culture definition follows the social science tradition of focussing on the “sharing” aspect of values. On the other hand, whereas economics recognizes the sharing aspect of culture (e.g., Henrich, Heine, & Norenzayan, 2010; Johnson & Mislin, 2011), it predominantly focusses on the intergenerational transmission aspect of values, for example, Guiso et al. (2006, 23) “define culture as those customary beliefs and values that ethnic, religious, and social groups transmit fairly unchanged from generation to generation” (italics original).

According to Herskovits (1948, p. 625), “[…]culture is essentially a construct that describes the total body of belief, behavior, knowledge, sanctions, values, and goals that mark the way of life of any people.” (Italics added.) House et al. (2004) define culture as shared understandings manifested in societal values. The term “culture”, therefore, covers two key aspects. One, it signifies one or more types of values, for example, the value of “harmony”. Two, it refers to a group of individuals who share common values, for example, people who value harmony. Given these two distinct features of culture, we cannot demarcate the boundary of both a priori. We must either (1) first define the culture group and then determine their culture values, or (2) first identify individuals with similar culture values and then classify them as a culture group.

In the first case, if the aim of the researcher is to study people who are members of a homogenous group, for example, a small tribe living in isolation, then the values these people share is an empirical question. We can discover their values by studying the tribe during their day-to-day lives. In the second case, if we are interested in a specific set of values, for example, those for tradition – that we define to have distinct conceptual meaning – then, we must empirically determine the boundary of the group of people who share these traditional values. However, in both cases our objective is the same, namely to identify a distinct group of people who share common values, where sharing implies relative consensus and homogeneity of their beliefs. If we accept this definition of culture as the shared values of a group, the approaches taken by the proponents of national culture create two problems. First, imposing national geographic boundaries on cultural space may or may not align with shared culture groupings within and across countries. Countries such as India or the USA may be built from many separate cultural groupings; each holding their own set of shared values (Sen, 2005; Woodward, 2012). Second, these approaches use averaging to arrive at a national score on a cultural dimension. It is, however, difficult to reconcile averaging of values with sharing of values unless there is a high degree of homogeneity of belief among all people within the nation (Maruyama, 1999). Averaging may also hide underlying differences in values among the individuals who are averaged (Kamakura & Mazzon, 1991; Venaik & Brewer, 2013).

Another issue with the national culture dimensions perspective is that culture is examined one dimension at a time rather than as a bundle of values that are manifested as a pattern. In their comprehensive review of the culture concept, Kroeber and Kluckhohn (1952) provide several definitions based on the patterning and organization of culture. For example, Kroeber and Kluckhohn (1952, p. 34) state that “The essential part of culture is to be found in the patterns embodied in the social traditions of the group, that is, in knowledge, ideas, beliefs, values, standards, and sentiments prevalent in the group.” And, Rokeach (1973, p. 5) distinguished the concepts of “value” and “value system” as follows: “A value is an enduring belief that a specific mode of conduct or end-state of existence is personally or socially preferable to an opposite or converse mode of conduct or end-state of existence. A value system is an enduring organization of beliefs concerning preferable modes of conduct or end-state of existence along a continuum of relative importance.” In both these conceptualizations of values, individual differences are reflected in differences in the pattern or configuration of the value systems rather than on single values. In the same vein, Triandis (1994, p. 2) states that “The elements of subjective culture are organized into patterns” and Hofstede (2001, p. 1) views “culture patterns are rooted in value systems.” Thus, most definitions refer to culture as “the integrated, complex set of interrelated and potentially interactive patterns characteristic of a group of people” (Lytle, Brett, Barsness, Tinsley, & Janssens, 1995, p. 170). Overall, there is strong support for viewing culture as a distinct pattern, configuration, or gestalt of values shared by a group of people. To standardize terminology, we use the word “configuration”, where a configuration is a group’s values expressed across multiple dimensions. In this chapter, we identify culture archetypes within and across countries through common configurations of cultural values shared by groups of people.

In sum, it is the sharing of values configurations among individuals that is the key characteristic of culture, and not the averaging of values one dimension at a time according to some predetermined boundary such as age, class, or a nation. National averages have certainly advanced our understanding of culture but, along with many other scholars, we believe there is significant heterogeneity in values configurations within nations and homogeneity across nations. Hence, the more interesting and important research question in culture is how we best conceptualize, measure, and analyze the richness and diversity of shared values within and across countries.

3. Culture Archetypes

The origin of the idea of an “archetype” commences with the work of Plato, in particular his theory of forms that distinguishes between abstract “universals” and concrete “particulars”. Archetypes themselves are a later development that builds on Plato’s ideas, often by defining classes of universals for specific fields of application, but a development that retains his idea of a pure form that embodies the fundamental characteristics of an object. Although the word “archetype” derives from Ancient Greek word whose roots, arkhe and tupos roughly translate as “starting or beginning model,” the modern definition of archetype is “a perfect example” (Merriam-Webster Dictionary). According to North (2005, p. 32), “The capacity to generalize from the particular to the general, and to use analogy is a part of … representational redescription and underlies not only creative thinking but belief systems generally.”

In this chapter, we focus on archetypes of culture values that are shared by groups of individuals within and across nations. In addition, we view these shared values through a configurational lens because individuals may hold many distinct and different types of values (e.g., Tsui et al., 2007). Using the Platonic origins of the word “archetype” and its etymology, we define a culture archetype as: a configuration of the fundamental values shared by a group of people and represented by a hypothetical individual who perfectly embodies these values (Venaik & Midgley, 2015, p. 1055). Since we do not observe it directly, a culture archetype represents a hypothetical individual. The configuration of values of each member of the group will then have varying degrees of resemblance to an archetypal configuration, analogous to the way cases of a shape such as a square relate to the universal idea of a square. To illustrate, Fig. 1 is an example of a values configuration for a hypothetical archetype (the bold line). The archetype is based on the 10 values in the Schwartz (1992) universal values model. This hypothetical archetype shows a cultural group that is weak on the self-oriented values of self-enhancement/openness to change (on the left) and strong on the society-oriented values of self-transcendence/conservation (on the right).

Fig. 1. 
An Example of a Culture Archetype.

Fig. 1.

An Example of a Culture Archetype.

The configurations of the archetypes and the exact sizes of the various subgroups are naturally dependent on the characteristics of the study population. How we define boundaries between subgroups also depends on our methodology. This raises the question of how we identify archetypes in the sorts of culture data we have available. To do so requires a different methodological approach to those typically used by researchers in economics, because we need to identify both the archetypes and their associated subgroups of individuals from the configurations of values in our data. Existing methods, such as those that classify or describe individuals by averages or group them a priori based on geographic factors, are inadequate for this task. Fortunately, a method for identifying archetypes in multivariate data exists and is increasingly being used in a range of physical and social sciences. We describe this method, AA (Cutler & Breiman, 1994), in the next section. There, we also show how this method is consistent with, and builds on, our conceptual discussion above.

To summarize, the following are our key conclusions from the theoretical discussion of culture archetypes. First, archetypes are clearer and sharper than group averages since they differentiate more between subsets of values in the configuration. Second, in diverse populations, multiple archetypes provide a parsimonious description of the population through a small number of perfect examples. Third, to identify culture archetypes in empirical data, we require a new and different methodology – AA. Overall, the construct of a culture archetype and the AA used to identify these from empirical data provide a novel approach to studying cultural heterogeneity.

4. How to Identify Archetypes From Data? the Archetypal Analysis Method

4.1 Archetypal Analysis

Cutler and Breiman (1994) introduced AA as a mathematical technique for identifying archetypes in multivariate data. AA builds on the theory we introduced earlier by providing a formal and general algorithm for identifying a small number of distinct archetypes from our culture values data. AA also provides a means for us to determine the similarity of each individual’s values to each of our archetypes, determine which of these archetypes best represents each individual, and through this classification of individuals identify groups with similar values. Thus, AA is the appropriate method to apply to our work here, since the end result is a small number of groups with shared values and a representation of each group’s value configuration; that is, the archetype of that group.

To illustrate how this algorithm works, we can imagine our data as a cloud in a multidimensional space described by the variables of interest. Each case in the data would then occupy a point in this cloud defined by its coordinates, that is, the magnitudes of the variables observed for that case. Archetypes are then defined as the coordinates of a small number of points on the frontier of this cloud – that is, the convex hull – and each case is described as a simple weighted composite of these archetypes. Using iterative optimization techniques, AA chooses both the coordinates of the archetypes and the individual case weights to best fit the totality of the data. Largely because of this topological approach, and in addition to the theoretical advantages listed above, AA provides two practical advantages over other techniques when our purpose is to describe value configurations in a clear and parsimonious manner. These are: imposing no strong “model” on the data (Li, Wang, Louviere, & Carson, 2003), and being robust to noise in the data (Chan, Mitchell, & Cram, 2003).

4.2. The AA Algorithm

The foundation paper for AA is Cutler and Breiman (1994). As explained by Eugster and Leisch (2009), the essential problem AA seeks to solve is as follows. Assume you have a matrix X of multivariate data with n observations and m variables. Further, assume you know the number of archetypes you wish to generate, k. Then, the algorithm seeks to find a matrix Z of k m-dimensional archetypes that satisfy two important conditions.

  • (1)

    The data are best approximated by the convex combinations of the archetypes that minimize the residual sum of squares; namely RSS = ||XαZ T ||2 where the coefficients α are >0 and sum to 1 across the k archetypes.

  • (2)

    The archetypes themselves are convex combinations of the data points; namely Z = X T β and the coefficients β are also > 0 and sum to 1 across the n rows.

The way these equations are solved to satisfy the two conditions is by an optimization procedure that alternates between finding the best α’s for a given set of archetypes and finding the best archetypes for a given set of α’s (Cutler & Breiman, 1994, p. 345). At each step, the algorithm needs to solve several convex least squares problems and linear equations. Cutler and Breiman (1994) show that the AA algorithm always converges to a minimum but not always the global minimum. Thus, the researcher has to try more than one starting point for the algorithm in order to be confident they have found the best fit.

In our first work (Venaik & Midgley, 2015), we used the software developed by Eugster and Leisch (2009), which is based on the original algorithm of Cutler and Breiman. However, since 2009 several other AA algorithms have appeared, one of which by Morup and Hansen (2012) we believe better suits the nature of the WVS data. Morup and Hansen replace the initial search for points on the convex hull with a “furthest sum” technique that considers groups of points. This, we believe is more appropriate for typical social science data measured by questionnaire scales with a small number of response categories, where many individuals may be assigned identical sets of numeric values. For fitting archetypes to data, Morup and Hansen also use a projection gradient technique which has, in our experience, proved more robust than the alternating least squares or matrix decomposition techniques used in earlier AA algorithms. Morup and Hansen’s algorithm is implemented in Python and available on gitHub (

4.3. Determining the Number of Archetypes

Given we can generate a solution for a given number of archetypes, k, how do we choose between different ks? In fact, this is done in the same way as for several other statistical methods, namely by examining how the residual sum of squares decreases as the number k increases and determining where this improvement shows an elbow or “knee point”. The choice of this knee point can be made subjectively by the researcher from graphical aids such as “scree plots.” However, objective methods for detecting knee points in noisy data do exist (Christopoulos, 2014) and we employ these here, thus avoiding any subjectivity in the choice of k. We use the extremum distance estimator from the R package “inflection” available on CRAN ( which provides the best estimate of where the decrease in the residual sum of squares reaches a plateau.

4.4. Associating Individual Cases with Archetypes

We can associate individual cases with the k archetypes either through the weights resulting from the AA or through the Euclidean distances between the archetype configurations and the Schwartz values for each individual case. Here we prefer the latter, because it better reflects the concept of shared values and it is also easier to interpret and implement, the latter especially being the case when multiple different AA solutions are to be compared. We use these Euclidean distances to assign each case to its nearest archetype, thus forming k subgroups within the overall sample.

4.5. Comparison of AA with Other Methods

There are several similarities between AA and other methods such as cluster or latent class analysis. All three methods use iterative optimization techniques to fit their models to data and all three produce multiple profiles across the variables of interest. All three require running from multiple start points to ensure best fit and all three require comparisons of fit at different numbers of archetypes, classes, and clusters to choose the overall solution. But there are also two striking differences between AA and the other methods. The first of these relates to objectives, the second to philosophy concerning the information content of a sample case. On objectives, most forms of cluster and latent class analysis seek to divide the sample into distinct subgroups (“clusters”, “classes”). In contrast, AA seeks to describe each sample case as a weighted combination of a small number of archetypes, with each archetype being a “perfect example” of a subgroup of configurations within the overall data. This leads directly to the second, philosophical, point. AA effectively treats cases on the convex hull of the multidimensional data cloud as more informative than cases in the middle of this cloud. It is from these cases that the archetypes are derived and to which (as weighted combinations of these archetypes) all cases in the data are connected. Cutler & Breiman also formally prove that if there are N data points that define the convex hull of the data, there are k (<N) archetypes that minimize the residual sum of squares between all the cases and the original data (Cutler & Breiman, 1994, p. 344). In contrast, cluster and latent class analysis treat each case equally regardless of its position in the cloud. The bottomline is that AA can be thought of as taking into consideration the topology of the whole sample; the other methods consider a more restricted local topology around their clusters or classes. In that sense, these methods may not be fully comparable.

4.6. Advantages and Limitations of AA in Comparison with Other Methods

Note we are not arguing here that one method is superior to any other; this clearly depends on the researcher’s objectives and data. Equally all three methods, cluster analysis, latent class analysis, and AA come not only with specific advantages but also limitations. Moreover, cluster analysis dates from the 1930s and latent class analysis from the 1960s, and consequently both have an extensive literature for researchers to draw on. Hence, the advantages and limitations of these two methods are better known than is the case for AA, which dates from the 1990s and thus has a smaller literature. The key merit of AA, as some authors argue, is that it produces sharper solutions than cluster analysis (e.g., Elder & Pinnel, 2003) and through simulation studies others show AA is robust to various types of noise in the data (e.g., Chan et al., 2003). Some authors also point out that AA does not impose a strong model or set of external assumptions on the data. In particular, it does not impose the sorts of artificial orthogonality constraints that underlie several cluster methods (Li et al., 2003). Morup and Hansen (2012) also conclude that AA avoids the “hard assignment” of most cluster methods, whereby a case must be unambiguously assigned to one cluster. In contrast, AA simultaneously considers the relationships between all cases and all archetypes.

The limitations of AA mentioned in this literature include the fact that the chance of finding local (suboptimal) fits to the data increase with the number of archetypes (Cutler & Breiman, 1994). However, this is also true for other statistical techniques using optimization methods such as some types of cluster analysis and latent class analysis, which is particularly prone to such problems (Goodman, 1974) as well as those of local dependence amongst variables (Hagenaars, 1988). More AA specific limitations come from our own experience here, of which we regard two as important to note.

First, relatively large samples are preferable so that single data points do not carry disproportionate influence on the eventual solution, and so that each archetype is associated with a reasonable number of cases lending overall stability to the solution. Thus, in our experience, the minimum sample size for a solution with five archetypes would be in the order of 500. These numbers are not too dissimilar to recommendations for other statistical techniques such as factor analysis or structural equation modeling. While our conclusions obviously need further investigation, perhaps through simulation studies, they are not a concern here as our sample size is much larger, over 68,000 cases.

Second, AA is relatively computing intensive; requiring powerful computers if run times are to be kept in a practical range or better yet using the multiple cores present in many desktop and laptop machines through parallel computing techniques. Such computing power is necessary to allow the number of replications from different start points that are necessary for the researcher to be confident they have a good fit to the data. For example, here we run 30 trials from different random starting points and examine solutions from 1 to 15 archetypes for our 68,000 cases, which took a total of approximately 22 h to complete on a Mac Pro (2013) computer using all 12 Intel Xeon cores. Notwithstanding these limitations, we believe AA better suits our purpose here, through its theoretical ability to produce a parsimonious and distinct set of value configurations that fully describe the data.

5. Illustration of Archetypal Analysis: Schwartz Values in the World Values Survey Wave Five

5.1. The World Values Survey

We extracted the data for our study from the WVS Wave Five covering the period 2005–2009. These data are available online at The WVSs have been conducted in six waves between 1981 and 2014. These surveys contain data from 97 countries representing nearly 90% of the world population. The data are collected from a representative sample of respondents in each country, via face-to-face interviews in the local language. However, the specific questions and the sample countries vary from one wave to another. In this chapter, we focus on the 10 universal culture values in the Schwartz model and these were included for the first time in the WVS Wave Five. The WVS Wave Five contains data for 58 countries (N = 83,975), with sample size per country ranging from 954 for New Zealand to 3,051 for Egypt. However, 6 countries contain no data for Schwartz values, which leaves 52 countries (N = 74,031) for our analysis. Further, excluding the cases that have missing values for one or more of the 10 Schwartz values, we are left with 68,322 complete cases. We use data for the 68,322 complete cases on the 10 Schwartz values across the 52 countries in the WVS Wave Five as input for our AA. While Wave Five represents only one point in time and the Schwartz model is only one of several models of culture values, we believe these data are adequate to demonstrate how AA can extend and enhance culture-related theory in economics.

5.2. The Schwartz Values Model

Several different models of cultural values have been discussed in the literature (e.g., Hofstede, 2001; House et al., 2004; Schwartz, 1992). Although “groups can exist on multiple dimensions (cognitions, behaviors, and values), cross-cultural research has focused on shared cultural values as the major source of differentiation among national groups” (Tsui et al., 2007, p. 431). In our chapter, we focus on the Schwartz values model for several reasons. One, the Schwartz values have an underlying theoretical model that builds on a stream of prior research, notably Rokeach (1973), and provide a holistic view of the individual. Two, the Schwartz model is one of the well-known models of values at the individual level. Three, the Schwartz Value Survey – from which the 10 items are drawn for the WVS, is found to “exhibit high construct equivalence across scores of countries covering all continents” (Steenkamp & Ter Hofstede, 2002, p. 197). Four, from a practical perspective, individual level data on the 10 Schwartz values dimensions at the individual level are available online for 52 countries in the WVS Wave Five ( We need such individual level data to identify the diverse culture archetypes within and across countries, and profile the values structure of the archetypes. In contrast, only national level aggregated data is available in the public domain for other models such as those of Hofstede and GLOBE.

Schwartz (1994, p. 21) defines “values as desirable trans-situational goals, varying in importance, that serve as guiding principles in the life of a person or other social entity.” According to Schwartz (1994), values reflect responses to three universal requirements: (1) as biological organisms, (2) for coordinated social interaction, and (3) for smooth functioning and survival of groups. These could be summarized into two broad needs: (1) self-survival of individual, biological organisms, and (2) sustenance of group, social organizations, in large part to support individual survival. From these universal requirements, Schwartz derives 10 motivational value types (Schwartz, 1992). These he organizes into four categories around a circle: (1) self-transcendence (reflecting universalism and benevolence values); (2) conservation (encompassing tradition, conformity, and security values); (3) self-enhancement (reflecting power and achievement values); and (4) openness to change (which includes the values of self-direction and stimulation). A 10th value – hedonism – straddles both self-enhancement and openness to change. Inherent in the model are also two tensions, self-transcendence versus self-enhancement and openness to change versus conservation. These tensions are also reflected between the five self-oriented values to the left of the circle, and the five society-oriented values on the right of the circle. (see Fig. 1 for details.)

5.3. Measurement of Schwartz Values in the WVS Wave Five

The WVS Wave Five measures the 10 Schwartz values with an importance rating on a scale of 1 to 6. First, the interviewer describes some people and asks whether the respondent considers that person is like them. The interviewer gives the respondent a card with a verbal description of each scale point to facilitate this process (1, Very much like me; 2, Like me; 3, Somewhat like me; 4, A little like me; 5, Not like me; 6, Not at all like me). Next, the interviewer reads the respondent a statement, for example, “it is important to this person to be rich, to have a lot of money and expensive things” and asks them to indicate from the card whether this person is like them or not. This procedure is repeated for the other nine items in the Schwartz values (see Table 1). Thus, following most research on values, the WVS measures how important this value is to the respondent (each description contains the words “it is important to this person”). Across the 10 representative items, we therefore obtain a profile of which values the respondents regard as important to them. Finally, for ease of interpretation, we reverse this scale in our data so that in this article, 1 means “Not at all like me,” and therefore unimportant, and 6 means “Very much like me,” and therefore important.

Table 1.

Schwartz’s Ten Motivational Values and Their Measures in the WVS Wave Five.

Schwartz Motivational Values Measures (Variable Number) in the WVS Wave Five
1. Universalism Looking after environment (V88)
2. Benevolence To help people (V84)
3. Tradition Tradition (V89)
4. Conformity To always behave properly (V87)
5. Security Living in secure surroundings (V82)
6. Power To be rich (V81)
7. Achievement Being very successful (V85)
8. Hedonism To have a good time (V83)
9. Stimulation Adventure and taking risks (V86)
10. Self-direction To think up new ideas (V80)

Sources: Adapted from Schwartz (1994: 22), Spini (2003: 5), Venaik and Midgley (2015), and the WVS Wave Five.

Notes: Each of the 10 value items in WVS Wave Five are measured on a 6-point scale, with 1, very much like me; 2, like me; 3, somewhat like me; 4, a little like me; 5, not like me; and 6, Not at all like me. The following statement precedes the set of 10 Schwartz values items in WVS Wave Five: “Now I will briefly describe some people. Using this card, would you please indicate for each description whether that person is (1) very much like you, (2) like you, (3) somewhat like you, (4) a little like you, (5) not like you, or (6) not at all like you?” For example, the item used to measure the “tradition” motivational value is: “Tradition is important to this person; to follow the customs handed down by one’s religion or family.” The scales are reversed in our analysis so that high score means “very much like me”. The 10 Schwartz motivational values represent two bipolar value dimensions: self-transcendence (universalism, benevolence) versus self-enhancement (power, achievement, hedonism); and conservation (tradition, conformity, security) versus openness to change (hedonism, stimulation, self-direction).

The WVS adopts various strategies to overcome the potential problems of bias in cross-cultural surveys. One, the survey contains over 250 questions, and the size, complexity and diversity of the questionnaire make it difficult for respondents to have a systematic bias in their responses (Baumgartner & Steenkamp, 2001). Two, the survey contains all types of scales, nominal, interval, and ratio, of varying lengths. Three, the scale anchors are mixed, both in wording and direction, which makes it difficult for respondents to answer questions in a biased manner: “The advantage of balanced scales is that they have a built-in control for stylistic responding because a high (low) score cannot be obtained simply because of yea-saying (nay-saying)” (Baumgartner & Steenkamp, 2001, p. 144). Four, according to Fisher (1993, p. 303), “an important technique used by researchers to mitigate the effects of social desirability bias is indirect (i.e., structured, projective) questioning.” The WVS uses a projective technique to overcome social desirability bias in answering the Schwartz values questions. For example, the interviewer asks the following question before presenting the 10 Schwartz values questions: “Now I will briefly describe some people. Would you please indicate for each description whether that person is very much like you, like you, somewhat like you, a little like you, not like you, or not at all like you?” Thus, instead of directly asking the respondents about their motivational values, the survey asks respondents if they are similar to or different from another person with a specific values profile. Finally, any biases that do exist are inherent to the methodology of the WVS which, whatever its shortcomings, is the only publicly available source of individual level values data from multiple countries on a large scale.

6. Results and Discussion

We use the respondent scores on the 10 Schwartz values as inputs for our AA. We examine the AA solutions from 1 to 15 archetypes, repeating each analysis from 30 random starting points to avoid the problem of local minima and identify the best fit between solution and data. We apply the extremum distance estimator to the sequence of 15 best fits we find from the 30 runs, which indicates that the knee point is found at five archetypes. We therefore chose this solution for our global archetypes. To check that this best-fit solution is not an outlier, we also compared the 10 best fitting of the 30 runs for 5 archetypes. The archetype configurations from these 10 runs turn out to be very similar (the coefficient of variation averages 10% across the 5 archetypes and 10 Schwartz values). Hence, we conclude that the five best fit archetypal configurations are a good representation of our data.

Table 2 summarizes these five global archetypes that we identify here with the Schwartz values data in the WVS Wave Five. These configurations are also presented as radar plots in Fig. 2, one radar plot for each of the five archetypes from A1 to A5. Looking at any one of these radar plots clockwise from top, the five values on the right of the circle from benevolence to security broadly represent society-oriented values, and those on the left from power to self-direction represent self-oriented values. As shown in the figure, archetype A1 represents individuals who consider all society-oriented values as well as the two self-oriented values of self-direction and achievement as particularly important. Archetype A2 is somewhat like A1 in that both regard all society-oriented values as important; the key difference being moderate to high importance for self-oriented value of hedonism in A2 versus high importance for the self-oriented values of self-direction and achievement in A1. In contrast with A1 and A2, archetype A3 represents individuals who consider all self-oriented values as important except power and to a lesser extent achievement. These archetypal individuals also give high importance to the society-oriented value of benevolence. Next, we have archetype A4 that represents individuals who give high importance to both self- and society-oriented values. Finally, archetype A5, which is almost a mirror image of A4, represents individuals who give low importance to all values except to power which they consider to be moderately important.

Table 2.

Five Global Archetype Profiles on the 10 Schwartz Values in the WVS Wave Five (52 Countries, n = 68,322 Complete Cases).

Archetype Universalism Benevolence Tradition Conformity Security Power Achievement Hedonism Stimulation Self-Direction
A1 6.0 6.0 6.0 6.0 5.8 1.6 6.0 1.0 1.0 6.0
A2 5.4 5.3 6.0 6.0 6.0 1.0 1.0 4.4 1.0 1.0
A3 4.8 5.7 1.0 1.3 2.5 1.0 3.9 6.0 6.0 6.0
A4 5.3 5.4 6.0 6.0 6.0 6.0 6.0 6.0 6.0 5.8
A5 1.0 1.0 1.5 1.2 1.3 3.6 1.4 1.0 1.2 2.0

Fig. 2. 
Five Global Archetypes: Profiles on the 10 Schwartz Values in the WVS Wave Five.

Fig. 2.

Five Global Archetypes: Profiles on the 10 Schwartz Values in the WVS Wave Five.

We next examine the heterogeneity in values configurations both within and across countries. Table 3 summarizes the proportion of each country’s sample that is associated with each of the five archetypes, A1–A5. We use Euclidean distances between the cases and the five archetypes to classify each respondent to the archetype nearest to their own value configuration. If we look at the pattern of countrywise proportion of individuals by archetypes, we do not find a single country out of the 52 in our sample where all individuals in the country are associated with a single archetype. This result clearly indicates that the notion of a single national culture that can be attributed to all individuals within a country is questionable. Instead, what we find is that all countries have a heterogeneous population of individuals who associate with one archetype or another to a greater or lesser degree. Further, the pattern of the proportion of country sample by archetypes varies widely from one country to another.

Table 3.

Distribution of Archetypal Cases by Country.

No. Country N a A1 b A2 A3 A4 A5 Min. c Max.
1 Andorra 995 0.12 d 0.26 0.35 0.25 0.02 0.02 e 0.35
2 Argentina 917 0.30 0.27 0.18 0.13 0.12 0.12 0.30
3 Australia 1364 0.24 0.27 0.24 0.13 0.13 0.13 0.27
4 Brazil 1467 0.32 0.38 0.14 0.15 0.02 0.02 0.38
5 Bulgaria 865 0.29 0.27 0.09 0.21 0.14 0.09 0.29
6 Burkina Faso 1234 0.39 0.06 0.06 0.44 0.04 0.04 0.44
7 Canada 2057 0.32 0.23 0.23 0.18 0.03 0.03 0.32
8 Chile 938 0.19 0.22 0.17 0.37 0.05 0.05 0.37
9 China 1828 0.36 0.23 0.08 0.20 0.13 0.08 0.36
10 Cyprus 1037 0.43 0.09 0.10 0.35 0.03 0.03 0.43
11 Egypt 3000 0.29 0.32 0.01 0.35 0.03 0.01 0.35
12 Ethiopia 1396 0.17 0.04 0.09 0.48 0.21 0.04 0.48
13 Finland 990 0.21 0.32 0.25 0.13 0.08 0.08 0.32
14 France 969 0.18 0.32 0.27 0.16 0.07 0.07 0.32
15 Georgia 1309 0.44 0.24 0.02 0.29 0.01 0.01 0.44
16 Germany 1918 0.21 0.26 0.26 0.14 0.12 0.12 0.26
17 Ghana 1465 0.35 0.02 0.01 0.60 0.01 0.01 0.60 f
18 Great Britain 1004 0.24 0.30 0.23 0.18 0.06 0.06 0.30
19 Hungary 989 0.25 0.28 0.08 0.35 0.04 0.04 0.35
20 India 1264 0.33 0.08 0.07 0.45 0.07 0.07 0.45
21 Indonesia 1833 0.20 0.31 0.07 0.39 0.03 0.03 0.39
22 Iran 2570 0.34 0.14 0.04 0.46 0.01 0.01 0.46
23 Japan 893 0.13 0.20 0.16 0.02 0.49 0.02 0.49
24 Jordan 1144 0.08 0.05 0.01 0.85 0.00 0.00 0.85
25 Malaysia 1197 0.24 0.16 0.10 0.36 0.15 0.10 0.36
26 Mali 1237 0.28 0.11 0.04 0.56 0.01 0.01 0.56
27 Mexico 1448 0.31 0.24 0.15 0.25 0.05 0.05 0.31
28 Moldova 989 0.34 0.21 0.09 0.26 0.09 0.09 0.34
29 Morocco 1069 0.28 0.10 0.06 0.51 0.05 0.05 0.51
30 Netherlands 1028 0.08 0.28 0.39 0.12 0.13 0.08 0.39
31 Norway 1013 0.21 0.29 0.37 0.08 0.05 0.05 0.37
32 Peru 1353 0.40 0.21 0.18 0.13 0.09 0.09 0.40
33 Poland 970 0.41 0.20 0.09 0.27 0.03 0.03 0.41
34 Romania 1428 0.32 0.27 0.05 0.28 0.08 0.05 0.32
35 Russia 1725 0.22 0.30 0.14 0.23 0.12 0.12 0.30
36 Rwanda 1306 0.29 0.14 0.04 0.47 0.07 0.04 0.47
37 Serbia/Montenegro 1100 0.18 0.27 0.14 0.22 0.19 0.14 0.27
38 Slovenia 966 0.32 0.20 0.20 0.24 0.04 0.04 0.32
39 South Africa 2880 0.18 0.09 0.05 0.65 0.02 0.02 0.65
40 South Korea 1198 0.09 0.22 0.24 0.23 0.21 0.09 0.24
41 Spain 1145 0.21 0.30 0.17 0.27 0.05 0.05 0.30
42 Sweden 983 0.22 0.24 0.39 0.11 0.05 0.05 0.39
43 Switzerland 1205 0.20 0.27 0.38 0.11 0.04 0.04 0.38
44 Taiwan 1218 0.20 0.51 0.11 0.13 0.05 0.05 0.51
45 Thailand 1504 0.18 0.18 0.14 0.29 0.21 0.14 0.29
46 Trinidad/Tobago 979 0.51 0.14 0.10 0.19 0.06 0.06 0.51
47 Turkey 1280 0.35 0.07 0.07 0.47 0.03 0.03 0.47
48 Ukraine 865 0.19 0.29 0.12 0.27 0.12 0.12 0.29
49 United States 1184 0.23 0.29 0.20 0.16 0.12 0.12 0.29
50 Uruguay 956 0.18 0.36 0.21 0.13 0.12 0.12 0.36
51 Viet Nam 1312 0.20 0.23 0.07 0.45 0.05 0.05 0.45
52 Zambia 1338 0.28 0.06 0.12 0.47 0.08 0.06 0.47
Minimum 865 0.08 0.02 0.01 0.02 0.00 0.00 0.24
Maximum 3000 0.51 0.51 0.39 0.85 0.49 0.14 0.85
Mean 1314 0.26 0.22 0.15 0.29 0.08 0.06 0.39
Std. Deviation 469 0.09 0.10 0.10 0.17 0.08 0.04 0.11

Notes: N is the sample size for each country (total 68,322 complete cases across the 10 Schwartz values).


A1–A5 are the five archetypes. A1, Traditional Achiever; A2, Traditional Materialist; A3, Postmaterialist; A4, Materialist; A5, Secular Minimalist.


Min. and Max. are the minimum and maximum proportion of sample in each country associated with an archetype.


The cell entries for columns A1–A5 are the proportion of each country sample associated with the respective archetype.


The underlined entries are proportions ≤ .02.


The bold entries are proportions ≥0.50.

However, we do find a few countries have a “dominant” archetype. When we scan the maximum column in Table 3 for the bold entries (reflecting proportions ≥ 0.50), we find that seven countries have 51% or more individuals in the respective country sample who are associated with just one of the five archetypes. The most common dominant archetype is A4 (for five countries), followed by A1 and A2 for one country each. Four of these seven countries are also part of the ten countries that have a negligible proportion of 0.02 or less of their sample individuals associated with one of the five archetypes (the underlined entries in the minimum column in Table 3). For example, in Jordan, 85% of the individuals in the sample are associated with A4, and conversely, a negligible proportion of the country’s sample is associated with archetypes A3 and A5. Thus, we can say that Jordan is the most culturally homogeneous country in the WVS Wave Five, based on the configurations of the 10 Schwartz values. However, even in this country 100% of the sample is not monocultural. Eight and five percent of the country’s sample is associated with archetypes A1 and A2 respectively. Six other countries – South Africa, Ghana, Mali, Morocco, Taiwan, and Trinidad and Tobago – have 51–65% of their respective samples associated with one of the five archetypes, and therefore can be considered to have a dominant, but not a single, archetype.

Next, we look at the pattern of association of archetypes in 45 countries that are relatively heterogeneous, that is, for countries that require two or more archetypes to represent a cumulative proportion of 51% or more of the country sample. One way to examine this heterogeneity is to compute the difference between the maximum and minimum proportion of a country sample associated with an archetype. Countries with large differences between their maximum and minimum proportion are those that have one relatively dominant archetype, for example, the seven countries discussed above. Conversely, countries with small differences between their maximum and minimum proportion are those that are relatively more heterogeneous and require two or more archetypes to represent at least 51% of the country’s sample. We find three countries – Serbia and Montenegro, South Korea, and Thailand – that require at least three archetypes to represent 51% or more of the country sample. For example, in South Korea, the maximum proportion associated with any archetype is 0.24 (archetype A3), and the proportions associated with the other archetypes are 0.23, 0.22, 0.21, and 0.09, corresponding with the archetypes A4, A2, A5, and A1, respectively. South Korea is therefore the most culturally heterogeneous country in the WVS Wave Five based on the configurations of the 10 Schwartz values. The remaining 42 countries require at least two archetypes to represent 51% or more of the country’s sample, for example, France. Fig. 3 summarizes the archetypal distribution for three countries – Jordan, France, and South Korea – that require one, two and three archetypes respectively to represent at least 51% of the respective sample.

Fig. 3. 
Archetypal Distributions for Jordan, France, and South Korea in the WVS Wave Five.

Fig. 3.

Archetypal Distributions for Jordan, France, and South Korea in the WVS Wave Five.

In addition to highlighting the important phenomena of intranational heterogeneity, our AA results also show a high level of transnational homogeneity in Schwartz values configurations. That is, every country shares similar values configurations with every other country in the sample, to a greater or lesser degree. Let us look at USA and Japan, the two countries that are often considered to be opposites in the existing literature, especially on the individualism–collectivism dimension. This dimension is similar to our self- versus society-oriented set of five values each on the left and right of the Schwartz circle respectively (see Fig. 1). As shown in Table 3, the percent of Japanese sample associated with each of the five archetypes from A1 to A5 is 13, 20, 16, 2, and 49%, respectively, and the corresponding percentages for the US sample are 23, 29, 20, 16 and 12%, respectively. Whereas the percent of the sample associated with A5 is very different between Japan and USA (49 vs 12%), that associated with A3 is somewhat similar (16 vs 20%). Thus, both countries have people with similar values to a greater or lesser degree (see Fig. 4). This result further reinforces the need to model nations as heterogeneous configurations of multiple values rather than as homogenous entities with a single, common set of values.

Fig. 4. 
Archetypal Distributions for Japan and the United States in the WVS Wave Five.

Fig. 4.

Archetypal Distributions for Japan and the United States in the WVS Wave Five.

In sum, our AA approach shows that no country can be represented with a single configuration of values. About 13% of our sample countries (7 out of 52) have one dominant archetype (i.e., one archetype represents 51–85% of the respective country population), Six percent of our sample countries (3 out of 52) require at least three archetypes to represent 51% or more of the respective country’s population. And a large majority of countries in our sample (81%, or 42 out of 52) require at least two archetypes to represent 51% or more of the respective country’s population. These results call into question the orthodoxy of treating countries as homogeneous entities that can be represented with one set of scores or a single archetype for all individuals in the country. In addition, the finding of similar culture archetypes across countries shows that values configurations are transnational and people with similar values are dispersed globally, albeit with varying proportions in different countries.

6.1. Linking Archetypes with Other Characteristics

We next examine if our five culture archetypes are associated with demographic and other values characteristics of individuals in the WVS Wave Five database. The results of this analysis will also help us to label the culture archetypes that we identify here. We test the individual-level relationships between archetypal membership and four demographic variables (gender, age, education, and social class), and four values indices based on the work of Inglehart and Welzel, namely, postmaterialist index (12 items), and autonomy, secular values and emancipative values indices (Inglehart, 1990; Inglehart & Welzel, 2005). The scores on these indices for each individual are available in the WVS Wave Five database. Each of these eight variables were regressed on archetype membership, using weighted effect coded dummy variable regression as an appropriate approach for observational data with differing group sizes as here (Te Grotenhuis et al., 2017).

We find archetype A1 is highly traditional but also achievement oriented. We therefore label A1 as a Traditional Achiever. Archetype A2 is relatively both traditional and materialist, hence we consider it as a Traditional Materialist. This archetypal group is also significantly older and less educated than the mainstream. Archetype A3 is relatively higher educated, and high on secular, emancipative and postmaterialist values, hence we regard A3 as a postmaterialist. Archetype A4 is much younger and, almost opposite to A3, has higher levels of materialist and survival values. Hence, we characterize A4 as a Materialist. Finally, archetype A4 is minimalist on most Schwartz values dimensions, but is also highly secular, hence we label A4 as Secular Minimalist. Overall, our analysis shows our five archetypes have strong discriminant validity on the Schwarz’s dimensions, and also have strong external validity based on their significant relationships with a set of WVS demographics and sociocultural indices.

7. Conclusion

In this chapter, we present the concept of culture archetypes, which we believe contributes to a better understanding of culture as a configuration of values within and across countries. Our archetypal approach allows us to represent cultural diversity across the world with a small number of global culture values configurations. Further, by examining the Euclidean distance between the values profile of individuals in our data and the profile of the different archetypes identified with our AA method, we can classify individuals into one of our five archetypes. The AA method thus allows us to capture culture values heterogeneity within countries in a parsimonious manner This is an important contribution to the current culture literature that largely assumes national cultures to be homogeneous. In addition, our AA enables us to identify transnational culture archetypes, that is, groups of individuals who share similar culture values configuration across countries. This is another important contribution to the culture-related discourse that is currently dominated by the popular stereotypical national culture perspective in economics and other disciplines. Both these findings also call into question Huntington’s (1993) thesis of “clash of civilization” between nations with starkly different values. We believe interactions among individuals in different archetypal groups within and across countries is likely to reduce differentiation and tensions and enhance mutual understanding, as people exchange and appreciate the values of others, and assimilate those that they find useful into their own value systems. Such interactions can also potentially give rise to new values configurations. Further, interactions among diverse culture archetypal groups can enhance creativity through conceptual expansion and integration in novel ways (Leung, Maddux, Galinsky, & Chiu, 2008).

Our study also shows that some of the archetypes and the individuals associated with these archetypes embody seemingly unrelated sets of self- and society-oriented values. This is somewhat at odds with the Schwartz values model that considers these two sets of values to be at the opposite ends of his bipolar dimensions. It also challenges the popular view that individuals and societies can hold either individualistic or collectivistic beliefs but cannot be ambidextrous and hold both types of beliefs simultaneously. Our finding of diversity of values within each archetype raises important philosophical questions in relation to values research in economics and other disciplines, namely, is there a tension between self- and society-oriented values? And how do individuals with divergent values manage and resolve this tension in their day-to-day practices? It is also important to highlight that the insight about individuals embodying divergent values would not be evident if we had combined similar sets of values into a single overall index, as is often done in economics and other fields. The unique ability of AA to identify diverse configurations that reflect the underlying richness and diversity of the data, without the straightjacket of a priori theoretical “tensions”, means that the novel AA methodology has the potential for application and novel discoveries across a range of disciplines including economics that examines a broad range of characteristics of nations, firms, and individuals.

Values and beliefs are at the core of the institutional theory of economic development. Thus, examining the changes in society’s values can help us understand the dynamics of change, an undertaking that is beyond the scope of the static neoclassical theory (North, 2005, p. 125). Whereas the human belief systems in the past were largely designed to deal with the uncertainties of the physical environment, current and future economic development is largely dependent on the ability of societies to deal with the complexities of the human environment (North, 2005, p. 44). Furthermore, unlike the physical environment that is largely ergodic in nature, the human environment is relatively nonergodic (North, 2005, p. 19). Thus, understanding the dynamics of human belief system that is complex and uncertain is critical for understanding the past trajectory and mapping the future direction of national growth and development. We believe our AA approach would be a useful tool to map the trajectory of configurations of societal values and beliefs, and link these to various aspects of economic growth and development.

In understanding the phenomena of culture change, the focus of economics has largely been on the intergenerational transmission aspect of values, although the sharing aspect of values is acknowledged. In this view, values are transmitted from one generation to another as social heritage, and individuals are regarded as “passive porters of a social tradition” (Dollard, 1939, p. 54). We believe the economic approach to modeling culture simply as transmission may be limiting as culture encompasses diverse aspects of values that may or may not involve intergenerational transmission from the older to the younger generation. The intergenerational aspect of culture was potentially relevant in earlier times when knowledge was largely passed on personally, supplemented with impersonal sources such as community gatherings, and later through the print, radio, and television media. However, with the rise of the internet, there has been an exponential growth in the production of information and the speed with which it can be accessed and transmitted around the world. Thus, in the contemporary world, large amounts of information is easily available at low cost to practically everyone around the world almost instantly.

The model of culture change in contemporary information-rich society, therefore may be more like the transmission and mutation of viruses rather than the transmission from one generation to another. Values and beliefs can mutate endogenously and rapidly, within a generation, instead of being isomorphic with the previous generation via transmission, as modeled in economics. Humans are “not only the carriers and creatures of culture – they are also creators and manipulators of culture” (Kroeber & Kluckhohn, 1952, p. 49). Culture evolves over both time and space, globally as well as locally, through the influence of travel and communication (e.g., Tsui et al., 2007). We are potentially in a new era where the intergenerational transmission of values occurs in reverse, from the younger to the older generation, on issues such as technology, same-sex marriage, importance of religious institutions, and a concern for sustainable development.

Finally, from a practical perspective of an experimentalist, a values questionnaire could be administered to the subjects of interest to identify their values profile and culture groups created with individuals having similar values configurations. This is unlike the popular approach wherein groups are created a priori based on nationality, gender or other characteristics and differences examined between these a priori defined groups. The extant approaches to measuring culture differences are seemingly inconsistent with the core idea of “shared values” underlying the culture concept. In addition, diverse values profiles and multiple culture groups are more likely to be identified in heterogeneous subject pools that vary by age, education, and other variables that influence values, and less likely to be identified in relatively homogenous subject pools such as students. At the end, the sample archetype profiles and groups could be compared and contrasted with the global archetypes profiles and similarities and differences explained with detailed, contextual data about the subjects under study.

Overall, our archetypal approach allows us to examine configurations of values without imposing a priori exogenous national or other boundaries on individuals’ values. By decoupling the identification of culture values archetypes and the groups who identify with these archetypes, we discover new configurations of values archetypes and novel compositions of countries with multiple diverse archetypes. We believe our archetypal culture perspective shows that the real world of cultural diversity is more complex than what is reflected in the stereotypical national culture perspective. In future research, extending our AA approach to individuals within each nation can potentially provide a more refined understanding of cross-national cultural differences and similarities. We may also identify novel archetypes that are unique to each country, providing us with a more nuanced and detailed understanding of subnational and transnational cultures. We are confident that our archetypal culture perspective will engender a more enlightened debate on how best to conceptualize and model the diversity of cultures we see around us. Finally, our archetypal approach to culture analysis can be extended to other areas of economics, such as identifying archetypes of nations based on a set of characteristics related to the specific problem of interest to the researcher.


Baumgartner & Steenkamp (2001) Baumgartner, H. , & Steenkamp, J. E. M. (2001). Response styles in marketing research: A cross-national investigation. Journal of Marketing Research, 38(2), 143156.

Beugelsdijk & Maseland (2011) Beugelsdijk, S. , & Maseland, R. (2011). Culture in economics: History, methodological reflections, and contemporary applications. Cambridge: Cambridge University Press.

Brewer & Venaik (2014) Brewer, P. , & Venaik, S. (2014). The ecological fallacy in national culture research. Organization Studies, 35(7), 10631086.

Chan, Mitchell, & Cram (2003) Chan, B. H. P. , Mitchell, D. A. , & Cram, L. E. (2003). Archetypal analysis of galaxy spectra. Monthly Notices of the Royal Astronomical Society, 338(3), 790795.

Christopoulos (2014) Christopoulos, D. T. (2014). Developing methods for identifying the inflection point of a convex/concave curve. Retrieved from Accessed on July 15, 2017.

Cutler & Breiman (1994) Cutler, A. , & Breiman, L. (1994). Archetypal analysis. Technometrics, 36(4), 338347.

Dollard (1939) Dollard, J. (1939). Culture, society, impulse, and socialization. American Journal of Sociology, 45(1), 5063.

Elder & Pinnel (2003) Elder, A. , & Pinnel, J. (2003). Archetypal analysis: an alternative approach to finding and defining segments. Proceedings of the Sawtooth Software Conference, Sequim, WA, June, 113132.

Eugster & Leisch (2009) Eugster, M. J. A. , & Leisch, F. (2009). From spider-man to hero – archetypal analysis in R. Journal of Statistical Software, 30(8), 123.

Fischer & Schwartz (2011) Fischer, R. , & Schwartz, S. H. (2011). Whence difference in value priorities?: Individual, cultural, or artifactual sources. Journal of Cross-Cultural Psychology, 42(7), 11271144.

Fisher (1993) Fisher, R. J. (1993). Social desirability and the validity of indirect questioning. Journal of Consumer Research, 20, 303315.

Geertz (1983) Geertz, C. (1983). Local knowledge: Further essays in interpretive anthropology. New York, NY: Basic Books.

Goodman (1974) Goodman, L. A. (1974). Exploratory latent structure analysis using both identifiable and unidentifiable models. Biometrika, 61, 215231.

Guiso, Sapienza, & Zingales (2006) Guiso, L. Sapienza, P. , & Zingales, L. (2006). Does culture affect economic outcome? Journal of Economic Perspectives, 20(2, Spring), 2348.

Guiso, Sapienza, & Zingales (2016) Guiso, L. Sapienza, P. , & Zingales, L. (2016). Long-term persistence. Journal of the European Economic Association, 14(6), 14011436.

Hagenaars (1988) Hagenaars, J. A. (1988). Latent structure models with direct effects between indicators: local dependence models. Sociological Methods and Research, 16(3), 379405.

Hall (1977) Hall, E. T. (1977). Beyond culture. New York, NY: Doubleday.

Harris (1979) Harris, M. (1979). Cultural materialism: The struggle for a science of culture. New York, NY: Random House.

Henrich, Heine, & Norenzayan (2010) Henrich, J. Heine, S. J. , & Norenzayan, A. (2010). The weirdest people in the world? Behavioural and Brain Sciences, 33(2/3), 175.

Herskovits (1948) Herskovits, M. (1948). Man and his works: The science of cultural anthropology. New York, NY: Alfred Knopf.

Hofstede (2001) Hofstede, G. (2001). Culture’s consequences: Comparing values, behaviors, institutions and organizations across nations. Thousand Oaks, CA: Sage.

House, Hanges, Javidan, Dorfman, & Vipin (2004) House, R. J. , Hanges, P. J. , Javidan, M. , Dorfman, P. W. , & Vipin, G. (2004). Culture, leadership, and organizations: The GLOBE study of 62 societies. Thousand Oaks, CA: Sage.

Huntington (1993) Huntington, S. P. (1993). The clash of civilizations? Foreign Affairs, 72(3), 2249.

Inglehart (1990) Inglehart, R. (1990). Culture shift in advanced industrial society. New Jersey, NJ: Princeton University.

Inglehart (2006) Inglehart, R. (2006). Mapping global values. Comparative Sociology, 5(2–3), 115136.

Inglehart & Welzel (2005) Inglehart, R. , & Welzel, C. (2005). Modernization, cultural change, and democracy: The human development sequence. New York, NY: Cambridge University.

Johnson & Mislin (2011) Johnson, N. D. , & Mislin, A. A. (2011). Trust games: A meta-analysis. Journal of Economic Psychology, 32(5), 865889.

Kamakura & Mazzon (1991) Kamakura, W. A. , & Mazzon, J. A. (1991). Value segmentation: A model for the measurement of values and value systems. Journal of Consumer Research, 18(2), 208218.

Kroeber & Kluckhohn (1952) Kroeber, A. L. , & Kluckhohn, C. (1952). Culture: A critical review of concepts and definitions. Cambridge, MA: The Peabody Museum.

Leung, Maddux, Galinsky, & Chiu (2008) Leung, A. K. , Maddux, W. W. Galinsky, A. D. , & Chiu, C. (2008). Multicultural experience enhances creativity. American Psychologist, 63(3), 169181.

Li, Wang, Louviere, & Carson (2003) Li, S. , Wang, P. , Louviere, J. , & Carson, R. (2003). Archetypal analysis: A new way to segment markets based on extreme individuals. ANZMAC Conference Proceedings, Adelaide, December, 1674–1679.

Lytle, Brett, Barsness, Tinsley, & Janssens (1995) Lytle, A. L. , Brett, J. M. , Barsness, Z. I. , Tinsley, C. H. , & Janssens, M. (1995). A paradigm for confirmatory cross-cultural research in organizational behavior. In L. L. Cummings & B. M. Staw (Eds.), Research in organizational behavior (Vol. 17, pp. 167214). Greenwich, CT: JAI Press.

Maruyama (1999) Maruyama, M. (1999). Heterogram analysis: Where the assumption of normal distribution is illogical. Human Systems Management, 18(1), 5360.

Morup & Hansen (2012) Morup, M. , & Hansen, L. K. (2012). Archetypal analysis for machine learning and data mining. Neurocomputing, 80(March), 5463.

North (2005) North, D. C. (2005). Understanding the process of economic change. Princeton, NJ: Princeton University Press.

Rokeach (1973) Rokeach, M. (1973). The nature of human values. New York, NY: Free Press.

Schwartz (1992) Schwartz, S. H. (1992). Universals in the content and structure of values: Theoretical advances and empirical tests in 20 countries. Advances in Expermental Social Psychology, 25, 165.

Schwartz (1994) Schwartz, S. H. (1994). Are there universal aspects in the structure and contents of human values? Journal of Social Issues, 50(4), 1945.

Sen (2005) Sen, A. (2005). The argumentative indian: Writings on indian history, culture and identity. New York, NY: Farrar, Straus and Giroux.

Spini (2003) Spini, D. (2003). Measurement equivalence of 10 value types from the Schwartz value survey across 21 countries. Journal of Cross-Cultural Psychology, 34(1), 323.

Steenkamp & Ter Hofstede (2002) Steenkamp, J. E. M. , & Ter Hofstede, F. (2002). International market segmentation: Issues and perspectives. International Journal of Research in Marketing, 19(3), 185213.

Taras, Steel, & Kirkman (2016) Taras, V. Steel, P. , & Kirkman, B. L. (2016). Does country equate with culture? Beyond geography in the search for cultural boundaries. Management International Review, 56(4), 455487.

Grotenhuis et al. (2017) Te Grotenhuis, M. , Pelzer, B. Eisinga, R. Nieuwenhuis, R. Schmidt-Catran, A. , & Konig, R. (2017). When size matters: Advantages of weighted effect coding in observational studies. International Journal of Public Health, 62(1), 163167.

Triandis (1994) Triandis, H. C. (1994). Culture and social behavior. New York, NY: McGraw-Hill.

Tsui, Nifadkar, & Ou (2007) Tsui, A. S. Nifadkar, S. S. , & Ou, A. Y. (2007). Cross-national, cross-cultural organization behavior research: Advances, gaps, and recommendations. Journal of Management, 33(3), 426478.

Tylor (1871) Tylor, E. B. (1871). Primitive culture: Researches into the development of mythology, philosophy, religion, art, and custom. London: John Murray.

Venaik & Brewer (2013) Venaik, S. , & Brewer, P. (2013). Critical issues in the Hofstede and GLOBE national culture models. International Marketing Review, 30(5), 469482.

Venaik & Midgley (2015) Venaik, S. , & Midgley, D. F. (2015). Mindscapes across landscapes: Archetypes of transnational and subnational culture. Journal of International Business Studies, 46(9), 10511079.

Woodward (2012) Woodward, C. (2012). American nations: A history of the eleven rival regional cultures of North America. New York, NY: Penguin.

World Values Survey (WVS) Wave Five (2005–2009) World Values Survey (WVS) Wave Five . (2005–2009). Retrieved from


This work was supported by a research and development grant from INSEAD.