Habitus,Globalization,and Cosmopolitanism: How Dispositions,Class,and Migration Shape Global Self-Identification Differently in 74 Countries

Abstract

This paper critically examines conceptualizations of “transnational,” “global” or “cosmopolitan” habitus in migration, globalization and inequality research. We conceptualize habitus as a configuration of basic dispositions, using items from the Schwartz values battery, and compare the effect of habitual dispositions, class, and migration on cosmopolitan self-identification via country-specific multiple regression models, based on data from the World Values Survey (2005–2015) of around 100,000 individuals from 74 countries. We show that migration experiences and habitual dispositions of openness to new and stimulating experiences rather than orientations toward tradition and security are associated with stronger global self-identification. We also find a tendency among the upper classes toward cosmopolitan identification; however, these effects vary between an economic and a cultural capital route as well as among societies. Our findings thus caution against the uncritical use of “one-size-fits-all” concepts such as transnational, global and cosmopolitan habitus independent from specific social and cultural contexts.

Introduction

The question of how globalization affects individuals and societies preoccupies public debate and the social sciences alike. In many societies across the world, this can be observed clearly in the form of increasing polarization and the formation of new cleavages between somewhat nationally and cosmopolitan-oriented groups, class fractions, and milieus, whose attitudes toward the nation-state, migration and, in particular, issues of (national) culture and social cohesion, differ significantly (see de Wilde et al., 2019; Ollroge, 2023). These developments raise the question of to what extent cosmopolitanism—understood as an individual's sense of belonging to humanity as a single community rather than identifying with the national society—is actually a lived reality across the globe and how it is structured among social groups and within social spaces (see Pichler 2012; Skrbiš and Woodward, 2013 for an overview and synthesis of empirical research and different conceptualizations of cosmopolitanism). ¹ The possible spread of cosmopolitan identities is also of sociological relevance because it can potentially facilitate collective responses to global problems affecting all of humanity (e.g., climate change, control of AI).

In the 1990s and 2000s, social theorists tended to assume that, by and large, people around the world would become ever more cosmopolitan in their orientations and identities as a result of ongoing globalization and transnationalization processes (Appadurai, 1996; Beck, 2006; Hannerz, 1990). Overall, however, empirical research since has not shown any such general trend: Not only do people mean very different things when they state that they see themselves as “citizens of the world” (Braun et al., 2018; Carmona et al., 2022; Pichler, 2012), but, more importantly, there also seem to be noticeable differences among as well as within national societies globally with regard to the extent to which people self-identify as cosmopolitan (see also Haller and Roudometof, 2010). Indeed, research (Bayram, 2015; Gorman and Seguin, 2018; Jung, 2008; Katz-Gerro et al., 2024; Pichler, 2008, 2012; Zhou, 2016) has indicated that global identification differs between countries by societal factors (level of affluence or globalization) and individual factors (e.g., age, class, education and place of residence); but these factors do not necessarily have the same effect across countries. ² While the literature thus agrees that cosmopolitan orientations and identities manifest themselves quite differently in different societies, there is much less agreement on how to account theoretically for this interplay of macro- and micro-level factors (cp. Jung, 2008: 579; Zhou, 2016: 154). To this end, research often relies on psychological notions such as group threat (Gorman and Seguin, 2018) and values (e.g., Bayram, 2015) that generally assume a relatively stable and “trans-situational” effect across diverse societal contexts (cp. Bayram, 2015: 453, 456).

Against this background, this article contributes empirically and conceptually to the debate on the relationship between globalization and cosmopolitanism. Conceptually, we draw on Bourdieu's (1984 [1979], 1990 [1980]) notion of habitus, which refers to people's sociostructurally-based patterns of thought, perception and evaluation, thus linking macro and micro levels. In research on migration, class and globalization over the past three decades, the idea of “transnational,” “global” or “cosmopolitan” habitus has gained prominence as a means of capturing how people's outlooks and belonging may change when transcending national borders (e.g., Carlson and Schneickert, 2021; Guarnizo, 1997; Illouz and John, 2003; Jung et al., 2017; Kelly and Lusis, 2006; Nava, 2007; Nukaga, 2013; Stahl et al., 2024). This predominantly qualitative research suggests that such habitus entail dispositions that are primarily related to migration experiences, but possibly also to basic orientations and social class. However, such relationships have, to our knowledge, not been investigated in large-scale (comparative) quantitative research. In this article, we therefore suggest operationalizing habitus as a set of basic dispositions as suggested by Carlson and Schneickert (2021) and use country-specific multiple regression models—drawing on comprehensive World Values Survey (2005–2015) data from almost 100,000 respondents across 74 countries—in order to examine the relationships between basic habitual dispositions (tradition, status, and experience orientation), class (cultural and economic capital), and migration background with global self-identification.

State of Research

Discussions about the relationship between globalization, cosmopolitanism and habitus have predominantly emerged within migration and stratification research as both fields are deeply invested in analyzing the drivers and effects of globalization on societies across the world. Consequently, as the following literature review shows, both strands of research have engaged intensively with the question of how class and migration might affect people's sense of global belonging using the concept of transnational habitus. We bring these research traditions together and propose an operationalization of habitus by continuing to distinguish analytically between migration, class, and habitual dispositions in the following, to explore its associations with global self-identification even if the relations among the former are taken into account.

Cosmopolitanism and Migration

Research on the link between cosmopolitanism and migration intensified in the early 2000s and assumed that the experience of migration would generally go hand in hand with more cosmopolitan attitudes and practices (e.g., Beck, 2006; Vertovec and Cohen, 2002). The reasoning tended to be that, as people are exposed to different cultural contexts, which go beyond their own local experiences and broaden their perspectives, they may engage with the “Other,” develop complex transnational affiliations (multiple frames of reference in different national and transnational contexts), and therefore adopt a more open and tolerant attitude. However, subsequent research has quickly yielded diverging conclusions on this matter. Based on a literature review of various migration studies that focus on the European context, Horst and Olsen (2021: 14) argue that migration and transnationalism can indeed be seen as “important pathways towards creating cosmopolitan outlooks” (see also Schlenker, 2017). By contrast, Glick Schiller et al. (2011: 404) stress that “experiences of migration, travel and pilgrimage do not necessarily produce either cosmopolitan sociabilities or identities.” Thus, one can just as well imagine that encounters with other people and cultures actually strengthen existing ethnic or national identities, rendering them more parochial than cosmopolitan. Ultimately, then, as Schlenker (2017: 327–328) comments, it is an open empirical question whether (first or second generation) migrants are more likely to see themselves as global citizens. ³

For this reason, our first research question (RQ 1) is: To what extent and in what direction are migration experiences associated with cosmopolitan identifications in different countries? Our general expectation is that experiences of migration and the transnational networks that may accompany them will lead to a stronger global self-identification and a weaker identification with national society (both the society of origin and of arrival). However, a compensatory effect could work in the opposite direction, with migrants over-identifying with one of the two societies; and these relationships may also depend on national migration and integration policies.

Cosmopolitanism as a Class Project?

Alongside research highlighting the possible effect of migration on global self-identification, class is another important factor. Indeed, research on the globalization and transnationalization of class often regards cosmopolitanism as the preserve of (corporate) elites and (transnational) upper classes (e.g., Calhoun, 2002; Carroll, 2010; Sklair, 2001; Schneickert, 2018). Additionally, research from Europe has shown that transnational activities and cosmopolitan attitudes are unequally distributed across society (Delhey et al., 2015; Fligstein, 2008; Mau, 2010), as is the acquisition of so-called cosmopolitan capital (Bühlmann et al., 2013), e.g., via educational practices (Carlson et al., 2017; Igarashi and Saito, 2014; Weenink, 2007). Thus, one may assume that there is a social gradient in global self-identification, considering that higher resource endowments reduce dependence on the nation-state and allow for increased transnational activity and orientation. This may concern economic capital, since transnational activity is usually more expensive than local activity, but it may also concern cultural capital, for instance, because language skills and intercultural competencies as well as knowledge of other cultural contexts are relevant (Carlson et al., 2017).

Nevertheless, this does not mean that cosmopolitan orientations and practices are necessarily a purely elite phenomenon, as studies on working-class cosmopolitanism have shown (Lamont and Aksartova, 2002; Werbner, 1999). Furthermore, when looking beyond the European context, empirical evidence on this point seems even more ambivalent. Pichler (2012: 36) has found that the stratification of cosmopolitan identities and attitudes differs among countries. In turn, Zhou (2016: 169–170) has identified a sociodemographic stratification of global self-identification (the young, the better educated, and the urban being more cosmopolitan than the older, the less educated and those living in rural areas), while regarding economic resources as seemingly less influential. Thus, our second research question (RQ 2) is: To what extent and in what direction are cultural and economic capital—as key indicators of social class—related to global self-identification in different countries? Our general expectation is that higher social classes will have a stronger global self-identification because they are less tied to the nation and welfare state. However, based on Bourdieu's (1984 [1979]) assumptions about social space, we expect differences between economic and cultural class fractions within the upper class. Local specific struggles for distinction or weak welfare states, in which the lower classes are particularly globally oriented, could also create opposing relations.

Cosmopolitanism and Habitual Dispositions

By examining cosmopolitan self-identification across the world in relation to Bourdieu's (1984 [1979], 1990 [1980]) habitus concept, we can view people's orientations and identities as socially structured, thereby linking micro and macro levels. Habitus is sociologically defined as a “system of dispositions” (Bourdieu 1990 [1980]: 53), referring to a person's patterns of thought, perception and evaluation. ⁴ These dispositional patterns also generate and organize action and behavior as well as attitudes, beliefs and general orientations, depending on the specific social position (in social space) and specific contexts (in social fields) of which an individual is part (Bourdieu 1984 [1979], 1990 [1980]). Thus, ultimately, whether people act in certain ways or manifest specific attitudes and beliefs (of a cosmopolitan kind, for example) is context-dependent.

Researchers, especially in migration studies, have employed habitus theory in global contexts, introducing concepts such as “cosmopolitan habitus” (Jung et al., 2017; Nava, 2007), “global habitus” (Illouz and John, 2003) and “transnational habitus” (Guarnizo, 1997; Kelly and Lusis, 2006; Nedelcu, 2012; Nukaga, 2013; Stahl et al., 2024; for an overview and critical perspective, see Carlson and Schneickert, 2021). On balance, all these concepts strive to capture the ways in which people increasingly move in social contexts that systematically transcend nation-state borders and thus incorporate different kinds of dispositions that may manifest themselves as cosmopolitan orientations, attitudes and/or practices. These studies helpfully alert us to the fact that migration and other forms of cross-border experiences may change people's frames of reference and basic orientations. In turn, they call into question the notion of unambiguous belonging to a single national context, and open up the possibility of self-identifying as cosmopolitan rather than national. On the other hand, it also seems plausible that the experience of migration per se is already structured by basic habitual dispositions.

To accommodate these considerations—albeit without presuming specific habitus formations—we suggest breaking down habitus into a set of basic dispositions (Carlson and Schneickert, 2021) thereby partly following Lahire's (2003, 2011) critique of habitus. This conceptualization avoids a conception of habitus as a necessarily homogenous and unifying totality (see Silva, 2016) and emphasizes that specific dispositions may manifest quite differently (e.g., as cosmopolitan identifications) depending on the precise social context people are part of, thus following Bourdieu's habitus-field theory. Such an approach is particularly valuable for global comparative research given the vast diversity of societal conditions that may play a role in making people identify as cosmopolitan. But it also seems reasonable given the fact that Bourdieu himself (1984 [1979], 1990 [1980]) conceptualized habitus as simultaneously structured by class and other markers of difference (such as gender and migration).

Furthermore, there are three reasons why such a conceptualization is a viable option in the context of this study. First, it facilitates empirical measurement, construction and thus quantitative-empirical comparison in large-scale, country-comparative population surveys. Second, an in-depth exploratory clustering of local configurations of habitus types using grouping algorithms such as cluster analysis or latent class analysis would create new problems for large-scale cross-national comparative studies. By contrast, a decomposition into (gradual and simultaneous) habitual dispositions makes it possible to compare these in a standardized way, while also allowing for the possibility that specific habitual dispositions have different meanings and impact across societies. Third, while in Bourdieusian research methodology dimension-reducing reconstructions of social spaces (e.g., geometric data analysis such as multiple correspondence analysis) and grouping algorithms (such as cluster analysis) have been dominant, exploring specific dispositions allows the estimation of independent (including opposing, intervening, moderating, etc.) effects of habitus traits.

A number of large-scale surveys include items that are suited to measure such habitual dispositions, mostly based on psychometric measurement instruments, including values (Schwartz, 2012) or personality (McCrae and Costa Jr, 1999). However, to date, sociologists seem hesitant about using these instruments, as an unfortunate mutual disregard between sociology and psychology has allowed the belief to prevail that the concepts of habitus, values and personality are incompatible. This issue is only slowly being discussed and resolved in recent research (see e.g., Kaiser and Schneickert, 2016; Schmitz and Barth, 2019; Schmitz and Bayer, 2017; Schneickert et al. 2024; Schneickert, 2025). In the following, we argue that it is possible to assess habitual dispositions using psychological measurement instruments without making far-reaching assumptions about trans-situational relevance (in the case of values) or pre-dispositional genetic determinations (in the case of personality), or claiming universal, comprehensive, time- and culture-independent validity (in the case of both). Moreover, employing such measurements of habitual dispositions does not necessarily lead to the conclusion that values or personality instruments capture habitus in its entirety (see online appendix B for a further conceptual discussion on measuring dispositions with items from value research).

That said, for the exploratory purposes of this paper, it seems reasonable to initially fall back on those measurements that have been used in a variety of sociological studies on social space, habitus, and milieus (for an overview, see Schneickert, 2023; Schneickert, 2025). In particular, three value-based basic orientations have been distinguished: 1) a tradition orientation that is geared toward conservation of tradition, conformity, and security; 2) a status orientation that aims at (material) prosperity and success; and 3) an experience orientation that strives for hedonism and new, exciting experiences (Schneickert, 2025).

In doing so, we link sociological theories of value change based on a sequence of cultural modernization (for the orientations discussed here, see in particular Inglehart 1997: 77–78; Schneickert et al., 2024; Schneickert, 2025) and the measurement of human values in the Schwartz (2012) instrument (conservation, self-enhancement, stimulation/hedonism). However, the transition from materialism to post-materialism (according to Inglehart) does not completely coincide with the distinction made here between tradition (which would correspond to the measurement of security values such as “maintaining order,” etc.) and status (which would correspond to the measurement of classic materialistic values) and thus represents a refinement of the theory with a focus on the individual level of habitual dispositions.

In principle, it can be assumed that people with a habitual basic orientation geared toward tradition, security and conformity tend to be particularly oriented toward their national society and the security of the national welfare state, while an orientation toward status and experiences can represent two (post)modern dispositional routes to transcending the nation-state. Status seeking is manifested through a career orientation focused on individual advancement and success, whereas experience seeking is exhibited in openness to new experiences and a search for stimulation. However, this basic configuration may differ from country to country, both because nation-states guarantee security and welfare in different ways (i.e., they have different “strengths”) and because the cultural significance of these basic orientations and hence their connection to cosmopolitan attitudes may vary.

Thus, our third research question (RQ 3) is: To what extent and in what direction are basic dispositions—as indicators of habitus—related to global self-identification in different countries? Our general expectation is that traditional orientations tend to go hand in hand with national self-identification, while modern and postmodern orientations tend to correlate with global self-identification. However, opposing tendencies could be caused by the degree of security that welfare state regimes offer their citizens (for the security-oriented traditionalists), the economic opportunity structure (for the status-oriented) or the prosperity and space of experiences (for the post-materialistic experience-oriented).

Data and Methods

Data: the World Values Survey Time Series Dataset

We use data from the World Values Survey (WVS 2005–2015, waves 5 and 6; see Inglehart et al., 2022). The WVS is a unique dataset in several regards. First, and most importantly, it comprises a broad variety of countries across the globe, including non-Western countries that tend to be excluded from other large-scale comparative surveys. Second, waves 5 and 6 include short versions of the Schwartz values battery. Unfortunately, the value items were no longer surveyed in more recent waves of the WVS, limiting us to the older data (2005–2015), and preventing us from mapping the current renationalization tendencies and crisis dynamics of globalization. On the other hand, this fact also keeps the sample more homogeneous; with respect to the aforementioned period, we are mapping the peak phase of globalization at the beginning of the twenty-first century.

Due to the heterogeneity and number of participating countries, the WVS has a complex data structure, especially with regard to the comparability of indicators and missing information. With the construction of our measures, we are confident that we have achieved a reasonable tradeoff between comparability and the inclusion of a wide variety of countries.

After listwise deletion of 10.7% (n = 11,820) of individual cases due to missing values, our final sample consists of 74 countries with 98,670 individuals (for country details, see Table A1 in the online appendix). We use the country weight to compensate for small deviations according to sex, age, rural–urban domicile and education, but only for the descriptive presentation of means. We do not weight the regressions as we control for these characteristics. Wherever we use weights, this is mentioned explicitly in the table or figure.

Measures

Central Dependent variable: Global Self-Identification

A large proportion of existing quantitative research on cosmopolitanism has made use of a specific item included in the WVS (e.g., Bayram, 2015; Jung, 2008; Pichler, 2012; but see Haller and Roudometof, 2010 drawing on data from the International Social Survey Program, ISSP). In a battery of five items, people are asked how they see themselves in relation “to the world,” with one item stating, “I see myself as a world citizen,” and another stating, “I see myself as part of the [country name] nation,” both measured on a four-point scale ranging from 1 = strongly agree to 4 = strongly disagree. With this item, there are two main ways in which cosmopolitan self-identification is constructed, each depending on the specific conceptualization of identity underlying it. Quite straightforwardly, the item of identification as a citizen of the world can be used in isolation, although the distribution of data on the four-point scale sometimes implies certain statistical restrictions. Typically, either multinomial or ordered logistic regressions are used, or the item is dichotomized (see, for example, Pichler, 2012). However, focusing solely on this item ignores its relationship to the other item from the battery: identification with the nation-state. If identities are seen as consisting of various relational layers or dimensions, as is true when we employ the concept of habitus, then the measures chosen should capture global identification relative to national identification (e.g., Gorman and Seguin, 2018; Zhou, 2016). Objections to this argumentation are conceivable as well (see Gorman and Seguin, 2018; Jung, 2008), and we agree that identities can overlap, as stated, in fact, by cosmopolitanism theory (Pichler, 2012). Nevertheless, we consider an indicator of relational difference as a suitable ideal-typical way of operationalizing the competing logic of habitus and identities. Following Zhou (2016), we therefore measure individuals’ global self-identification gradually and in relation to self-identification with the nation-state by taking the difference of two scores, which then ranges from −3 (national self-identification) to 3 (global self-identification). ⁵ We add 3 to this difference so that the final measure is on a seven-point scale ranging from 0 to 6. A higher score indicates more intense global (relative to national) self-identification; see Table A2 in the online appendix for descriptive statistics of all relevant indicators.

Central Independent Variables: Habitual Dispositions

Empirical-quantitative habitus analysis is not particularly advanced, even compared to other Bourdieusian concepts such as forms of capital, which is surprising given the popularity of the habitus concept. However, recent theoretical studies have argued that habitus can be conceptually decomposed into dispositions (Carlson and Schneickert, 2021) and operationalized via indicators of personality or values (e.g., Kaiser and Schneickert, 2016; Schmitz and Barth, 2019; Schmitz and Bayer, 2017).

We therefore propose to measure basic dispositions of habitus with items from the Schwartz values battery in the WVS. For the exploratory purpose of this study, however, we only use those items that have been repeatedly considered in sociological studies of social space, class, and milieus, specifically measuring three basic habitual dispositions: tradition-, status-, and experience-seeking (Schneickert, 2025). Given the heterogeneity of the cultural contexts covered by our study, this seems complex enough, as extensive exploratory factor analyses (not reported here) indicate that neither the universal value nor personality or milieu structures are identified identically in all countries. The construction of the three suggested basic orientations is supported by a pooled factor analysis (Table A3 in the online appendix), which confirms a fundamental separation into tradition versus (post)modern orientations. Thus, we interpret the two factors in line with recent studies that place habitual dispositions in the context of cultural transformation processes (Schneickert et al., 2024; Schneickert, 2025), the latter consisting of two subdimensions that other studies have distinguished as status-seeking (achievement and power; see also Delhey et al., 2022) and experience-seeking (stimulation and hedonism). ⁶ As a consequence, we construct three basic orientations as unweighted mean indices ⁷ : tradition-seeking (from the Schwartz items: tradition, security, and conformity), (materialist) status-seeking (success and wealth), and experience-orientation (adventure and having a good time). Higher values indicate stronger agreement in the respective construct.

Central Independent Variables: Class Indicators

Following Bourdieu (1984 [1979]), we operationalize cultural and economic capital as two main indicators of class. High economic capital: The subjective income measurement in the WVS is a very rough indicator that asks people to place themselves on country-specific income groups. ⁸ This leads to quite heterogeneous distributions in the countries. To construct a simple and comparable measurement, we use a relative measurement across three quantiles (lowest third, middle, highest third of the distribution) that refers to relative economic position in each national context respectively and add a category for missing values; in each case, we focus on the effect of high economic capital compared to the reference group of low economic capital. High cultural capital: We measure high cultural capital via a dichotomous variable that compares individuals with tertiary (“high”) education to all other individuals. This approach circumvents the very different forms of educational qualifications below tertiary education in different countries. High education means tertiary, university education (with or without a degree) across all countries. ⁹

Central Independent Variables: Migration Background

We measure people's migration background via parental immigration history as a dichotomized variable that takes the value 1 if at least one parent immigrated from another country. We do not use information on the respondents own immigration status, because in too many countries there is missing information. Still, in many countries there is no or insufficient information (n < 50 or more than 15% missing on this variable) even on the proxy of parental immigration, so we estimate models without migration only for these countries (n = 27).

Control Variables

All models control for gender (1 = female), age (divided by 10 so that effects are 10-year differences), and religiosity (attendance at religious service). Given that it is difficult to compare very religious with very secular countries, we use a rough measure with three categories measuring frequency of attendance at religious events (0 = no information, 1 = once a year or less, 2 = once a month or more, with 1 as the reference category). In Qatar, Morocco and Kuwait, however, this question was not asked at all; for these countries, the models are estimated without religiosity.

Analytical Strategy

Our main analytical strategy is based on country-wise, multiple linear (ordinary least squares, OLS) regression models and comparing the explanatory power (r-squares) and unstandardized b-coefficients of the central independent variables (see Figure 1). We prefer to compare regression models separated by country rather than using a multi-level model for the following reasons: First, the scope of our paper is exploratory and focuses on the assumed heterogeneity of different countries in order to question the assumptions discussed in the literature, which to date is mainly based on evidence from “Western, educated, industrialized, rich, democratic” (WEIRD) countries. A single multi-level model could smooth out these differences. Second, comparing models allows for a more intuitive understanding of differences between national contexts, especially when comparing total explanatory power. Furthermore, the possibilities of graphical representation are simpler and easier to understand than the more abstract results of a multi-level analysis and therefore may encourage researchers from interdisciplinary and methodologically diverse backgrounds who are only interested in specific countries or regions to deepen their analysis on the basis of the presented findings. Third, the use of separate regression models allows us greater flexibility in modeling, as certain indicators are not applicable in some countries, which would imply either omitting the countries or the indicators (see comments on migration or religiosity above).

Figure 1.

Conceptual model.

We proceed in three steps. First, we examine the overall distribution of global self-identification and the explanatory power of our full models. Second, we examine the countries in the detailed country regressions with a focus on the marginal effects (unstandardized b-coefficients), that is, the association of the respective independent variable (class, migration, dispositions) on global-self-identification, while keeping the other independent variables and the control variables constant. Although it is true that in Bourdieu's theoretical framework class, migration and habitus (and others, cp. Bourdieu, 1984 [1979]: 106) are seen as inseparable, with respect to the comparative and explorative approach of this article, it seems important to first separate these concepts analytically in order to assess their effects individually. Only in this way can we understand and explore the overall configurations in different countries without starting out from Eurocentric assumptions (see Bhambra, 2010) about how we expect these combinations of migration, class, and dispositions to play out.

Findings

How Cosmopolitan is the World?

First of all, our data confirm a general finding of other studies (cp. Zhou, 2016: 162), using older WVS data: Populations in most countries prefer to identify with their nation rather than as world citizens. Our scale of global self-identification ranges from 0 (national relative to global) to 6 (global relative to national), with the theoretical midpoint 3 (balanced identification). As Figure 2 shows, in most countries, the population averages are well below this midpoint, the exceptions being Andorra and the Philippines. The pooled mean (unweighted) is around 2.5. This reminds us that although most people identify to some degree with different frames of identification, the nation state continues to play a major role in the identity of people around the world.

Figure 2.

Global self-identification by country.

Comparative Country Analyses

The explanatory power of our country regression models differs significantly, and this information is already quite informative in itself from a comparative perspective. Figure 3 shows the share of total variance in global self-identification explained by our full model (see Tables A4 and A5 in the online appendix for a pooled OLS and multi-level models of the full model, and Figure A1 for regression diagnostics of the OLS model across all countries). Figure 3 demonstrates that in some countries, global self-identification is quite well explained, while in others, it is not at all. Interestingly, the highest explanatory power is found in Germany and Switzerland, two countries in which the thesis of cosmopolitanism as a class or elite project is particularly prominent (see e.g., Delhey et al., 2015; Hartmann, 2020; Mau, 2010). Beyond that, however, no clear country patterns emerge, although WEIRD countries seem to show at least some degree of difference among social groups, and therefore tend to be found in the upper half of the figure.

Figure 3.

Explained variance in global self-identification by full model (per country).

Cosmopolitanism and Migration

For migration, we now compare the unstandardized b-coefficients from the full models. This means that differences in class, habitus and other sociodemographic factors between migrants and non-migrants are held constant so that the effect instead reflects the migration experience (of the parents) per se. ¹⁰

As Figure 4 shows, in most countries, individuals whose parents are migrants self-identify more globally than their fellow citizens. However, in Chile, Pakistan, Andorra, and Rwanda, the situation is exactly the other way around, as migrants seem to be more nationally oriented there. We cannot delve into the details of each country here, but it would certainly be interesting to explore more deeply the specific characteristics of the cultural frameworks and migrant populations that lead to these unique associations.

Figure 4.

Marginal effects of migration on global self-identification. Note: Unstandardized b-coefficients from full-model OLS country regressions. Black filled bubbles indicate significant effects at p < 0.05.

Cosmopolitanism as a Class Project

Next, we examine the two main indicators of social class: economic and cultural capital. Again, we interpret marginal effects, a particularly relevant issue here, since economic and cultural capital are positively correlated. We are therefore interested in the extent to which these forms of capital have independent and possibly different effects on global self-identification.

As can be seen in Figure 5, global self-identification is associated with economic and cultural capital in many countries, but the effects of both forms of capital differ and the overall picture does not allow us to conclude that cosmopolitanism is a class project everywhere. Indeed, while it may be the case in some countries (such as Germany, where this idea is particularly prominent in the social sciences), in others, either the effects are not significant, or they are even reversed (such as China, where high-income groups appear to be more nationally oriented). The relationship between cultural and economic capital is also revealing, as the two indicators by no means have the same effect in all countries. This shows that it is worth looking at the marginal effects of the two class indicators separately. Thus, some strong effects stand out, where the more highly educated are more cosmopolitan, as true of almost all wealthy industrialized countries, but not Japan and South Korea.

Figure 5.

Marginal effects of class indicators on global self-identification. Note: Unstandardized b-coefficients from full-model OLS country regressions. Black filled bubbles indicate significant effects at p < 0.05.

Overall, positive effects of cultural and economic capital on global self-identification are only present in two countries: Spain and Germany. Otherwise, the country groups do not overlap. Economic capital is associated with stronger global self-identification in Nigeria, Spain, Pakistan, Serbia, Ethiopia, Jordan, Germany, Kuwait, Uruguay, Armenia, South Africa, Malaysia, Libya, and South Korea. On the other hand, cultural capital (tertiary education), affects global self-identification in Germany, Spain, Switzerland, Cyprus, Finland, Turkey, Azerbaijan, Palestine, Norway, Australia, Russia, Poland, Sweden, and Canada. In addition, interesting opposing effects are apparent: In Switzerland and Norway, cultural capital is positively correlated, while economic capital is negatively correlated with global self-identification. The reverse is true in South Africa and Armenia, where economic capital seems to reinforce cosmopolitan identification, while the more highly educated tend to identify more nationally. Here, in-depth case studies would be worthwhile to explain these patterns in greater detail.

Cosmopolitanism and Habitual Dispositions

Finally, we analyze the effects of the basic habitual orientations, which we also consider separately, keeping the other basic orientations and class, migration and the control variables constant (Figure 6). First, with respect to tradition-seeking, in most of the countries analyzed here, more traditional individuals are more strongly tied to their nation-state, while those who are more open to new experiences and are more hedonistic identify globally instead. Second, status-seeking is not associated with global self-identification in most countries. However, where it is, it tends to be rather positive, meaning that people who strive for (materialistic) status and are ambitious are more likely to self-identify as cosmopolitan. Further, in some countries (Kazakhstan, Switzerland, Nigeria, and Iraq), status seekers rather tend to be oriented toward the nation. Third, experience-seeking is clearly associated with a stronger global self-identification: People who are more open to new experiences, more hedonistic, and more risk-taking and adventurous are more likely to self-identify as cosmopolitan.

Figure 6.

Marginal effects of habitual dispositions on global self-identification. Note: Unstandardized b-coefficients from full-model OLS country regressions. Black filled bubbles indicate significant effects at p < 0.05.

Considering the effect of different habitual dispositions, we can thus conclude that it is not the status-oriented elite who particularly drives cosmopolitanism, but rather the milieus of postmodern experiential orientation. ¹¹ These results largely confirm Bayram's (2015: 470) findings, which, following Schwartz's psychological theoretical framework, show that self-transcendence, self-enhancement, regard for diversity, and independence advance people's self-identification as world citizens, while interest in power and conservation values hinder this. Bayram has also confirmed country differences in the effects of values on cosmopolitanism (cp. Bayram, 2015: 470). By decomposing habitus into dispositions, our results indicate that it is specific dispositions that give rise to global self-identification.

Discussion and Conclusion

Using data from the World Values Survey (2005–2015), we have examined the effects of class, migration, and habitual dispositions on global-self-identification in 74 countries with nearly 100,000 respondents.

Conceptually, we proposed an operationalization of habitual dispositions via value items, and operationalized class via separate indicators of cultural and economic capital. This article thus also represents a global application of Bourdieu's theoretical framework, which ties in with current debates in globalization and migration research on transnational or cosmopolitan habitus. It is obvious that conceptual modifications are necessary when expanding the scope and analytical framework of a theory in this way.

Empirically, we have found that there are substantial differences across countries, and it can be assumed that the mechanisms either bringing forth or hindering global self-identification differ by country. This suggests a relational and comparative approach, as migration, capital, and habitus appear to operate differently in different national societies. Nevertheless, general patterns (albeit with some exceptions) can also be observed: First, the populations in almost all the countries studied here are, according to our measurements, more likely to self-identify nationally than globally, a finding that points to the continued importance of the nation-state.

Second, habitual dispositions are associated with cosmopolitan attitudes over and above the effects of migration, class, and sociodemographic factors. More traditional and security-minded individuals are less globally oriented in almost all countries; conversely, individuals who tend to be more hedonistic and adventurous are more globally oriented almost everywhere in the world. This varying effect of different habitual dispositions can only be observed thanks to our conceptualization of habitus as a set of dispositions and its ensuing empirical decomposition. However, this goes beyond Bourdieu and immediately points to a modified application of habitus-field-theory for comparative analyses of societies at the global level: it shows that similar or even identical dispositions can have different effects depending on the societal context. It is therefore not the “upper-class habitus”, the “migrant habitus”, or “transnational habitus” as a totality that determines outcome Y, but rather, depending on the context (societies, culture, class, networks, spaces, fields), different configurations of dispositions and societal contexts that correlate with a particular form of self-identification.

Third, we find effects of class position, with cultural and economic capital differing in their effects by country. Cosmopolitanism can therefore be described as an “elite project” in some countries, but not in all, and in most countries, the cultural and economic upper-class fractions differ in their global self-identification. This finding thus cautions against simply generalizing Western ideas about cosmopolitanism as a “class project” to the global level, corroborating earlier findings by Haller and Roudometof (2010) who pointed out Europe's particularity in terms of cosmopolitan attitudes compared to other industrialized countries across the world.

Fourth, we find that migration background (i.e., parental migration) is associated with higher global self-identification in many countries, independently of class position, habitus patterns, and sociodemographic characteristics.

The present study is not without limitations. In the methods section, we argued that while the WVS offers unique opportunities, the price of being able to compare such diverse countries often lies in rather crude indicators. The country sample is highly heterogeneous, which is precisely what we considered interesting, yet we lack other rich European countries (for example) to confirm some of the assumptions regarding geographic patterns. Methodologically, we decided to use an exploratory-comparative design to allow sufficient space for country differences. Now that the relationships have been laid out in an explorative and descriptive manner, multi-level models with possible cross-level interactions (see Gorman and Seguin, 2018; Zhou, 2016) may be appropriate to explore the micro- and macro-level relationships of class, migration and habitus with sociodemographic characteristics in greater detail. Geometric data analyses in the tradition of Bourdieu could help to focus not only on marginal effects, but also on socio-spatial proximity within countries or even transnational spaces, without presupposing a strict separation of independent and dependent variables. In addition, detailed country analyses and, in particular, in-depth case studies seem useful to elaborate on some of the counterintuitive cases we have identified.

Conceptually, we have also made a contribution to the empirical operationalization of habitus, suggesting a need to focus on habitual dispositions rather than transnational, cosmopolitan or global habitus as such, and to measure these using psychometric constructs (such as values and personality). ¹² We hope that this will stimulate further research elaborating the relationship between habitus, values, and personality, and in particular to explore it empirically and in international comparison. We believe that such empirical research also holds enormous potential for theory development as it allows for examination of potentially opposing or moderating effects of individual dispositions.

In conclusion, our analyses show that, despite profound globalization and transnationalization processes, it would be premature to assume that habitual dispositions have the same effect in different societies. Although there are indeed certain general tendencies and associations between forms of capital, migration experiences, habitual dispositions and global self-identification, these have rather different effects in different societies. For this reason, one-size-fits-all notions such as “transnational,” “global,” and “cosmopolitan” habitus should be used with caution, as they tend to gloss over the actual empirical variety of the configuration of dispositions and contexts. With these insights, we hope to stimulate future research that follows this path with more refined methods and more comprehensive data. Our study has focused on the main phase of globalization at the beginning of the twenty-first century. It will be particularly exciting to see how cosmopolitan attitudes have evolved during the numerous crises and renationalization trends that have occurred over the past decade.