Schemas of the Nation in Modern Democracies

This study develops a novel analytical approach for studying popular conceptions of the nationstate that accounts for both withinand between-country heterogeneity and avoids a priori assumptions about the national boundedness of culture. I identify widely shared attitudinal patterns among a pooled sample of over 27,000 respondents from thirty countries and only subsequently examine those respondents’ national affiliations. Having established the robustness of the resulting four-fold typology of nationalist beliefs using multiple strategies, including outof-sample replication, I relate these cultural schemas to the respondents’ political beliefs. The results reveal four characteristics of nationalism in settled times: 1) meanings attributed to the nation are far more heterogeneous than is suggested by existing theories; 2) the same four cultural schemas of the nation are found in all countries, though their relative prevalence varies; 3) the content—but not the distribution—of the schemas is stable over time; and 4) schemas of the nation are highly predictive of other political attitudes. The paper makes a substantive contribution to research on political culture and offers a general analytical approach for the comparative study of collective identification. The symbolic meanings attributed by individuals to their nations are associated with a variety of outcomes of interest to sociologists and political scientists. Restrictive conceptions of a nation’s social boundaries, strong collective identification with the nation, high levels of national pride, and feelings of national superiority, for instance, have been linked to heightened levels of ingroup favoritism, out-group prejudice, and support for authoritarian politics (Blank, Schmidt, and Westle 2001; Ceobanu and Escandell 2008; Kunovich 2009; Schatz, Staub, and Lavine 1999). These attitudinal associations have been shown to influence voting and support for public policies, particularly when the nation is made salient in political discourse (Sears 1993; Citrin et al. 1990). Historical studies have also demonstrated that political elites’ favorability toward Bonikowski Schemas of the Nation 2 exclusionary public policies has been historically shaped by idealized conceptions of the nation’s character (Smith 1997). Despite the proliferation of research on contemporary nationalism in settled times—with nationalism understood broadly and inclusively, as the heterogeneous domain of dispositions and practices that reflect and reproduce the primacy of the nation-state as a fundamental unit of identification and governance (Brubaker 1996, 2004)—the theoretical contributions of this field remain fragmented. This research tends to either privilege practice-based approaches, which emphasize the richness and contextuality of multilayered ethnic and national identities at the cost of empirical generalizability (Fox and Miller-Idriss 2008), or variable-based survey research, which focuses on individual attitudes without considering how such attitudes may aggregate into overarching cultural worldviews (e.g., Kunovich 2009; Hjerm and Schnabel 2010). Neither approach has generated a systematic typology of popular conceptions of the nation-state that can enable comparative research on the causes, consequences, and changing form of contemporary nationalism (Dekker, Malová, and Hoogendoorn 2003). In contrast, an older research tradition on the essential content of national identities did catalogue distinct varieties of nationalism (Kohn 1944), but its impact was limited by a reductive understanding of culture that ignored the multivocality of the nation’s meanings within each national community (Smith 1997). In this paper, I seek to build on these three approaches by developing a systematic typology of cultural schemas of the nation-state among 27,790 respondents from thirty countries. In particular, I draw inspiration from practice-based nationalism research (Fox and Miller-Idriss 2 It is important to note that I do not define nationalism here as a coherent elite ideology (as does Gellner [1983], for instance) or as a specific set of normative attitudes toward the nation (as do scholars of chauvinism [e.g., Kosterman and Feschbach 1989]). Instead, I favor Brubaker’s (1996:10) view of nationalism as a “set of ‘nation’-oriented idioms, practices, and possibilities that are continuously available or ‘endemic’ in modern cultural and political life,” the normative content of which can vary widely. 3 My use of the term “nation-state” does not imply a political territory that is ethnically homogeneous (i.e., a national state). Instead, it highlights two related objects of collective identification: the sociocultural community Bonikowski Schemas of the Nation 3 2008), which views the meaning of collective identities as variable within national communities, and from survey-based research (Kunovic 2009), which makes distributional claims about such attitudinal variation. While my analytical strategy also makes use of distinctions made in classic research on national character, I steer clear of some of the limitations of this tradition by avoiding a priori judgments about the national boundedness of culture. Finally, in contrast to both survey-based and national character research, I treat cultural schemas of the nation as primary units of analysis. This distinction is a subtle but important one: the objective is not to correlate disembodied attitudes or to argue for fundamental cultural differences between countries, but to map communities of shared meaning, whose members view their nation-states through similar interpretive lenses and whose boundaries may not be congruent with national borders (Fine 1979; Zerubavel 1999; Eliasoph and Lichterman 2003; Goldberg 2011). In so doing, the approach offered here privileges persons over variables, polysemy over singularity of meaning, and relational configurations of beliefs over individual attitudes (Abbott 1988, 1992). The results demonstrate that cross-national differences in popular conceptions of the nation-state are best understood in terms of the relative salience of multiple cultural models within countries rather than in terms of essential country characteristics (cf. Lamont and Thévenot 2000). Respondents who subscribe to these distinct cultural models have contrasting attitudes on social and political issues, a finding that attests to these models’ substantive importance for politics. Furthermore, the content of the cultural schemas that define these groups is remarkably stable, as evidenced by an out-of-sample replication performed with comparable data collected eight years earlier. 4 This is not to say that a variable-based approach is not useful for demonstrating the robustness of the cultural schemas that define such communities of shared meaning. Indeed, in this paper, I use regression analyses to demonstrate that other politically relevant beliefs are systematically distributed across the observed cultural communities. Bonikowski Schemas of the Nation 4 RESEARCH ON NATIONALISM IN SETTLED TIMES Much nationalism research has focused on those aspects of the phenomenon that have been the most destabilizing for existing social and political institutions: the emergence of collective identification in newly forming nation-states, the struggle for statehood by separatist, irredentist, and post-colonial nationalist movements, and the periodic eruptions of nativism and xenophobia in otherwise stable societies. There is no doubt that such work has made essential contributions to the scholarly and popular understanding of the contemporary social and political order. Yet, for all its rich insights, the nationalism literature has suffered from an analytical blind spot, stemming from the implicit assumption that in the absence of explicit conflict, nationalism in established democracies is simply a fait accompli, rather than a constitutive frame of reference that continually shapes the course of social and political change. Over the past twenty years, however, an emerging scholarly tradition has challenged this position. Focusing on nationalism in settled times, this research has asked two sets of questions: (1) how do people use the nation in everyday practice to make sense of the world, maintain a collective identity, and facilitate social interaction? and (2) how do attitudes toward the nation vary within national communities and how are they associated with other beliefs? Both approaches are predicated on the pervasiveness of a taken-for-granted cognitive and affective orientation that reproduces the primacy of the nation-state in everyday life (Billig 1995; Brubaker 2004; Collins 2012), but the former privileges the contingency of micro-level processes embedded in rich social contexts, while the latter abstracts from such contexts in an effort to map attitudinal variation in nationally representative samples. In advancing a constructivist understanding of culture, these two modes of research differ sharply from Bonikowski Schemas of the Nation 5 antecedent research on nationalism in settled times, which sought to specify the core values that ostensibly define each country’s fundamental political identity. Methodological Nationalism and Reductive Individualism in Nationalism Research Influenced by a functionalist understanding of culture as a coherent system of consensus values, traditional research on nationalism in settled times assumed that each nation has an essential character—a set of principles that defines the collective identity of its people and the logic of its institutions. The task of the analyst was then to uncover these principles and, often, to compare them to the identities of other nations. Thus, for instance, Lipset (1990) argued in Continental Divide that national identity in the U.S. embodies the core features of the American Creed: antistatism, individualism, populism, and egalitarianism. In contrast, national self-understanding in Canada, he argued, is based on deference to authority, collectivism, elitism, and particularism (also see Huntington 2004). The primary legacy of the macro-level approach to nationalism research came in the form of a distinction between ethnic and civic nationalism that originated in the writing of Meinecke ([1907] 1970) and became popularized by Kohn (1944). This binary typology, which continues to be salient in a variety of forms, focuses on one of the most salient features of national identity, namely the principles that determine legitimate national membership. Ethnic nationalism is based on ascriptive criteria, such as race, ethnicity, ancestry, language, and in some cases religion, while civic nationalism is based on elective criteria, such as commitment to the country’s values, subjective identification with the national community, and respect for the nation’s laws and institutions. In keeping with the view of culture as a collective attribute of social groups, this dichotomy has been used to classify countries and world regions into two distinct ideological Bonikowski Schemas of the Nation 6 camps, with Germany epitomizing ethnic nationalism and France and the United States exemplifying civic nationalism (Kohn 1944). Despite its merits, the problem with the classic research on political culture was its tendency toward “methodological nationalism”, that is, a taken-for-granted view of the nationstate as a naturally bounded cultural unit of analysis (Calhoun 1999; Wimmer and Schiller 2002). Even though the ethnic-civic typology has become less rigid in its more recent applications, the assumption that nation-state borders provide natural bounds for homogenous nationalist ideologies persists in contemporary comparative research (e.g., Smith 1991; Ignatieff 1993; Schopflin 1995). As Brubaker (2004) has persuasively argued, such “groupism”—the conflation of analytical categories (e.g., the nation-state) with empirical groups (e.g., a community with shared beliefs)—leads to theoretically problematic and empirically inaccurate conclusions. In reality, national populations comprise multiple competing belief systems and the tension between them is an important driver of political change (Smith 1997). Whether those cultural repertoires are similar or distinct across nations should be a matter for empirical investigation, not an untested assumption. Furthermore, such beliefs systems can vary over time, both in their content and in their relative dominance in any given country, a fact that is largely ignored by traditional nationalism research. Nonetheless, when stripped of its essentialist trappings, the ethnic-civic distinction produced by this literature may be useful for classifying subsets—though not the full range—of attitudes that constitute popular understandings of the nation-state. In contrast to traditional scholarship on national character, more recent research has sought to challenge the notion that the meaning of the nation is consistent in content and stable in salience across the national community. Rather than treating national political culture as a set of widely shared values, sociologists have stressed the continent nature of cognitive and affective Bonikowski Schemas of the Nation 7 dispositions toward the nation, as they are put into practice in everyday life (Brubaker 2004; Brubaker et al. 2007). The question for these scholars is not what constitutes a particular country’s essential national identity but what use do people make of the national frames of reference as they navigate their social world—that is, under what circumstances and to what ends does thinking, talking, and acting with the nation become relevant in specific social contexts (Fox and Miller-Idriss 2008). In so doing, this work has shown that the meaning of the nation is itself highly variable, not just across situations but also across individuals. The microinteractional orientation of this tradition, however, along with its tendency to rely on ethnographic and interview-based methods, has precluded it from systematically mapping and classifying the variation in popular understandings of the nation. If classic studies of national political culture have tended toward methodological nationalism and practiced-based approaches have privileged situational particularity, quantitative analyses of survey data in sociology and political psychology—though attentive to withincountry heterogeneity—have been prone to reductive individualism. The units of observation in such research are individuals, but the mode of analysis is variable-based: the regression models used in this work abstract specific attitudes from concrete persons and correlate those attitudes with other variables of interest, including anti-immigrant sentiments (Citrin et al. 1990), authoritarianism (Schatz, Staub, and Lavine 1999), support for aggressive foreign policy (Kosterman and Feshbach 1989), and skepticism toward supranational institutions (Lubbers and Scheepers 2007). This analytical approach does not consider the intersubjective dimensions of culture that produce patterns of shared meaning within groups of likeminded individuals belonging to distinct thought communities (Zerubavel 1999). 5 To be clear, Zerubavel’s (1999) concept of “thought communities” does not imply that people with shared cultural understandings constitute “classes-for-themselves.” Likewise, despite my objective of identifying common patterns Bonikowski Schemas of the Nation 8 Thus, while survey-based research has generated a rich set of results that demonstrate the political relevance of and within-country variation in attitudes toward the nation, it has not been able to produce a set of conceptual tools that lend themselves to meaningful aggregation beyond the individual level and—if we are to understand common conceptions of the nation in more general terms—that enable systematic cross-national comparison. Developing such tools is one of the central objectives of the present paper. A systematic typology of the varieties of cultural frameworks through which individuals understand their nation-states can serve as a foundation for a more complete understanding of the relationship between nationalism and political change, from electoral outcomes and policy reforms to social movement mobilization. Attitudes, Cognitive Schemas, and Shared Representations A core assumption of the present study is that a typology of nationalist beliefs should rest on realistic assumptions about the cognitive foundations of meaning structures (DiMaggio 1997). Survey studies in this field have tended to emphasize specific attitudes (typically measured by specific variables), such as identification with the nation or pride in the nation’s heritage, to the exclusion of others, thereby assuming that the meaning of such attitudes is self-evident. Yet, a long tradition in cultural sociology, influenced by structural linguistics (Saussure 1916 [1960]), has demonstrated that symbols (and other cultural objects) derive their meaning from their relationship to other symbols (and objects) and not from their own inherent properties (Emirbayer 1997; Mohr 1998). Consequently, to understand symbolic structures—that is, of dispositions toward the nation, I do not assume that such patterns characterize groups of individuals with shared experiences, common political goals, or even conscious awareness of cultural commonality. 7 In keeping with a domain-based definition of nationalism (Brubaker 1996), I use “nationalism” and “nationalist beliefs” as shorthand for the wide range of popular dispositions toward the nation without implying any particular normative content of those dispositions. This assumes that most people employ the nation as a frame of reference in their everyday lives at least some of the time, even if the meanings they attribute to it differ markedly. This is consistent with Billig’s (1995:6) argument that “nationalism, far from being an intermittent mood in established nations, is the endemic condition.” Bonikowski Schemas of the Nation 9 culture—we must take into account their constituent parts, as well as the relations of similarity and opposition among those parts. A particularly useful analytical tool for thinking relationally about meaning making in specific cultural domains is the concept of a cognitive schema, which originates in cognitive psychology (Fiske and Linville 1980) and has made its way into cultural sociology (DiMaggio 1997). Cognitive schemas (or schemata) are networks of association that impart coherence and order onto the messy and rapid flow of sensory information to which individuals are exposed in their daily lives. In addition to organizing and interpreting lived experience, schemas feature affective and evaluative components that enable individuals to respond to stimuli in a manner consistent with their past experience and future aspirations. The schematic processing of information occurs rapidly and without much deliberation (Lieberman et al. 2002). As it is typically used in cognitive psychology, however, the cognitive schema concept underemphasizes the fact that meaning structures are produced in social interaction and institutionalized through collective narratives, rituals, and symbolic practices (Collins 2004). Indeed cognitive schemas are most appropriately thought of as individualized instances of overarching shared representations of particular domains of social life (Durkheim [1895] 1982; Moscovici 1984; Thompson and Fine 1999). Such representations are shared because they reflect collective belief patterns found within thought communities and they are representations because they consists of relational systems of symbolic elements that give meaning to social experience. To emphasize these intersubjective and relational features of belief systems, I refer to the observed patterns of meaning as cultural schemas. This paper is concerned with a specific type of cultural schema, namely that through which people give meaning to their membership in imagined communities defined and governed 8 For stylistic reasons, I use the terms “cultural schema” and “cultural model” interchangeably. Bonikowski Schemas of the Nation 10 by formal institutions. Whether such groups are nation-states, religions, professions, or federated social movements, they share a core set of features that can be systematically studied using the approach suggested here: they serve as objects of collective identification (Tajfel and Turner 1979), they are delineated by social and symbolic boundaries that determine group membership (Lamont and Molnár 2002), they are dually constituted by both their membership and their institutional governance structure (Gellner 1983), their members feel a sense of mutual belonging, despite the lack of face-to-face contact (Calhoun 1997), and they function within a broader social field populated by other groups against which they vie for resources (Fligstein and McAdam 2011). Crucially, group members often disagree, whether actively or not, about the meaning of the group, with important implications for the group’s internal dynamics and relations with external actors. What this paper seeks to demonstrate is that the underlying heterogeneity in meaning can be systematically studied by identifying competing cultural schemas using inductive survey analysis methods. Which Attitudes Are Relevant for Measuring Cultural Schemas of the Nation-State? Scholars of nationalism in political psychology, where survey-based research on the topic is most prevalent, study a wide variety of attitudes, often treating subsets of them as constitutive of nationalism as a whole (as opposed to patriotism, for instance). Setting aside the resulting definitional disagreements, it is possible to identify four broad attitudinal groupings that have received the most scholarly attention in this work—and that feature in widely available cross9 Given that the cultural schemas of interest in this paper concern the meanings attached to the collective community by its members, they are a variant of what can be termed “collective identification schemas." Because “collective identification schemas of the nation-state” is a rather awkward formulation, however, I refer to these meaning structures as, interchangeably: cultural schemas of the nation(-state), cultural models of the nation(-state), shared representations of the nation(-state), or more generically, popular conceptions or popular understandings of the nation(-state). Bonikowski Schemas of the Nation 11 national survey data. In this paper, I will jointly consider all of these attitudes and examine how they systematically co-vary within subsets of respondents with shared attitudinal profiles. For some political psychologists, nationalism corresponds to the identification with the nation over and above other collective entities (Sidanius et al. 1997; Li and Brewer 2004; Huddy and Khatib 2007). A standard (if imperfect) survey item used to capture this sentiment asks respondents to rate how close they feel to the nation. These responses are then typically compared with degree of identification with suband supra-national communities. For others, the most important feature of nationalism consists of the membership criteria that define the nation’s symbolic boundaries, with a particular focus on elective (or civic) and ascriptive (or ethnic) characteristics (Jones and Smith 2001; Hjerm 2003; Ceobanu and Escandell 2008)—an approach that adapts the binary typology from macro-level research on national character to individual-level attitudes. Survey items typically ask respondents to rate the importance of ancestry, birth in the country, language, religion, subjective identification, and other criteria for being a “true” member of the nation (e.g., a true American). The third approach focuses on the specific aspects of the nation-state that are positively evaluated by respondents (De Figueiredo and Elkins 2003; Parker 2010; Green et al. 2011); some label this patriotism, but I will refer to it as national pride in order to avoid unnecessary normative connotations. Survey measures of this dimension of nationalism typically ask respondents how proud they are of the nation’s achievements in domains like sports, science, democratic rule, or the egalitarian treatment of groups. Finally, the fourth dimension of nationalism consists of comparisons of the nation and its people with the rest of the world (Kosterman and Feshbach 1989; Blank and Schmidt 2003; Huddy and Khatib 2007). Scholars often interpret excessively boastful comparisons of this sort Bonikowski Schemas of the Nation 12 as constitutive of nationalism in general, but I prefer the more precise labels of chauvinism or national hubris. Standard measures ask respondents about their preference for their country over others, whether the world would be better if everyone was more like the country’s residents (e.g., more like Americans), along with a variety of other evaluative statements. It is worth emphasizing that these four areas of research capture different dimensions of a single overarching phenomenon: the repertoire of intersubjective meanings through which contemporary nation-states are understood and reproduced. My analysis incorporates indicators of all four dimensions of nationalism, because the meaning attributed by respondents to each indicator is likely to depend on their overall relational configuration. DATA Given that cultural schemas are held at the individual level but shared across subsets of a given group’s members, it is possible to study them using survey data. Survey data have the dual advantage of facilitating systematic comparisons across respondents that can yield patterns of broadly shared beliefs and enabling inferences about the distribution of the resulting cultural models in the population of interest. Because the objective of this paper is to measure the heterogeneity of nation-state schemas within national populations and compare the resulting patterns across countries, survey data are the most appropriate choice for the task. In particular, the analyses in this paper rely on data from the 1995 and 2003 National Identity Supplements to the International Social Survey Programme (ISSP), a representative multinational survey administered independently in each participating country (in the U.S., it is 10 Although hubris is often defined as excessive pride, there is an important distinction between the two concepts, in that pride can be self-oriented while hubris necessarily involves a comparison of one’s attributes to those of others. In fact, in the original Greek, hubris was a legal term referring to the act of exercising power through the shaming of a victim, often by way of sexual violence (Cohen 1991). 11 The limitations of survey data for studying meaning are of course well known and should be kept in mind when interpreting the results of this paper. For a recent critique of interview-based methods, including surveys, see Jerolmack and Khan (2014). Bonikowski Schemas of the Nation 13 incorporated into the General Social Survey). The survey probes respondents’ attitudes on a variety of topics and features a wide selection of sociodemographic covariates. The Aspects of National Identity II supplement was administered in 34 countries between 2003 and 2005. Residents of former East and West Germany were sampled separately, as were Israeli Jews and Arabs, bringing the number of separate samples to 36. For the purposes of the analysis, the East and West German samples were combined using the appropriate sample weights and four countries were excluded: Bulgaria, Latvia, and Israel were dropped because their questionnaires omitted a number of nationalism items and Taiwan was excluded because of insufficiently complete covariate data. In the interest of comparability, respondents under 18 and over 65 years of age were omitted from the analyses, as were non-citizens and respondents with missing values on more than two nationalism items. The final sample size consisted of 27,790 observations from 30 countries, with an average of 926 respondents per country. In addition to analyzing the 2003 sample, the paper features an out-of-sample replication that makes it possible to examine change over time in the content and distribution of nationalist cultural schemas. To do so, I take advantage of a previous wave of ISSP data collected in 1995 as part of the ISSP National Identity I supplement. The 1995 data feature 20 countries, with a total sample size of 18,613 and country samples ranging from 608 to 1,767 respondents. The replication analysis relies on the 1995 data along with a subset of the 2003 sample restricted to the 20 countries that were polled in both years. The reduced 2003 sample consists of 17,574 respondents. The two national identity supplements include twenty-six indicators of national identification, membership criteria, national pride, and national hubris, which are listed in detail in Appendix A, and all of which will be used to identify the cultural schemas of the nation. The Bonikowski Schemas of the Nation 14 variables were recoded so that higher scores correspond to stronger national identification, greater importance attributed to each criterion of national membership, greater degrees of pride, and greater levels of hubris. The observations were weighted using individual-level sampling weights provided by the ISSP, as well as population weights to ensure that all countries contributed equally to the model estimation. The data also include a variety of sociodemographic covariates and indicators of other social and political attitudes. The former will serve as controls, while the latter will be used to examine the association between the cultural schemas of the nation and other sets of beliefs. MEASURING CULTURAL SCHEMAS The primary challenge in identifying cultural schemas using survey data lies in the selection of appropriate analytical methods. Given the theoretical approach to culture employed in this paper, the method of choice must meet the following requirements: it must be relational because the meanings it is intended to capture are themselves inherently relational (Mohr 1998); it must be person-centered rather than variable-centered, so that the respondent-level dependencies between attitudes are preserved (Muthén and Muthén 2000); it must allow for the aggregation of individual-level data into discrete clusters defined by shared patterns of survey responses, so that distinct thought communities can be identified; and it must enable the classification of respondents into those clusters so that the distribution of the resulting cultural models can be examined within and across groups (in this case, within and across countries). Those requirements are partially met by a variety of relational and combinatorial methods, ranging from semantic network analysis (Carley and Palmquist 1992) and qualitative 12 The logic of the last two objectives is similar to that employed in optimal matching (Abbott and Hrycak 1990) and other clustering techniques: first identify overarching patterns in complex data and then match observations to those patterns to obtain a pattern-by-observation classification. Bonikowski Schemas of the Nation 15 comparative analysis (Ragin 1987) to hierarchical clustering and multidimensional scaling (Breiger, Boorman, and Arabie 1975). One approach in particular, however, satisfies all four criteria: latent class analysis (LCA), a well-documented survey analysis method that groups respondents based on the similarity in the pattern of responses to multiple survey items (Lazarsfeld and Henry 1968; Goodman 1978). LCA has been used in a variety of fields, from medicine and marketing to the sociological study of cultural consumption (Van Rees, Vermunt, and Verbrood 1999). Cultural sociologists, however, are only beginning to realize its utility for mapping cultural schemas (DiMaggio and Bonikowski 2008; Gross 2013; cf. Goldberg 2011). Latent class analysis, a variant of structural equation modeling, estimates one or more unobserved categorical (latent) variables by modeling the associations between a set of observed indicators. Like factor analysis, it is often used as a data reduction technique, but factor analysis produces continuous latent variables, while LCA generates categorical classes that capture distinct patterns of survey responses. This unique feature allows LCA to classify observations into distinct response sets, which—in the context of attitudinal surveys—can be interpreted as clusters of respondents with similar cultural understandings of a particular social domain. As such, LCA is an ideal method for developing empirically grounded conceptual typologies (McCutcheon 1987). Formally, a latent class model with four indicators can be represented as follows (Magidson and Vermunt 2004): π!"#$% = π!π!" π!" π!" π!" !|! (1) In this notation, X is the latent categorical variable with t classes and A through D are the observed indicators with i through l response categories. Consequently, π!is the probability of Bonikowski Schemas of the Nation 16 an observation belonging to each latent class, while π!" is the probability of a response i to variable A conditional on membership in latent class t. In estimating the model parameters, LCA uses iterative maximum likelihood estimation to best approximate a solution where the indicators become conditionally independent of one another—that is, any residual associations between them can be attributed to chance alone—within each level of the latent variable. The algorithm then calculates a posterior probability of class assignment for each observation in the data set and classifies the observations as belonging to the class with the modal posterior probability. If we analyze pooled data from multiple countries, we can use LCA to inductively generate groups (i.e., classes) of respondents who share similar response patterns—that is, similar ways of understanding the nation-state—regardless of their national affiliations. Having generated the latent class structure, we can then ask subsequent questions, such as what predicts class membership, how is class membership related to other political attitudes, how are the classes distributed across and within countries, and how does the content and distribution of the classes change over time. The nestedness of the individual observations within countries, however, violates the assumption of independent errors in the regression models used to estimate the latent classes. To account for this, a country-level variable is added as a covariate to the original LCA model. The probability structure of the extended model is restricted so that the country clustering of individual observations affects the distribution of classes within each country, but it does not affect the response probabilities of particular indicators within each latent class. This model, known as the partial homogeneity model, is formally expressed as follows (with G indicating the country grouping variable): π!"#$%|! !"#$%|! = π!|! π!" π!" π!" π!" !|! (2) Bonikowski Schemas of the Nation 17 The same model can be further extended by including individual-level sociodemographic variables as additional covariates. These covariates improve the ability of the model to assign observations to classes, but they do not affect the class content. In the present analysis, I include age, education, partisan ideology, gender, marital status, religiosity, urban residence, and parental immigration status as covariates. All of the variables have significant effects on the classification of observations into latent classes. THE FOUR CULTURAL SCHEMAS OF THE NATION When estimating latent class analysis models, it is up to the analyst to decide how many classes should be identified. This decision is typically based on theoretical grounds, as well as on measures of goodness of fit, such as the Bayesian information criterion (BIC). My analyses of the ISSP data consistently demonstrated that a four-class solution appropriately captures the cross-national variation in nationalist attitudes. This conclusion is based on a wide range of detailed robustness checks, including comparisons of BIC scores and classification errors across models, comparisons of results across models with a variety of classes, and a replication of the analysis with both a subset of the 2003 data and a completely independent survey sample collected in 1995. For more details on model fit, see Appendix B. The replication analysis will be presented in the final section of the paper, because its implications are both methodological and substantive. To interpret the four types of nationalism, it is possible to examine how respondents in each class structure their understandings of the nation-state. LCA calculates the probabilities of specific survey responses conditional on class assignment; based on those probabilities, it then predicts the distribution of responses to each nationalism measure in every class. By examining these predicted responses, we can get a sense of the attitudinal profile of each class (or what I Bonikowski Schemas of the Nation 18 refer to as class content). Given that there are twenty-six variables, each of which has between four and five response categories, the easiest way to compare the classes is to use variable means. The class means for the twenty-six nationalism measures are presented in Figure 1. To aid in the interpretation of the results, I have labeled the four classes as liberal, disengaged, restrictive, and ardent. Class 1: Liberal nationalism. The defining characteristics of the first class, which comprises 36 percent of the sample, are its moderate scores on the identification, membership, and hubris variables, combined with a high degree of pride in all domains of the nation-state. Respondents in this class feel close to their region and country and are ambivalent about their identification with their continent. Their notion of who is a legitimate member of the nation tends toward civic nationalism, with more emphasis placed on elective criteria, like respect for institutions and laws, language ability, and subjective feeling than on ascriptive criteria, like religion, ancestry, and birth. The pattern of responses to the pride items stands in contrast to the moderate values on the identification and membership criteria variables: members of Class 1 exhibit a high degree of pride in all aspects of the nation-state, which is only surpassed by Class 4. Finally, members of Class 1 exhibit moderate attitudes on measures of chauvinism, shame, and unconditional support for their countries. Because the model of the nation-state espoused by members of this class consists of moderate identification with the nation, relative open-mindedness about the nation’s social boundaries, and a fairly strong sense of pride in the nation-state’s accomplishments without 13 The appropriateness of reporting mean values for ordinal data has been debated because the distances between the individual categories may not be equal. I use means here in the interest of parsimony. Having compared each mean value with the underlying variable distribution, I am confident that the means satisfactorily capture the response patterns found in each of the four LCA classes. 14 It is important to remember that the values shown on the graphs represent central tendencies. Consequently, given that most of variables were measured using a forced-choice four-point scale, the 2.5 mark on the graph represents not an individual’s lack of agreement or disagreement with a particular survey question, but an underlying distribution of positive and negative responses among all the respondents assigned to that class. Bonikowski Schemas of the Nation 19 Figure 1. Variable means by latent class, ISSP 2003. a Id = Identification; Mem = Membership Criteria, Prd = Pride, Hub = Hubris 1.0 2.5 4.0 A): Country