Who Supports Global Cooperation? Cooperative Internationalism at the Intersection of Social Class and Economic Development

Throughout the twentieth century, the world has seen a rapid increase in global social, economic, and political integration. According to many studies, attitudes toward international organizations and international cooperation have also grown more positive, particularly among elites and in the affluent, densely connected countries of the global core. Using survey responses on 18 different questions from six cross-national attitude surveys, we find that “cooperative-internationalist” attitudes, though widely popular, are no more common in the global core than on the periphery. Furthermore, we find elites are more likely to hold proglobal attitudes than non-elites only in wealthy core countries. These results indicate that scholars may have incorrectly assumed that (modest) class differences in cooperative-internationalist attitudes in Western countries generalize globally, both within and between countries. We conclude with a call to theorize cooperative internationalism as a function of how different groups of people interpret their own costs and benefits of global cooperation.

T HE dramatic global catastrophes of the first half of the twentieth century sparked the creation of an international world order characterized by increasing cooperation, trade, and communication between countries and international organizations (Federico and Tena-Junguito 2017;Meyer et al. 1997). Since then, the widespread acceptance, and even celebration, of international cooperation has been an integral component of this system (Rathbun 2020;Rathbun et al. 2016). Yet, substantial numbers of people hold reservations about global cooperation as evidenced by the recent backlash against "globalism" in countries around the world (Bonikowski et al. 2019), and we still know little about what produces cooperative-internationalist attitudes. Moreover, as we show here, existing theories of proglobal attitudes are probably mistaken. We ask: where and why do cooperative-internationalist attitudes take hold?
Most studies see cooperative-internationalist attitudes as a result of a particular structural location. These studies emphasize factors that expose individuals to foreigners and global cultural scripts-such as urban residence, wealth, education, national-level economic development, and their nation-state's position in global networks-as determinants of holding cooperative-internationalist attitudes (Brett and Moran 2011;Calhoun 2002;Inglehart 1997;Sassen 2001). On the other hand, the poor, people in rural and economically disadvantaged areas, and populations in developing countries are seen as more likely to view international cooperation and the institutions that promote it as threats to their livelihood and way of life (Beck 2011;DiGrazia 2017;Hainmueller and Hiscox 2010;Hurwitz, Peffley, and Seligson 1993;Norris 2000;Rosenau 2003). Some argue that the defining class division in the contemporary world is between a transnational alliance of urban-dwelling economic and cultural elites, who construct and benefit from the international system, and a broad global underclass (Kanter 2003:120;Sassen 2001:339).
There has been some empirical validation of the argument that transnational connections and economic prosperity lead to cooperative-internationalist attitudes. Numerous studies have shown that economic, political, and cultural elites in Western countries are considerably more proglobal than their conationals Hiscox 2006, 2007;Mau, Mewes, and Zimmermann 2008;Oldendick and Bardes 1982;Rössel and Schroedter 2015;Wittkopf and Maggiotto 1983). However, the scant empirical work that has investigated pro-and antiglobal attitudes in developing countries suggests that the link between development and cooperativeinternationalist orientations may not match theoretical expectations. For instance, although more educated people tend to be more proglobal than conationals in relatively wealthy countries, these differences are more muted in relatively poor countries (Zhou 2016). Moreover, studies have shown that marginalized groups and those on the periphery use global networks and international organizations to advance their local political interests (Keck and Sikkink 1998;Paschel 2010;Tsutsui 2017Tsutsui , 2018. For example, self-identification as a "world citizen" is more common on the global periphery than in the developed core (Pichler 2011) as people in peripheral contexts look to the global order for protection or other aid (Gorman and Seguin 2018).
We test three hypotheses drawn from theoretical work that conceptualizes cooperative-internationalist attitudes as resulting from a diffusion process or as a function of class position. To do so, we analyze 18 different response items from six cross-national attitude surveys, covering a total of 116 countries that together account for 88 percent of the world's population. First, we find little variation in cooperative-internationalist attitudes between residents of countries situated at the core and periphery of the global system. Second, we show that class differences in cooperative-internationalist attitudes within countries are generally minimal. In fact, despite the suggestion of large class differences in cooperative-internationalist attitudes in theoretical and empirical studies, we find that elites are significantly more proglobal than non-elites in only 41 percent of tests at the country-wave level. Moreover, even statistically significant differences are generally substantively small: we can reject the hypothesis that elites are one response category higher than non-elites on a five-category response scale in more than 99 percent of our tests. Third, we find that class divides in cooperative-internationalist attitudes are almost exclusively found in core countries. We conclude with a call for more attention to how individuals interpret the costs and benefits of the global order for their local communities.

Cooperative Internationalism
Cooperative internationalism, an "orientation toward international affairs that stresses concern for others abroad, with whom one should work toward common goals" (Rathbun et al. 2016:125; see also Rathbun 2020), is perhaps the most common orientation toward the outside world among Western publics in the post-World War II era-even in countries with historically isolationist tendencies like the United States (Hainmueller and Hiscox 2007;Holsti 1979;Norris 2000;Oldendick and Bardes 1982;Pichler 2009;Wittkopf 1990;Wittkopf and Maggiotto 1983). Although relatively few longitudinal studies have assessed the popularity of this orientation (for one example using limited data, see Wittkopf 1986), there is suggestive evidence that it has become more popular over time as a result of both the diffusion of a cooperative "global culture" facilitated by international organizations and other global actors (Meyer et al. 1997) and cohort replacement, with younger generations coming of age in an increasingly prosperous, peaceful, and interconnected world (Holsti andRosenau 1980, 1986;Norris 2000;Pichler 2011;Woodward, Skrbis, and Bean 2008;Zhou 2016).
Despite its overall popularity, not everyone is enthusiastic about global cooperation. Previous scholarship has generally identified the rural, poor, and uneducated on the global periphery as most likely to resist the Western-dominated global order (Eschle and Maiguashca 2005;Igarashi and Saito 2014;Norris and Inglehart 2009;Tarrow 2005). On the other hand, many people with ethnonationalist leanings in Western countries tend to be "globophobic" (Bonikowski 2017;Bonikowski et al. 2019;Bonikowski and DiMaggio 2016;Held and McGrew 2007:2) and commonly embrace conspiracy theories about international organizations such as the United Nations (DiGrazia 2017). Events ranging from Brexit to the rise of the Italian Five Star Movement to the elections of Viktor Orbàn and Donald Trump suggest that antiglobal attitudes may be more common in the wealthy, well-connected countries of the global core than many studies acknowledge (Bremmer 2018;Stokes 2018;Zúquete 2018). In what follows, we outline theories concerning who holds cooperative-internationalist attitudes and test several hypotheses derived from these theories.

Who Are the Cooperative Internationalists?
Transnational and Cross-Cultural Contact One set of explanations, based loosely on intergroup contact theory (Pettigrew 1998), suggests that personal experience with foreign people and cultures fosters attitudes that are accepting of "diversity, hybridity, and otherness" (Skrbis and Woodward 2013:27), including positive evaluations of cooperation with foreign actors (Skrbis and Woodward 2011). The mechanisms thought to drive this cross-cultural contact include international travel (Mau et al. 2008), transnational interpersonal and business relationships (Brett and Moran 2011), living in "global cities" like Tokyo and New York (Sassen 2001), access to communication technology (Verboord 2017), and exposure to educational materials that confer knowledge about other cultures and the benefits of global cooperation (Hainmueller and Hopkins 2014:228;Oldendick and Bardes 1982;Wittkopf 1986;Wittkopf and Maggiotto 1983). Although each of these mechanisms has been shown to be related to cooperative-internationalist attitudes empirically, the vast majority of studies in this area have focused on populations in Western countries.
Urban elites tend to have more transnational connections and exposure to foreign cultures. Thus, cooperative internationalism is expected to primarily be a feature of the wealthy, educated, urban-dwelling frequent traveler who "knows seven kinds of sushi, recognizes the sound of the digeridoo, can recite verses from the Koran, handles chopsticks with dexterity, and enjoys the costumed spectacles of Indian cinema" (Boli 2005:397; see also Calhoun 2002). Working class and rural populations that "do not trade, work, love, marry, or do research internationally" (Lamont and Aksartova 2002:2), especially those living on the global periphery, are theorized to be less likely to hold proglobal attitudes (Mau 2010;Mewes and Mau 2013;Norris and Inglehart 2009).

Teleological Isomorphism: Diffusion and Modernization
A related set of explanations sees cooperative-internationalist attitudes as part of a package of cultural values that originated in the wealthy countries of the global core and is on the rise around the globe. The first of these draws on world polity theory. Scholars in this area argue that ever-deepening networks of connection between countries facilitate the diffusion of a singular "global culture" oriented toward values including rationality, democracy, citizenship, human rights, education, and international cooperation (Meyer et al. 1997). Here, exposure to global cultural scripts through contact with foreigners via international organizations like the United Nations provides models for interests, goals, and behaviors, resulting in isomorphism in cultures around the globe (Boli 2005;Hafner-Burton 2005;Longhofer and Schofer 2010). This global culture and its associated values are legitimated by international organizations and the powerful global actors that promote them; the spread of this culture, in turn, further legitimates the international system (Boli 2005;Meyer and Jepperson 2000;Tsutsui 2017).
World polity scholars suggest that global culture diffuses across network ties through two primary mechanisms (Downey et al. 2020). Some understand international organizations and the scientists, lawyers, and other professionals that manage and promote them as the primary generators and diffusers of global culture (Frank, Hironaka, and Schofer 2000;Kentikelenis and Seabrooke 2017;Meyer and Jepperson 2000). Here, global cultural elements, including positive attitudes toward international cooperation, are transmitted along global network ties facilitated by educated, urban-dwelling elites with access to these networks (Tarrow 2005;Tsutsui 2017). A more critical approach to the world polity sees global political networks as dominated by the well-connected countries of the global core and argues that so-called global cultural scripts have their origins in Western cultures and are intentionally designed to benefit wealthy, Western countries (Beckfield 2003(Beckfield , 2010Hughes et al. 2009;Kentikelenis and Babb 2019;Paxton, Hughes, and Reith 2015). Despite disagreement about whether global culture spreads through international organizations or connection to the most powerful members of the global community, world polity scholars arrive at the same conclusion: exposure to global cultural scripts, which is a function of global connection-particularly to the wealthy countries of the densely connected global core-tends to produce cooperative-internationalist attitudes.
Scholars in the modernization tradition also understand cooperative-internationalist attitudes as part of an increasingly common bundle of cultural values around the world. However, these scholars tend to focus on economic development, rather than diffusion along network ties, as the driving factor behind increasing cultural isomorphism (Inglehart 1997;Inglehart and Welzel 2005). They argue that economic development fosters a sense of existential security that promotes a shift from traditional materialist values that focus on survival to postmaterialist values, including a willingness to tolerate and cooperate with foreigners (Inglehart and Welzel 2010:564;Welzel and Inglehart 2010:45). Rural and poorer populations that lack the security wrought by economic prosperity, on the other hand, are more likely to adhere to traditionalist values that lend themselves to intolerance of difference, isolationism, and belligerence (Inglehart and Baker 2000). Global connection may even create existential anxieties among people in developing societies, making the international community a specific target for cultural ire (Norris 2000).

Global Class Conflict
Political economy approaches to cooperative internationalism focus on the distribution of the material benefits of the international system. Drawing on Marx and Engels' dictum that the bourgeoisie operates with a "cosmopolitan character" (Marx andEngels [1848] 1906:18), scholars in this tradition see cooperative internationalism as an element of the neoliberal ideology of the transnational capitalist class that dominates international organizations like the International Monetary Fund, the World Bank, and the United Nations (Giddens 1991;Gowan 2001;Kentikelenis and Babb 2019). Here, because global interconnectedness is primarily beneficial to "First World societies and the elites of the periphery" (Canclini 2014:151) (see also Hardt and Negri [2000]; Sassen [2001]; Sener [2007]) and may be harmful to poor people in peripheral countries (Dollar 2005;Rudra and Tobin 2017), the "global directorate" (Held and McGrew 2007:115) promotes cooperative-internationalist ideologies in an attempt to secure the global underclass' acquiescence in the "global war of position" (Carroll 2007:38). This proglobal neoliberalism flourishes among top managers in multinational corporations and consulting firms (Calhoun 2002:169) and eventually "trickles down to become a fashion of the middle classes" in developed countries (Skrbis and Woodward 2013:21). The "global proletariat" (Mittelman and Chin 2000)-poor, working class, and rural populations, especially on the global periphery-may be less likely to hold cooperative-internationalist attitudes because of economic precariousness in the face of distant and uncaring global forces (Beck 2011;Giddens 1991;Mayda 2006;Rosenau 2003;Scheve and Slaughter 2004).

Local Uses and Interpretations of Globalization
Although many scholars have argued that people on the global periphery should be particularly resistant or hostile to the global order, there are reasons to believe they may be more enthusiastic members of the global community than once thought. First, although the international order was constructed by wealthy Western countries, less powerful actors have historically used global networks as leverage against the powerful states that dominate the international system. For example, following the adoption of the Universal Declaration of Human Rights (UDHR) by the United Nations General Assembly, newly independent states in Africa and Asia "took center stage, pressing for racial, social, and economic equality in the world" by promoting treaties like the International Convention on the Elimination of All Forms of Racial Discrimination (ICERD) against the wishes of Western governments (Tsutsui 2009:12). Similarly, peripheral countries have formed their own international organizations-such as the Organization for Islamic Cooperation-to promote their shared interests via alternative channels of access to the international system that run parallel to the Western-dominated, secular-rational global order (Arjomand 2004;Boyle, Kim, and Longhofer 2015). Smaller, less powerful countries can also band together to counter the hegemony of more powerful countries. Examples include the Non-Aligned Movement (NAM), which challenged the Cold War-era bipolar system (Kullaa 2012) and the Association of Southeast Asian Nations (ASEAN), which was founded to ensure security for its members in the face of Chinese regional hegemony (Kim 2012). For these reasons, residents of countries on the global periphery may not necessarily feel alienated from control of international organizations.
Second, some scholars see proglobal orientations not only as "a lifestyle choice" of jet-setting professional elites but also as "the tragic involuntary condition" of marginalized people who look to the outside world for allies in domestic struggles (Beck 2011(Beck :1358; see also Landau and Freemantle 2010). Scholarship has shown that marginalized communities in the developing world use linkages with international organizations and powerful global actors to pressure domestic elites to alter repressive policies or at least force them to pretend to adhere to global standards (Gorman 2016;Hafner-Burton, Tsutsui, and Meyer 2008). Keck and Sikkink's (1998) classic work shows how activists formed international networks to challenge repressive domestic policies, such as the Argentine military junta's use of forced disappearances against political dissenters in the late 1970s and early 1980s. More recently, Tsutsui (2017) and Paschel (2010) illustrate how racial and ethnic minorities in Japan and Colombia used ties with international organizations and powerful global actors to pressure domestic governments into recognizing them as protected groups.
Finally, it is possible that differences in cooperative-internationalist attitudes between urban professional elites and non-elites is greater in the global core than on the periphery despite increased access to transnational connections, exposure to global cultural scripts, and economic development in these contexts. Because people likely understand the impact of global connection on their local communities as relative rather than absolute, those in core countries whose communities face declining prospects or importance in the global order may have stronger (and more negative) reactions than people in communities that are objectively lower on the global sociopolitical hierarchy but that have recently experienced increasing prospects or importance on the world stage (Alvaredo et al. 2018;Milanovic 2016). Poorer, less-educated, and more rural people in core countries may therefore have negative opinions about international cooperation due to resentment about being "replaceable" by foreigners (Beck 2011;Kertzer and Zeitzoff 2017;Mansfield and Mutz 2009;Scheve and Slaughter 2004). Empirical scholarship has shown that local leaders' antiforeigner political rhetoric and policies can reinforce these attitudes (Hopkins 2010), especially when framed as protecting local cultures against foreign contamination (Hainmueller and Hiscox 2010).
The handful of studies comparing cooperative internationalism between residents of core and peripheral countries are inconclusive but are more consistent with the idea that people form their interpretations of globalization based on local considerations. Most of these studies have found no difference in cooperativeinternationalist attitudes and global identification on the basis of country-level factors that include per capita gross domestic product (GDP) and various measures of economic, political, and social globalization (Zhou 2016), with a handful indicating that such attitudes may be more common on the global periphery (Gorman and Seguin 2018;Hurwitz et al. 1993;Pichler 2011). Even fewer studies have systematically compared proglobal orientations between elites and non-elites in core and peripheral contexts, although Gorman and Seguin (2018) find that repressed minorities are often more likely to identify as world citizens. Taken together, scholarship in this area suggests that people on the global periphery may use global connections to further their local goals, and may, therefore, be more supportive of international cooperation than once thought. Moreover, rural, less-educated, and poorer residents of core countries may be less enthusiastic about international cooperation despite-or perhaps because of-their exposure to global cultural and economic forces.

Hypotheses
Theories centering transnational contact, international organization, modernization, and global class conflict offer different mechanisms, but all predict that similar groups of people will adopt cooperative-internationalist attitudes. At the country level, whether through interpersonal contacts, embeddedness in global networks, economic development, or ideological hegemony, all suggest that people who live in developed, well-connected countries should be more likely to hold cooperative-internationalist orientations than those in developing countries on the global periphery. Despite the theoretical consistency of this expectation, results of empirical studies have been inconsistent, with different measures of cooperativeinternationalist attitudes having positive (Pichler 2009), negative (Gorman and Seguin 2018), null (Zhou 2016), or mixed (Pichler 2011) statistical relationships with various measures of development and global connection. Given its centrality to theory and mixed empirical support, whether residents of the global core are more proglobal remains a critical unanswered question. We propose the following hypothesis: H1: Individuals in wealthier countries will be more likely to hold cooperative-internationalist attitudes than individuals in poorer countries.
At the individual level, theories generally see the central division in cooperativeinternationalist attitudes as being between elites "with global connections" and non-elites "who are stuck in one place" (Kanter 2003:120). Although the theorized mechanisms linking urban elites and cooperative-internationalist orientations vary, empirical work demonstrates this link within the countries of the global core. Individuals with high levels of education Hiscox 2006, 2007;Mansfield and Mutz 2009), who travel frequently (Mau et al. 2008;Rössel and Schroedter 2015), live in urban areas (Bayram 2015;Zhou 2016), and are relatively wealthy (Hainmueller and Hiscox 2010;Pichler 2011) are more likely to see international cooperation in a positive light. As such, we propose the following hypothesis: H2: Urban professional elites will express higher levels of cooperativeinternationalist orientations than non-elites or elites living outside of urban areas.
Combining individual-and contextual-level determinants of cooperative-internationalist orientations leads to the expectation that country-level position in the global system moderates the effect of urban elite status on proglobal attitudes. Because inequalities in access to transnational connections, exposure to global cultural scripts, and existential security between elites and non-elites is larger at the global periphery than the global core (Norris and Inglehart 2009), theories strongly suggest that the gap in cooperative-internationalist attitudes should be greater in poorer countries. Likewise, some class-based approaches theorize the existence of a transnational elite class that includes entire societies on the global core but only elites on the global periphery (Canclini 2014;Hardt and Negri 2000;Koo 2016;Mittelman and Chin 2000). This leads us to hypothesize: H3a: The gap in cooperative-internationalist orientations between urban professional elites and non-elites will be greater in poorer countries than in richer countries.
Alternatively, some theories suggest that people interpret the benefits of global cooperation according to their perceptions of how it helps or hinders in reaching local goals. This research has tended to show that members of marginalized groups and those on the global periphery are no less likely-and perhaps more likely-to hold proglobal attitudes than elite conationals and residents of the densely connected global core (Gorman and Seguin 2018;Pichler 2011;Zhou 2016). Furthermore, there is ample evidence that rural, working class, and less-educated people in core countries are less likely to view international cooperation in a positive light than their urban-dwelling, wealthy, and highly educated conationals Hiscox 2006, 2007;Hurwitz and Peffley 1987;Mau et al. 2008;Oldendick and Bardes 1982;Rössel and Schroedter 2015;Wittkopf 1986;Wittkopf and Maggiotto 1983). Therefore, we also consider an alternative hypothesis: H3b: The gap in cooperative-internationalist orientations between urban professional elites and non-elites will be greater in wealthier countries than in poorer countries.

Data and Method
To test these hypotheses we analyze individual-level responses to six cross-national survey data sets: the International Social Survey Programme's (ISSP) first, second, and third waves on national identity; the fifth and sixth waves of the World Values Survey (WVS); the third and fourth waves of the Asia Barometer; the third and fourth waves of the Arab Barometer; the 2017 wave of the Latinobarómetro; and the fourth wave of the Afrobarometer. Altogether these data describe responses from 353,610 individuals across 116 countries at all levels of economic development and global integration from 1994 to 2017 and account for 88 percent of the world's population (see Appendix A in the online supplement for information on countries included in each survey wave).

Measuring Cooperative Internationalism: Dependent Variables
We measure cooperative-internationalist attitudes using survey items that tap into respondents' attitudes toward international cooperation, opinions about foreign influence on local communities, desired level of openness to the outside world, and self-identification as a world citizen. We intentionally avoid any survey items that mention specific countries or organizations, as these may evoke attitudes coming from particular histories with specific organizations rather than overall feelings about global cooperation. 1 Our final list of survey items totals 18 as follows: eight from the ISSP, one from the WVS, two from the Asia Barometer, two from the Arab Barometer, one from the Latinobarómetro, and four from the Afrobarometer. Although we would ideally use a common set of survey items across all surveys, there are no surveys with items that meet our criteria and also cover a relatively representative swath of the globe. Analyzing data from both global surveys like the ISSP and WVS (in which core countries are overrepresented) and regional surveys allows us a globally comparative view that has been missing in previous scholarship.
We analyze the following 18 survey items: • ISSP, waves 2 and 3: Large international companies damage local businesses (reverse-coded so that 1 = strongly agree, 5 = strongly disagree).
• ISSP, waves 1, 2, and 3: [Country] should limit the import of foreign products in order to protect its national economy (reverse-coded so that 1 = strongly agree, 5 = strongly disagree).
• ISSP, waves 1, 2, and 3: [Country] should follow its own interests, even if this leads to conflicts with other nations (reverse-coded so that 1 = strongly agree, 5 = strongly disagree).
• ISSP, waves 2 and 3: International organizations are taking too much power from the [country nationality] government (reverse-coded so that 1 = strongly agree, 5 = strongly disagree).
• ISSP, waves 2 and 3: In general, [Country] should follow the decisions of international organizations to which it belongs, even if the government does not agree with them (1 = strongly disagree, 5 = strongly agree).
• ISSP, wave 3: I feel more like a citizen of the world than of any country (1 = strongly disagree, 5 = strongly agree) • WVS, waves 5 and 6: I consider myself a world citizen (1 = strongly disagree, 4 = strongly agree).
• Asia Barometer, waves 3 and 4: We should protect our farmers and workers by limiting the import of foreign goods (1 = strongly agree, 4 = strongly disagree).
• Asia Barometer, waves 3 and 4: Given the chance, how willing would you be to go and live in another country? (reverse coded so that 1 = not willing at all, 4 = very willing).
• Arab Barometer, waves 3 and 4: Do you think that the increase in global connectivity is a good thing or a bad thing for society? (reverse-coded so that 1 = very bad, 5 = very good).
• Arab Barometer, wave 3: Is it better for your country to. . .
(1 = open up to the outside world to a greater extent, 3 = decrease its level of openness to the outside world).
• Afrobarometer, wave 4: In your opinion, how much do international donors and nongovernmental organizations help your country? (0 = do not help, 3 = help a lot).
• Afrobarometer, wave 4: In your opinion, how much do international businesses and investors help your country? (0 = do not help, 3 = help a lot).
• Afrobarometer, wave 4: Do you think international donors and nongovernmental organizations have too little, too much, or about the right amount of influence over your government? (0 = far too little, 5 = far too much).
• Afrobarometer, wave 4: Do you think international business and investors have too little, too much, or about the right amount of influence over your government? (0 = far too little, 5 = far too much).
We code variables across all data sets so that higher values indicate attitudes that are more positive toward international cooperation.

Independent Variables and Controls
At the country level, our key independent variable is position in the stratified global system. There is considerable scholarly debate over how to measure this. Scholars in the world polity tradition tend to use variables, including membership in intergovernmental organizations (IGOs) (Swiss and Longhofer 2016), the number of international nongovernmental organizations (INGOs) operating in a given country (Longhofer and Schofer 2010), and signatory status in international treaties (Cole 2015), as measures of embeddedness in global institutional networks. Others, drawing on political economy theories, counter that the wealthy countries of the global core tend to occupy central positions in global institutional and trade networks (Beckfield 2003(Beckfield , 2010Downey et al. 2020;Hughes et al. 2009;Paxton et al. 2015), have outsized influence on the policies and activities of international bodies, and serve as headquarters for powerful multinational corporations (Barrett, Kurzman, and Shanahan 2010;Kentikelenis and Babb 2019). Thus, although national wealth may not directly correspond to global connection, the two are very highly correlated. 2 Because of spatiotemporal limitations in data on global networks, 3 we operationalize global connectivity through the proxy of national wealth. We use the World Bank's World Development Indicators measure of per capita GDP 4 (logged) in current US dollars. As a robustness check, we reran all models using the KOF Globalisation Index (Dreher 2006) with substantively similar results (see Appendix B in the online supplement). Because previous work suggests that younger people (Norris 2000), women (Fite, Genest, and Wilcox 1990), and noncitizens (Gorman and Seguin 2018) may be more likely to hold cooperative-internationalist attitudes, we include controls for age (in decades) and gender in all models, and citizenship status of the respondent in the models that use ISSP data. 5 At the individual level, our key independent variable is a measure of whether or not a respondent belongs to the urban professional elite. Previous theory suggests that this is a configurational, rather than an additive, distinction-that is, educated, urban professionals should be more enthusiastic about international cooperation than others net of the main effects of either individually. Although these populations are typically wealthy, they need not be: relatively poor artists and intellectuals qualify as urban elites, whereas wealthy (but less-educated) business owners or landowners in rural settings do not. As such, we measure belonging to the urban elite as a dummy variable indicating that a respondent lives in an urban setting 6 and meets one of the following three criteria: (1) has at least some college education, (2) has a household income 7 in the top quintile in a given country-wave, or (3) has a head of household whose profession is classified as "service class one" on the 10-category EGP scale (examples include scientists, engineers, medical doctors, legal professionals, university professors, authors, journalists, designers, artists, performers, economists, financial professionals, and administrators of large firms (see Ganzeboom, De Graaf, and Treiman 1992). 8 In general, the literature understands urban professional elite status as absolute rather than relative; as such, some countries should have more urban professional elites than others. Differences in rates of urban professional elite respondents between countries should not affect our individual-level results comparing elite and non-elite respondents within countries. Nevertheless, differences in survey methodology and local contexts can make cross-national measurements unstable. In order to ensure that idiosyncrasies in our operationalization of urban professional elite status are not driving our results, we conducted a series of supplemental analyses as robustness checks. We reran all of our models using the following changes to our criteria for urban professional elite status: (1) omitting income as a secondary criterion, (2) omitting profession as a secondary criterion, and (3) using education-our most comparable variable across all surveys-as the only criterion. Results of all models using these alternative specifications are substantively very similar to our reported results, with only minor changes in significance levels (see Appendixes C, D, and E in the online supplement for results).

Modeling
We test for the effects of both individual-and country-level independent variables on cooperative-internationalist attitudes using a series of statistical models. First, we run a series of hierarchical linear regression models with country-wave as the level 2 unit of analysis and standard errors clustered on countries. Because theory suggests that the effect of belonging to the urban professional elite on cooperative-internationalist attitudes may vary across countries in idiosyncratic ways, we include a random slope term for the urban professional elite variable in all hierarchical models. 9 For each of our 18 dependent variables, we test all three of our hypotheses with two sets of analyses. Our first set of analyses is a series of linear regression models for each of the 248 country-waves in our sample with the controls from the previous section included. The result is a total of 885 statistical tests for class divides within countries. Our second set of analyses consists of 18 hierarchical models with individual-and country-level, as well as cross-level, interactions. All models use robust standard errors.
Because our dependent variables are ordinal, readers may reasonably wonder whether nonlinear models would be more appropriate. We use linear models because it is difficult to interpret results from nonlinear models. As a robustness check, we reran all analyses using ordinal logistic regression models with substantively similar results (see Appendix F in the online supplement).

Missing Data
For simplicity's sake, we ran all models using list-wise deletion of observations with missing data. As a robustness check, we imputed missing data points using iterative chained equations, creating 10 imputations and transforming all variables before imputing (von Hippel 2009) with substantively similar results (see Appendix G in the online supplement).

Results
We test each of our hypotheses using descriptive comparisons across country-waves and two different kinds of regression model for each of our 18 dependent variables. Our first hypothesis is that individuals in wealthier countries should be more likely to hold cooperative-internationalist attitudes than individuals in poorer countries. The gray fit lines in the quiver plots in Figures 1 and 2 illustrate the relationship between per capita GDP (logged scale) on the x axis and country-wave level mean values for each dependent variable on the y axis (standardized to range from 0 to 1 to aid in comparative interpretation). Figure 1 shows results from the global survey data; Figure 2 shows results from the regional survey data. The gray fit lines in both figures show that cooperative-internationalist attitudes are generally popular across all country-waves but have an inconsistent relationship with per capita GDP. 10 The correlation between country-wave mean values of our dependent variables and per capita GDP is -0.15, casting doubt on H1. We test this hypothesis more fully in the nested multilevel regression models; the results of models using the global survey data are presented in Table 1 and the results of models using the regional surveys are presented in Table 2. The relationship between per capita GDP and cooperative-internationalist attitudes is decidedly mixed, ranging from statistically significant and positive at the p < 0.001 level to statistically significant and negative at the p < 0.01 level, with the modal coefficient statistically null.
Our second hypothesis is that urban professional elites should express higher levels of cooperative-internationalist orientations than non-elite conationals. To test this, we first ran separate regressions for each of our 18 dependent variables in every country-wave with available data, which resulted in 885 separate regressions. Because of the complexity of presenting and interpreting results of so many regressions in tabular form, we use the quiver plots in Figures 1 and 2 to illustrate the results. Each arrow represents results from a country-wave. On the y axis, we standardized the coefficients from our models to indicate the proportion change in the dependent variable associated with belonging to the urban professional elite; each arrow begins at the constant term and ends at the sum of the constant term and the urban professional elite coefficient. GDP per capita (logged scale) is on the x axis. The color and direction of each arrow indicates the sign and significance level of the coefficient. Upward-facing blue arrows represent positive and significant coefficients, downward-facing red arrows represent negative and significant coefficients, and yellow arrows represent coefficients that do not achieve statistical significance at the p < 0.05 level. Darker shades of the blue and red arrows indicate lower p values.
The prevalence of dark blue arrows in Figure 1 shows that in many countrywaves for many of our dependent variables in the global survey data, urban professional elites are more likely to hold cooperative-internationalist attitudes than non-elite conationals. This is true for a majority of country-waves for the dependent variables measuring attitudes about large international corporations (55 percent), import limits (76 percent), the power of international organizations relative to national governments (61 percent), and teaching foreign languages in schools (57 percent). Likewise, the results of the nested models in Tables 1 and 2   The gray fit line tests H1 by illustrating the relationship between GDP per capita and the mean value of each dependent variable (scaled to range from 0-1) for a given country-wave. Colored arrows test H2 and H3, with standardized regression coefficients on the y axis indicating the proportion change in the dependent variable associated with urban professional elite status in each country-wave. Blue arrows indicate positive and significant coefficients, red arrows indicate negative and significant coefficients, and yellow arrows indicate coefficients that do not achieve statistical significance at the p < 0.05 level. Darker shades indicate smaller p values. cally significant and positive relationship between urban professional elite status and cooperative-internationalist attitudes for 14 of 18 dependent variables across both global and regional surveys (minimum p < 0.05), with two at the p < 0.10 level. These results provide initial support for H2. However, there is some evidence that the effect of urban professional elite status on cooperative-internationalist attitudes is much weaker than theory would suggest. First, in the country-wave level analyses of the global survey data, the urban professional elite variable fares far worse in the other five dependent variables, with significant and positive coefficients in only 18 percent to 28 percent of cases. Second, Figure 2 shows that the relationship between urban professional elite status and cooperative-internationalist attitudes is more tenuous in the regional surveys, as  Figure 2: Quiver plots illustrating the effect of urban professional elite status on cooperative-internationalist attitudes using results of country-wave level regressions on regional survey data. Note: This figure displays results of 171 individual linear models regressing urban professional elite status on each dependent variable for each country-wave in the regional surveys. The gray fit line tests H1 by illustrating the relationship between GDP per capita and the mean value of each dependent variable (scaled to range from 0-1) for each country-wave. Colored arrows test H2 and H3, with standardized regression coefficients on the y axis indicating the proportion change in the dependent variable associated with urban professional elite status in each country-wave. Blue arrows indicate positive and significant coefficients, red arrows indicate negative and significant coefficients, and yellow arrows indicate coefficients that do not achieve statistical significance at the p < 0.05 level. Darker shades indicate smaller p values.
the number of tests with negative and null coefficients (73 percent) dwarfs those with positive and significant coefficients (27 percent). Even where coefficients are positive and significant, effect sizes are generally small-for the tests with positive and statistically significant coeffients, belonging to the urban professional elite is associated with an average increase of 7 percent in the dependent variable, and elites never score one category higher than non-elites on a five-category response scale (judged by whether the 95 percent confidence interval of the coefficient is greater than 0.20 on the standardized 0 to 1 scale). Finally, the results of nested models using regional survey data (see Table 2) show that urban professional elite status is statistically related to cooperative-internationalist attitudes for only five out of nine dependent variables. Note: All models using data from multiple survey waves include wave-level fixed effects and robust standard errors clustered at the country level. All models using data from a single survey wave include robust standard errors clustered at the country level. * p < 0.05; † p < 0.01.  Note: All models using data from multiple survey waves include wave-level fixed effects and robust standard errors clustered at the country level. All models using data from a single survey wave include robust standard errors clustered at the country level. There are no variables measuring citizenship in any of the surveys used in this table. * p < 0.05; † p < 0.01.
Third, we test two competing hypotheses: H3a expects that the gap in cooperativeinternationalist orientations between elites and non-elites should be greater in peripheral countries than in core countries, whereas H3b posits the reverse relationship. Regional surveys are less skewed toward high-income countries than the global surveys, and country-wave level results from these surveys (see Figure 2) suggest that statistically significant class differences in cooperative-internationalist attitudes are more common in higher-income countries, with null and negative coefficients primarily clustered in lower-and middle-income countries. For example, in the Afrobarometer, in which 2008 South Africa tops the per capita GDP list at just under $6,000, a mere 8 percent of country-wave level tests indicate a significant and positive effect of urban professional elite status on cooperative-internationalist attitudes. These results provide initial support for H3b over H3a.
To test these hypotheses explicitly, we add an interaction effect between the urban professional elite dummy and per capita GDP in full models (see Tables 1  and 2). For 8 of our 18 dependent variables, the coefficient for the interaction term is positive and significant, with the remaining 10 being statistically null. Because previous research indicates that statistically significant interaction terms are inadequate for evaluating cross-level interactions (Brambor, Clark, and Golder 2006), we include Figures 3 and 4 to illustrate the marginal effect of belonging to the urban professional elite on our measures of cooperative-internationalist attitudes across levels of per capita GDP. These show that class differences in attitudes become more likely as per capita GDP increases. Results from the WVS and Afrobarometer, both of which have relatively large samples of low-and middle-income countries, indicate no relationship between urban professional elite status and cooperativeinternationalist attitudes in the world's poorest countries. In fact, the negative (but not statistically significant at the p < 0.05 level) coefficients suggest that urban professional elites in relatively poor countries may be less supportive of global cooperation than their conationals. Taken together, these results support H3b over H3a.

Discussion and Conclusion
The most prominent existing theories understand cooperative-internationalist attitudes as an elite phenomenon. As such, these theories expect variation in these attitudes to run parallel to class divides both between and within countries. Scholars in disparate theoretical traditions ranging from political psychology, world polity, modernization, and international political economy all agree on a central guiding principal: the key division in proglobal attitudes should be between elites "with global connections" and non-elites "who are stuck in one place" (Kanter 2003:120;see also Boli 2005;Canclini 2014;Mewes and Mau 2013;Welzel and Inglehart 2010). Because the mechanisms theorized to lead to proglobal attitudes-transnational relationships, international connection, and economic prosperity-are more widely available in the countries of the global core, these theories suggest that "the affluent and educated urban elite living in developing countries may come to share similar attitudes with their counterparts living in Europe or North America," resulting in   Table  1. The figure shows that gaps between elites and non-elites are more common and larger in higher-income countries.
"greater polarization between elites and more conservative groups living in traditional rural communities" in peripheral contexts (Norris and Inglehart 2009:21). Results from empirical tests describing people from 116 countries representing more than 88 percent of the world's population, however, tell a different story. We find no consistent difference in support for cooperative internationalism between countries situated at the core and periphery of the global system, as measured by per capita GDP. Although results of nested models indicate that urban professional elites are, on average, slightly more likely to hold cooperative-internationalist attitudes, cross-level interactions indicate a class divide only in the wealthy, wellconnected countries of the global core; in the poorest countries in our sample, the elite/non-elite gap is not statistically different from zero. Taken together, these results cast serious doubt on a central tenet of many of the most prominent theoretical approaches to cooperative internationalism: that it is primarily "linked to resource endowment" (Rössel and Schroedter 2015:86).   Table 2. The figure shows that gaps between elites and non-elites are more common and larger in higher-income countries.
The lack of a class divide in proglobal attitudes outside of wealthy countries suggests that people in varying contexts develop proglobal orientations in different ways and for different reasons. Urban professional elites around the world may indeed perceive themselves as beneficiaries of globalization due to their membership in a "transnational capitalist class" or "cosmocracy" and adopt and promote cooperative-internationalist attitudes accordingly (Calhoun 2002;Canclini 2014;Carroll 2007;Hardt and Negri 2000;Mittleman and Chin 2000;Sassen 2001;Sener 2007). At the same time, non-elites may understand global connection and cooperation as beneficial to them, providing avenues for redress of grievances against repressive domestic states (Gorman and Seguin 2018), opportunities to connect with similarly marginalized groups abroad (Tsutsui 2018), or access to lucrative foreign markets (Alvaredo et al. 2018). Research on cooperative internationalism outside of wealthy countries has been minimal, and more research is needed to develop a full understanding of the mechanisms producing these attitudes on the periphery.
Our dependent variables capture attitudes about many aspects of global cooperation: economic, political, and cultural. The results of our analyses show variation in responses to these questions along these lines. Economic protectionist attitudes, which appear to be more popular on the periphery than in the global core and less popular among elites, tend to follow theoretical expectations more closely than attitudes about political globalization or openness to foreign people and cultures, which show little variation across countries or elite status. This is consistent with our assertion that people interpret globalization according to its varying impacts on local contexts. Someone may, without contradiction, view international organizations and transnational social exchanges positively but think that core-periphery economic relationships are inherently exploitative. Likewise, it is possible to understand economic development through international trade and investment as beneficial while simultaneously resenting foreign cultural influence and loss of national sovereignty to international organizations. We urge future scholars to investigate how constellations of pro-and antiglobal attitudes shift as a function of local conditions.
Our results do not mean that contact and economic development play no role in the development of cooperative-internationalist attitudes for two reasons. First, even if these connections do not diffuse such attitudes, cooperative internationalism is an interpretation of contact and global development and therefore is predicated largely on their existence. Second, connections to the outside world through communication technologies, trade, and international organizations may have reached a saturation point, such that economic class and country position no longer describe much meaningful variation in international connection. Meyer and colleagues' (1997) classic example of a hypothetical new-found island nation's immediate inundation with foreign aid, global development models, and international organizational linkages suggests as much: global connection is something that "no one in any corner of the world whatsoever can avoid" (Beck 2011(Beck :1352. We hope future scholars will take up the question of how residents of this new-found island nation would interpret such an onslaught of connection with the outside world and where those interpretations would lead them.

Notes
1 For example, a question in the World Values Survey probes respondents' confidence in the United Nations. Although the country-wave level mean rate of respondents reporting a "great deal" of confidence across the data is 14 percent, more than 56 percent of Bangladeshi respondents in the 1999 to 2004 wave expressed this level of enthusiasma clear outlier. Rather than reflecting an appreciation of international cooperation per se, this result is likely driven by Bangladesh's record as among the top five contributors to United Nations peacekeeping efforts (Kathman and Melin 2017). In return, Bangladesh is provided financial resources by the United Nations for its role as "good international citizen," leading many Bangladeshis to have positive views of the United Nations in particular (Murthy 2007:158). More generally, Kiley and Vaisey (2020) find that general political attitudes are more stable than those describing specific entities, so a person may have an authoritarian orientation but favor limits on executive authority when their party is out of power.
2 In our sample of country-years, logged GDP per capita correlates with the INGO network centrality score developed by Paxton et al. (2015) at 0.78 and the KOF Globalisation Index at 0.85.
3 For example, the most commonly used INGO and IGO network scores are only available in 5-year increments (see Paxton et al. 2015;Gorman and Seguin 2015). 6 The ISSP, Asia Barometer, Arab Barometer, and Afrobarometer all include dummy variables that capture urban/rural residence. For the WVS and Latinobarómetro, we coded anyone living in a city with more than 100,000 inhabitants as urban. For countrywaves in the WVS in which no respondents reported living in a city of this size, we coded anyone in the highest population category for that country-wave as urban.
7 The fourth wave of the Arab Barometer has an 11-point household income scalehere we coded any respondent in the top three categories as rich. We used alternative measures for two survey-waves that did not have measures of household income. In the Latinobarómetro, we coded any respondent who self-placed into the "highest" social class as rich. In the Afrobarometer, we coded anyone who said they had never gone without a cash income as rich.
8 For most country-waves in most surveys, we were able to categorize professions based on recorded EGP categories or International Standard Classification of Occupation codes. For the handful of country-waves with locally specific coding schemes, we hand-coded available profession responses as belonging to EGP service class 1 based on our judgment. Because the fourth wave of the Afrobarometer does not contain data on profession, we coded any respondent who said that they or their spouse had a full-time job as a professional. Because profession item was not asked in 45 country-waves in the WVS data (54 percent of country-waves), we do not include profession as a secondary criterion in WVS models.
9 Models without the random slope term produce substantively similar results.
10 With all of the dependent variables standardized to range from 0 to 1, the country-wave means range from 0.32 (Georgia, ISSP wave 3: "I feel more like a citizen of the world than of any country") to 0.93 (Latvia, ISSP wave 1: "[Country] schools should make much more effort to teach foreign languages properly") and the overall mean for all country-wave-variables is well above the midpoint at 0.61.