Ethnicity, Democracy, Trust: A Majority-Minority Approach

Ethnicity, Democracy, Trust: A Majority-Minority Approach Abstract Why do ethnic and racialized minorities have lower trust? While previous research emphasizes individual factors such as the national and cultural origins of ethnic groups, this paper draws attention to the ethnic majority-minority relationship. We argue that ethnic differences in trust are a function of the power dynamics underlying this relationship and that these dynamics are particularly salient in democratic political systems. To test this argument, we develop new measures of ethnic majority-minority status, which for the first time allows for global cross-national comparison of heterogeneous ethnic groups at the micro level. Using the World Values Survey, we test the majority-minority argument, showing that, while democracy increases generalized trust across the board, it also leads to a gap in trust that favors the majority group. This gap remains even after the inclusion of controls for country differences in factors such as ethnic diversity and GDP. Introduction Recent tensions with Muslims in France, ongoing Eastern European resistance to Syrian and North African refugees, and racially motivated crimes in Scandinavia remind us that in many countries ethnicity and race continue to be a major axis of social, economic, and political inequality. These challenges are often accompanied by a crisis of trust—seen through a significant difference in the trust in generalized others shown by ethnic minority and majority populations. Indeed, studies of trust in European countries as well as Canada and the United States clearly show that multiple measures of ethnicity (including race, national origin, and language) predict generalized trust, that is, the belief that most people can be trusted. One explanation of these differences focuses on the fact that minority groups are often immigrants. Differences in trust stem from the fact that immigrants have lower levels of economic and social capital and because they often come from less democratic countries, where trust tends to be scarce (De Vroome, Hooghe, and Marien 2013; Dinesen 2012, 2013; Uslaner 2008). However, ethnic gaps in trust cannot simply be attributed to immigration from less democratic to more democratic societies. As illustrated by the cases of Black Americans in the United States and people of color and Quebecers in Canada, members of minority groups born in democratic societies also have lower trust (Smith 2010; Stolle and Harell 2013; Wilkes 2011). This presents a puzzle. Democracy is widely held to be beneficial for trust across the board, including for ethnic minorities (Rothstein 2011; Rothstein and Uslaner 2005). How, then, can theoretical arguments about the benefits of democracy for trust be reconciled with the evidence of ethnic gaps in trust in democratic societies? The argument presented in this paper is that, in addition to the role of individual characteristics and place of origin, the power denoted in majority-minority group position needs consideration. That is, while ethnicity denotes particular socio-cultural and/or racial categories, these categories also reflect the experience of political, economic, and social power. Power is an important precursor to trust (Catala 2015; Yamagishi 2011). We further posit that this experience of majority-minority position is likely to be more salient in democratic societies than in non-democratic societies. By definition, the majority group has more potential for power, especially political power, relative to the minority in democratic systems. In non-democratic societies, as there is no principle of majority rule, either the power gap between the majority and minority is less significant or else the situation is even reversed—the minority group holds political and economic power. A test of this argument requires considering ethnic groups in both democratic and non-democratic societies. Studies to date have yet to do so, focusing mostly on comparisons among largely democratic societies (Kotzian 2011; Mewes 2012; Ziller 2017). The lack of attention to non-democratic societies might be because of a theoretical interest in European immigration and race relations in Western societies. This lack of attention is compounded by a major empirical challenge to doing so. Globally, different countries use different categories to denote ethnicity. The United States, for example, tends to use race-based categories of identification such as White, Black, and Asian, whereas in Ghana the major ethnic groups include Ga Afangbe, Akan, Ewe, Dagbani, and Guan. These differences mean there is little way to compare ethnic groups across countries except in terms of macro-level diversity. Here we present new micro-level measures of majority-minority status based on differences stemming from ethnic self-identification, language, and immigration that allow us to surmount this challenge.1 We provide a detailed description of how to create these new measures with the World Values Survey data. The use of these measures means that this paper provides the first global cross-national comparison of ethnic groups at the micro level. We then use the new majority-minority measures to analyze the relationship between ethnicity, trust, and democratization. The results show that democracy does condition the effect of ethnic minority-majority position on trust. Although everyone trusts more in democracies, the majority-minority gap in trust is also larger. These results hold irrespective of whether majority-minority group status is defined on the basis of ethnic majority-minority self-identification, majority-minority language, or immigration status. Trust, Ethnicity, and Democratization Trust is often understood as “a generalized expectancy held by an individual or a group that the word, promise, verbal, or written statement of another individual or group can be relied on” (Rotter 1971, 444). Trust in unknown others of this kind is of vital importance to the interactions between citizens, their leaders, and their institutions. It is seen as foundational to all aspects of society, operating as a form of social glue (Lewis and Weigert 1985, 2012; see also Wu and Wilkes 2016) and therefore leads to important societal and political benefits, among them personal and social well-being (Helliwell and Wang 2010), policy support (Hetherington and Globetti 2002), and political participation (Uslaner and Brown 2005). For this reason, trust is one of the most central concerns across the social sciences (Nannestad 2008; Smith 2010). Survey research shows that in Europe responses to the widely used question indicating generalized trust—whether “most people can be trusted”—often vary on the basis of immigration status, with immigrants less trusting than the native born (Doerschler and Jackson 2012; Hooghe et al. 2009). Cross-European studies show a trust gap between immigrants and non-immigrant groups more generally (Kotzian 2011; Mewes 2012, 2014), as do country-specific studies of Denmark (Bjørnskov 2008) and the Netherlands (De Vroome, Hooghe, and Marien 2013) (see also Dinesen [2012, 2013]; Ljunge [2014]; Nannestad et al. [2014]; Ziller [2015]). Several studies have attributed the lower trust of immigrant groups, especially those coming to Europe, to the fact that they came from less democratic societies that engender lower human capital as well as a culture of distrust (Dinesen 2012, 2013; De Vroome, Hooghe, and Marien 2013; Ljunge 2014; Ziller 2015). In democratic systems, laws are put in place to ensure that different groups of citizens are treated relatively equally, that there is a distribution of resources subject to “binding consultation,” and that there is protection “against arbitrary action by government agents” (Tilly 2001, 31). This is why factors such as the length of democracy, institutional effectiveness, and institutional impartiality all increase trust (Rothstein and Stolle 2008; Rothstein and Uslaner 2005). In contrast, in less democratic societies corruption and nepotism rule the day (ibid.). This explains why generalized trust is, on average, higher in more democratic countries, with the Scandinavian countries—Sweden, Norway, Denmark—having the highest mean global trust levels (Delhey and Newton 2005; Rothstein and Stolle 2001). Writing about ethnicity, Tilly (1998, 243) says that democracy “attenuates the performance effects of previously categorically differentiated experiences and reduces the transmission of categorically organized advantages from generation to generation.” The scope of who can be trusted is expanded and, hence, ethnic outgroup trust tends to be much higher in democratic societies (Delhey, Newton, and Welzel 2011). The problem is that we also see ethnic gaps in trust in democratic and non-democratic societies that cannot be attributed to migration from less democratic societies (Putnam 1993; Smith 2010; Uslaner 2008). Indeed, in democratic societies these ethnic gaps in trust exist even though democratic institutions are supposed to ensure fairness and equality. Blacks in the United States trust less than White Americans (Smith 2010). Indigenous people and visible minorities have lower trust than non-Indigenous people and non-visible minorities (Hwang 2013; Stolle and Harell 2013). Quebecers have lower trust than the rest of Canada (Grabb et al. 2009; Wu and Wilkes 2017). Elsewhere, the research also shows that Croats are more trusting than Bosniaks in Bosnia and Herzegovina (Håkansson and Sjöholm 2007) and that, in the Russian Federation, Russians have higher trust in most people than do Tatars or Yakuts (Bahry et al. 2005). Our argument is that, in addition to considering the role of democracy, the relative power denoted in majority-minority group position needs consideration. Indeed, since trust is an interpersonal construct and a social relationship (Rotter 1971; Yamagishi 2011), power is one of its important precursors (e.g., Bachmann 2001; Cook et al. 2005; Farrell 2004; Nunkoo and Ramkissoon 2012; Schilke, Reimann, and Cook 2015; Yamagishi 2011, 26). Power matters because it regulates risk and uncertainty (Guseva and Rona-Tas 2001; Lévi-Strauss 1969; Molm, Whitham, and Melamed 2012). More powerful people tend to trust more because they are more capable of handling the negative consequences should the trust relationship falter (Nunkoo and Ramkissoon 2012; Oskarsson et al. 2009; Öuberg and Svensson 2010; Wu and Wilkes 2016; Yamagishi 2011). As Cook et al. (2005, 140–42) write, power disparities create “fertile ground for distrust” and “commonly block the possibility of trust’” (see also Nunkoo and Ramkissoon 2012, 1005; Öuberg and Svensson [2010, 145]). While much of the social exchange literature on power and trust treats power as a characteristic of a specific and relatively short-term exchange relationship (Cook and Yamagishi 1992; Molm, Whitham, and Melamed 2012; Schilke, Reimann, and Cook 2015), the dynamics associated with ethnicity are a more enduring form of inequality (Tilly 1998). That is, while ethnicity denotes socio-cultural and/or racial categories of identification, these categories also reflect the experience of political, economic, and social power (Okamoto and Mora 2014; Pokharel 2013, 1) and the “valuations of the actors involved” (Ridgeway and Smith-Lovin 1999; Roscigno 2011, 352). We further argue that the majority-minority relationship is unique in democracies because, by definition, the majority group in democracies is likely to have more power relative to the minority group. This is not to say that actual powerlessness on the part of minorities is greater in democracies. Rather, the argument is that while democracy does increase individual political rights, democratic systems must also contend with the “tyranny of the majority” whereby the democratic process enables the majority to dominate the minority (Brittain 1996; Gamble 1997). Although there are clearly checks and balances in place (Lewis 2013), its potential may nevertheless be a major reason why minority groups in democracies have lower social trust than the majority group. At the group level, cases in democratic societies such as Black Pete in the Netherlands, the Washington Redskins in the United States, and Tintin (Belgium) with its depiction of Congolese people lead to “a deep divide between majority and minority groups” in terms of how they understand social practice and this, in turn, could affect trust (Catala 2015, 424). This relative imbalance is accompanied and reflective of heightened expectations around equality that come with democracies. As Kymlicka writes, in today’s democracies, members of historically subordinated groups demand equality, and demand it as a right. They “believe they are entitled to equality, and entitled to it now, not in some indefinite or millenarian future” (2002, 8). Therefore, while rights are accorded on an individual basis, the ideology of equality may also create a heightened expectation of fair treatment on the part of minority groups (Ziller 2017). In contrast, in less democratic systems the power manifested in the ethnic majority-minority relationships is often quite different. Expectations around fairness and equality may not exist in less democratic societies, where policies are top down and do little to mitigate the advantage held by some ethnic minority groups (Chua 2004; Horowitz 1995). The majority ethnic group does not de facto have more power. The Chinese in Malaysia, for example, have been economically advantaged, and the Sunnis in Lebanon have occupied positions of political privilege (Davis and Moore 1997). Certainly in many African countries “minority-group elites, with the backing of their ethnic kin or of ethnic coalitions in which their own group was dominant, have often been disproportionately represented or influential players in many of Africa’s authoritarian regimes” (Prempeh 2013, 441). Thus it is that advantaged minorities such as the Kikuyu in Kenya, despite comprising a relatively small proportion of the overall population, have produced three out of four presidents and have dominated the local economy (Bezemer and Jong-A-Pin 2013; Bienen 2015). Taken together, these points suggest that the nature of the majority-minority relationship is likely to be very different in democratic societies—although everyone will trust more in democratic societies, the gap in trust may also be larger. A test of this argument requires comparing individual members of majority and minority ethnic groups in democracies with individual members of majority and minority ethnic groups in other kinds of political systems. Therefore, we need to be able to consider both individuals as well as political systems. Table 1 illustrates three general frameworks—macro; micro; and micro-macro—currently used to study ethnicity and trust. We use table 1 to illustrate that with only a few exceptions, current research has yet to consider individual ethnic minority groups across political systems. Table 1. Frameworks for the study of the relationship between ethnicity, democracy, and trust, showing limited number of studies in fourth column Framework  Macro  Micro  Micro-macro  Sample research questions  Is average trust higher in more democratic countries?  Does ethnicity affect trust?  Is there an effect of ethnicity on trust controlling for state/country-level factors?  Does the effect of ethnicity on trust vary in democratic and non-democratic societies?  Dependent variable           Trust  yes  yes  yes  yes  Independent variables           Ethnicity            Individual-level ethnicity  X  yes  Xa  yes    Regional/country-level ethnic diversity  yes  X  Xa  yes   Democracy  yes  X  Xa  yes  Comparative dimension  yes  X  yes  yes  Sample studies using framework  Knack and Keefer (1997); Bjørnskov (2008); Tsai et al. (2011); Ziller (2015)  Bahry et al. (2005); Håkansson and Sjöholm (2007); Wilkes (2011); Dinesen (2012)  Putnam (2007); Uslaner (2008); Sturgis et al. (2011); Dinesen and Sønderskov (2015)  Ziller (2017) (European countries)  Framework  Macro  Micro  Micro-macro  Sample research questions  Is average trust higher in more democratic countries?  Does ethnicity affect trust?  Is there an effect of ethnicity on trust controlling for state/country-level factors?  Does the effect of ethnicity on trust vary in democratic and non-democratic societies?  Dependent variable           Trust  yes  yes  yes  yes  Independent variables           Ethnicity            Individual-level ethnicity  X  yes  Xa  yes    Regional/country-level ethnic diversity  yes  X  Xa  yes   Democracy  yes  X  Xa  yes  Comparative dimension  yes  X  yes  yes  Sample studies using framework  Knack and Keefer (1997); Bjørnskov (2008); Tsai et al. (2011); Ziller (2015)  Bahry et al. (2005); Håkansson and Sjöholm (2007); Wilkes (2011); Dinesen (2012)  Putnam (2007); Uslaner (2008); Sturgis et al. (2011); Dinesen and Sønderskov (2015)  Ziller (2017) (European countries)  aFor those studies that include both individual and macro-level variables, the limitation is that if there is a measure of individual-level ethnicity, then it is typically a within-country study, whereas if there are multiple countries, then the individual ethnicity measure is either absent or limited. Table 1. Frameworks for the study of the relationship between ethnicity, democracy, and trust, showing limited number of studies in fourth column Framework  Macro  Micro  Micro-macro  Sample research questions  Is average trust higher in more democratic countries?  Does ethnicity affect trust?  Is there an effect of ethnicity on trust controlling for state/country-level factors?  Does the effect of ethnicity on trust vary in democratic and non-democratic societies?  Dependent variable           Trust  yes  yes  yes  yes  Independent variables           Ethnicity            Individual-level ethnicity  X  yes  Xa  yes    Regional/country-level ethnic diversity  yes  X  Xa  yes   Democracy  yes  X  Xa  yes  Comparative dimension  yes  X  yes  yes  Sample studies using framework  Knack and Keefer (1997); Bjørnskov (2008); Tsai et al. (2011); Ziller (2015)  Bahry et al. (2005); Håkansson and Sjöholm (2007); Wilkes (2011); Dinesen (2012)  Putnam (2007); Uslaner (2008); Sturgis et al. (2011); Dinesen and Sønderskov (2015)  Ziller (2017) (European countries)  Framework  Macro  Micro  Micro-macro  Sample research questions  Is average trust higher in more democratic countries?  Does ethnicity affect trust?  Is there an effect of ethnicity on trust controlling for state/country-level factors?  Does the effect of ethnicity on trust vary in democratic and non-democratic societies?  Dependent variable           Trust  yes  yes  yes  yes  Independent variables           Ethnicity            Individual-level ethnicity  X  yes  Xa  yes    Regional/country-level ethnic diversity  yes  X  Xa  yes   Democracy  yes  X  Xa  yes  Comparative dimension  yes  X  yes  yes  Sample studies using framework  Knack and Keefer (1997); Bjørnskov (2008); Tsai et al. (2011); Ziller (2015)  Bahry et al. (2005); Håkansson and Sjöholm (2007); Wilkes (2011); Dinesen (2012)  Putnam (2007); Uslaner (2008); Sturgis et al. (2011); Dinesen and Sønderskov (2015)  Ziller (2017) (European countries)  aFor those studies that include both individual and macro-level variables, the limitation is that if there is a measure of individual-level ethnicity, then it is typically a within-country study, whereas if there are multiple countries, then the individual ethnicity measure is either absent or limited. The first column shows the macro-level framework. A typical research question for this framework would be “Is average trust higher in more democratic societies?” To answer this question requires comparing the average levels of trust across places—these could be national and/or regional—and assessing whether places with greater democratization have higher or lower trust (e.g., see Knack and Keefer 1997; Delhey and Newton 2005; Bjørnskov 2008; Ziller 2015). This approach can consider the impact of place-based ethnic diversity via ethnic fractionalization measures (Alesina and La Ferrara 2002). Macro-level measures of ethnic fractionalization can only denote the level of ethnic diversity within societies. They contain no information about economic or cultural differences between groups (Baldwin and Huber 2010). The macro-level measures of ethnic BGI (between group (income) inequality) and CLF (linguistic cultural fractionalization) developed by Baldwin and Huber are a highly innovative solution to this problem, though at present these remain limited to 46 countries. What the macro-level approach cannot do, however, is consider the individual-level impact of ethnicity on trust. Such a task has, however, been taken up by micro-level research, the framework of which is illustrated in the second column. Here the focus is on a host of individual-level variables and how they affect trust, and key among these is ethnicity. This research clearly documents ethnic gaps in trust within societies (e.g., see Bahry et al. 2005; Dinesen 2012; Håkansson and Sjöholm 2007; Wilkes 2011). Information as to the individual-level effects of ethnicity is limited in that these studies focus on a single country—therefore, we have little sense of how country-specific factors may or may not matter. The third column shows the micro-macro framework. In this framework there is a focus on both individual micro-level characteristics such as age and education as well as macro-level characteristics such as geographically based ethnic diversity. Here a typical question might be “Do individuals show differences in trust even when they live in different contexts?” or “Does place-based ethnic diversity affect individual-level trust?” (see, e.g., Dinesen and Sønderskov 2015; Putnam 2007; Sturgis et al. 2011; Uslaner 2008). While this approach typically has a measure of individual ethnicity and/or of democracy and diversity, only one study in the European context (Ziller 2017) has a measure of both. That is, as illustrated in the fourth column, while there are a few studies that consider whether ethnic diversity conditions the effect of ethnicity (e.g., Soroka, Helliwell, and Johnston [2003] consider the interaction between the level of diversity in a census tract and visible minority status in Canada) by and large these remain constrained to single-country studies. Those that have more country-level information (e.g., Kesler and Bloemraad 2010) typically do not have an individual-level measure of ethnicity.2 Ultimately, table 1 shows that there is little research that has simultaneously considered ethnicity, democratization, and trust.3 These concepts—and associated variables—are needed in order to consider the majority-minority argument. One reason why no study has done so to date might be that such an endeavor requires an individual-level measure of ethnicity that is comparable across countries so that country-level differences in democratization can also be included. This remains a major challenge because currently there is no way to measure ethnicity cross-nationally with categories that are commensurable across countries—at least not globally. That is, in many cases the standard markers of ethnicity—for example ethnic self-identification or language—are incommensurable across most countries. In, for example, the World Values Survey data, the ethnic self-identification categories for the United States include White/Caucasian, White Hispanic, Black African, Other, and Mixed. In Mexico the groups include colored (light), colored (dark), and white. Singapore’s categories include Chinese, South Asian, Malay, and Eurasian. Kyrgyzstan’s categories include Kyrgyz, Russian, Uzbek, and Turkish. While these ethnic categories are reflective of the classifications that countries believe are important (Patsiurko, Campbell, and Hall 2012), they cannot be matched across countries. One solution to this problem is to make use of a measure such as visible minority status or foreign born. This solution might only be relevant to certain national contexts. Dinesen and Sønderskov (2015) (Denmark) and Helliwell, Wang, and Xu (2014) (global) have a measure of whether the respondent is an immigrant. Still, it remains the case that ethnic minorities, particularly in non-democratic societies, are not all immigrants. In the following discussion of the empirical application of the majority-minority approach, we provide a solution to this problem. Data and Variables The primary focal measures are trust, ethnicity (majority-minority), and democratization. Most of these measures are from the most recent two waves of the World Values Survey (2005–2009, 2010–2014)—variables from other sources are appended. Table 2 provides the documentation, coding, and summary statistics for all variables used in the analyses.4 Table 2. Measures of key variables in analysis Variable  Survey question  Coding  Mean (%)  Individual level         Trust  WVS (A165): Most people can be trusted or you can’t be too careful  Binary (0, 1)  25.89   Majority-minority  WVS (X051): Belong to the non-dominant group in society  Binary (0, 1)  19.34  WVS (G016): Speak non-dominant language at home  Binary (0, 1)  16.36  WVS (G027A): Self-identified as an immigrant  Binary (0, 1)  4.55   Age  WVS (X003): in years  in years, 15–99  41.78   Gender  WVS (X001): 1 = female, 0 = male  Binary (0, 1)  52.18   Income  WVS (X047): Scale of income  Scale (1–10)  4.74   Life satisfaction  WVS (A170): Satisfaction with your life  Scale (1–10)  6.77   Education  WVS (X025): Highest educational level attained  Scale (1–8)  4.81   Political trust  WVS (E069_11) Confidence in the government  Scale (1–4)  2.442  Macro level         Democratization  Economist Intelligence Unit Democracy Index; Authoritarian: (1–3.99), Hybrid democracy: (4–5.99), Flawed democracy: 6–7.99), Full democracy: (8–10)  Index (1–10)  6.08   Ethnic fractionalization  Fearon (2003) ethnic fractionalization index  Index (0–1)  0.426   Gross Domestic Product  World Bank GDP per capita  Per capita constant dollars ($245–$83,556)  $16,330  Variable  Survey question  Coding  Mean (%)  Individual level         Trust  WVS (A165): Most people can be trusted or you can’t be too careful  Binary (0, 1)  25.89   Majority-minority  WVS (X051): Belong to the non-dominant group in society  Binary (0, 1)  19.34  WVS (G016): Speak non-dominant language at home  Binary (0, 1)  16.36  WVS (G027A): Self-identified as an immigrant  Binary (0, 1)  4.55   Age  WVS (X003): in years  in years, 15–99  41.78   Gender  WVS (X001): 1 = female, 0 = male  Binary (0, 1)  52.18   Income  WVS (X047): Scale of income  Scale (1–10)  4.74   Life satisfaction  WVS (A170): Satisfaction with your life  Scale (1–10)  6.77   Education  WVS (X025): Highest educational level attained  Scale (1–8)  4.81   Political trust  WVS (E069_11) Confidence in the government  Scale (1–4)  2.442  Macro level         Democratization  Economist Intelligence Unit Democracy Index; Authoritarian: (1–3.99), Hybrid democracy: (4–5.99), Flawed democracy: 6–7.99), Full democracy: (8–10)  Index (1–10)  6.08   Ethnic fractionalization  Fearon (2003) ethnic fractionalization index  Index (0–1)  0.426   Gross Domestic Product  World Bank GDP per capita  Per capita constant dollars ($245–$83,556)  $16,330  Note: WVS, World Values Survey (2005–2008, 2010–2014). Table 2. Measures of key variables in analysis Variable  Survey question  Coding  Mean (%)  Individual level         Trust  WVS (A165): Most people can be trusted or you can’t be too careful  Binary (0, 1)  25.89   Majority-minority  WVS (X051): Belong to the non-dominant group in society  Binary (0, 1)  19.34  WVS (G016): Speak non-dominant language at home  Binary (0, 1)  16.36  WVS (G027A): Self-identified as an immigrant  Binary (0, 1)  4.55   Age  WVS (X003): in years  in years, 15–99  41.78   Gender  WVS (X001): 1 = female, 0 = male  Binary (0, 1)  52.18   Income  WVS (X047): Scale of income  Scale (1–10)  4.74   Life satisfaction  WVS (A170): Satisfaction with your life  Scale (1–10)  6.77   Education  WVS (X025): Highest educational level attained  Scale (1–8)  4.81   Political trust  WVS (E069_11) Confidence in the government  Scale (1–4)  2.442  Macro level         Democratization  Economist Intelligence Unit Democracy Index; Authoritarian: (1–3.99), Hybrid democracy: (4–5.99), Flawed democracy: 6–7.99), Full democracy: (8–10)  Index (1–10)  6.08   Ethnic fractionalization  Fearon (2003) ethnic fractionalization index  Index (0–1)  0.426   Gross Domestic Product  World Bank GDP per capita  Per capita constant dollars ($245–$83,556)  $16,330  Variable  Survey question  Coding  Mean (%)  Individual level         Trust  WVS (A165): Most people can be trusted or you can’t be too careful  Binary (0, 1)  25.89   Majority-minority  WVS (X051): Belong to the non-dominant group in society  Binary (0, 1)  19.34  WVS (G016): Speak non-dominant language at home  Binary (0, 1)  16.36  WVS (G027A): Self-identified as an immigrant  Binary (0, 1)  4.55   Age  WVS (X003): in years  in years, 15–99  41.78   Gender  WVS (X001): 1 = female, 0 = male  Binary (0, 1)  52.18   Income  WVS (X047): Scale of income  Scale (1–10)  4.74   Life satisfaction  WVS (A170): Satisfaction with your life  Scale (1–10)  6.77   Education  WVS (X025): Highest educational level attained  Scale (1–8)  4.81   Political trust  WVS (E069_11) Confidence in the government  Scale (1–4)  2.442  Macro level         Democratization  Economist Intelligence Unit Democracy Index; Authoritarian: (1–3.99), Hybrid democracy: (4–5.99), Flawed democracy: 6–7.99), Full democracy: (8–10)  Index (1–10)  6.08   Ethnic fractionalization  Fearon (2003) ethnic fractionalization index  Index (0–1)  0.426   Gross Domestic Product  World Bank GDP per capita  Per capita constant dollars ($245–$83,556)  $16,330  Note: WVS, World Values Survey (2005–2008, 2010–2014). Trust is measured with the question “Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people?” coded on a 0–1 scale (1 denotes trust) for the WVS data (see, e.g., Fairbrother [2014]; Hooghe et al. [2009]; and Johnson and Mislin [2012] on the use of this measure). Ethnic majority-minority group status is coded in three ways: based on ethnic self-identification (X025), language use (language spoken at home, G016), and immigrant group status (self-identification as an immigrant, G027A). While self-identification as an immigrant is comparable across countries, this is not the case with either the ethnic self-identification or language spoken at home measures. Different countries have different sets of categories of ethnic self-identification and home language.5 There is no consistent overlap across countries for these categories. Using multiple different measures of ethnic majority-minority status further ensures the robustness of our argument. Table 3 illustrates this problem using the example of the United States, Sweden, and Taiwan. For each country, the table shows the WVS country categories for ethnic group and language at home. The US ethnic categories include Black/African, Spanish/Hispanic, White/Caucasian, Other, and Mixed Races; Sweden’s include White/Caucasian and Asian- Central (Arabic), Asian-East (Chinese, Japanese), Asian-South (Indian, Hindu, Pakistani), and Black African; and Taiwan’s categories includes Minnanese from Taiwan, Hakka from Taiwan, China, and Aborigine. It might be argued that White/Caucasian is common, but this is only trust for the United States and Sweden, not for Taiwan. The language variable also shows little overlap: the United States has six languages, Sweden has seven, and Taiwan has five. Further is that while English (spoken in the United States and Sweden) might be one of the world’s global languages, not only is it irrelevant in the Taiwanese context, but the meaning of speaking English is clearly different in Sweden than it is in the United States. Table 3. Example of the majority-minority coding strategy (WVS 2010–2014), majority in bold United States  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  Black African  127  10.17  No answer  33  1.48  Spanish; Hispanic  124  9.93  Chinese  7  0.31  White/Caucasian White  922  73.82  English  2,016  90.32  Other  36  2.88  French  2  0.09  Mixed races  40  3.2  Japanese  3  0.13        Spanish; Castilian  145  6.5        Other  26  1.16  Total  1,249  100  Total  2,232  100  Sweden  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  Missing; Not specified  7  0.58  Missing  1  0.08  Don´t know  16  1.33  No answer  2  0.17  Asian-Central (Arabic)  28  2.32  Chinese  1  0.08  Asian-East (Chinese, Japanese)  7  0.58  English  13  1.08  Asian-South (Indian, Hindu, Pakistani)  30  2.49  Finnish  3  0.25  Black African  8  0.66  French  2  0.17  White/Caucasian White  1,091  90.46  Spanish; Castilian  9  0.75  Other  19  1.58  Swedish  1,114  92.37        Other  61  5.06  Total  1,206  100  Total  1,206  100  Taiwan  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  No answer  7  0.57  No answer  2  0.16  Don´t know  10  0.81  Don´t know  1  0.08  Other  39  3.15  Mandarin  528  42.65  TW: Hakka from Taiwan  110  8.89  Other  51  4.12  Minnanese from Taiwan  987  79.73  TW: Taiwanese (Minnanese)  628  50.73  TW: China  73  5.9  TW: Hakka  27  2.18  TW: Aborigine  12  0.97  TW: Aborigine language  1  0.08  Total  1,238  100  Total  1,238  100  United States  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  Black African  127  10.17  No answer  33  1.48  Spanish; Hispanic  124  9.93  Chinese  7  0.31  White/Caucasian White  922  73.82  English  2,016  90.32  Other  36  2.88  French  2  0.09  Mixed races  40  3.2  Japanese  3  0.13        Spanish; Castilian  145  6.5        Other  26  1.16  Total  1,249  100  Total  2,232  100  Sweden  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  Missing; Not specified  7  0.58  Missing  1  0.08  Don´t know  16  1.33  No answer  2  0.17  Asian-Central (Arabic)  28  2.32  Chinese  1  0.08  Asian-East (Chinese, Japanese)  7  0.58  English  13  1.08  Asian-South (Indian, Hindu, Pakistani)  30  2.49  Finnish  3  0.25  Black African  8  0.66  French  2  0.17  White/Caucasian White  1,091  90.46  Spanish; Castilian  9  0.75  Other  19  1.58  Swedish  1,114  92.37        Other  61  5.06  Total  1,206  100  Total  1,206  100  Taiwan  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  No answer  7  0.57  No answer  2  0.16  Don´t know  10  0.81  Don´t know  1  0.08  Other  39  3.15  Mandarin  528  42.65  TW: Hakka from Taiwan  110  8.89  Other  51  4.12  Minnanese from Taiwan  987  79.73  TW: Taiwanese (Minnanese)  628  50.73  TW: China  73  5.9  TW: Hakka  27  2.18  TW: Aborigine  12  0.97  TW: Aborigine language  1  0.08  Total  1,238  100  Total  1,238  100  Table 3. Example of the majority-minority coding strategy (WVS 2010–2014), majority in bold United States  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  Black African  127  10.17  No answer  33  1.48  Spanish; Hispanic  124  9.93  Chinese  7  0.31  White/Caucasian White  922  73.82  English  2,016  90.32  Other  36  2.88  French  2  0.09  Mixed races  40  3.2  Japanese  3  0.13        Spanish; Castilian  145  6.5        Other  26  1.16  Total  1,249  100  Total  2,232  100  Sweden  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  Missing; Not specified  7  0.58  Missing  1  0.08  Don´t know  16  1.33  No answer  2  0.17  Asian-Central (Arabic)  28  2.32  Chinese  1  0.08  Asian-East (Chinese, Japanese)  7  0.58  English  13  1.08  Asian-South (Indian, Hindu, Pakistani)  30  2.49  Finnish  3  0.25  Black African  8  0.66  French  2  0.17  White/Caucasian White  1,091  90.46  Spanish; Castilian  9  0.75  Other  19  1.58  Swedish  1,114  92.37        Other  61  5.06  Total  1,206  100  Total  1,206  100  Taiwan  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  No answer  7  0.57  No answer  2  0.16  Don´t know  10  0.81  Don´t know  1  0.08  Other  39  3.15  Mandarin  528  42.65  TW: Hakka from Taiwan  110  8.89  Other  51  4.12  Minnanese from Taiwan  987  79.73  TW: Taiwanese (Minnanese)  628  50.73  TW: China  73  5.9  TW: Hakka  27  2.18  TW: Aborigine  12  0.97  TW: Aborigine language  1  0.08  Total  1,238  100  Total  1,238  100  United States  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  Black African  127  10.17  No answer  33  1.48  Spanish; Hispanic  124  9.93  Chinese  7  0.31  White/Caucasian White  922  73.82  English  2,016  90.32  Other  36  2.88  French  2  0.09  Mixed races  40  3.2  Japanese  3  0.13        Spanish; Castilian  145  6.5        Other  26  1.16  Total  1,249  100  Total  2,232  100  Sweden  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  Missing; Not specified  7  0.58  Missing  1  0.08  Don´t know  16  1.33  No answer  2  0.17  Asian-Central (Arabic)  28  2.32  Chinese  1  0.08  Asian-East (Chinese, Japanese)  7  0.58  English  13  1.08  Asian-South (Indian, Hindu, Pakistani)  30  2.49  Finnish  3  0.25  Black African  8  0.66  French  2  0.17  White/Caucasian White  1,091  90.46  Spanish; Castilian  9  0.75  Other  19  1.58  Swedish  1,114  92.37        Other  61  5.06  Total  1,206  100  Total  1,206  100  Taiwan  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  No answer  7  0.57  No answer  2  0.16  Don´t know  10  0.81  Don´t know  1  0.08  Other  39  3.15  Mandarin  528  42.65  TW: Hakka from Taiwan  110  8.89  Other  51  4.12  Minnanese from Taiwan  987  79.73  TW: Taiwanese (Minnanese)  628  50.73  TW: China  73  5.9  TW: Hakka  27  2.18  TW: Aborigine  12  0.97  TW: Aborigine language  1  0.08  Total  1,238  100  Total  1,238  100  However, if we consider whether these categories denote ethnic majority-minority status, they can be compared in a meaningful way. We aggregated each variable to the country level and then used the country percentages to identify the numeric majority group and language.6 In table 3, the majority group is bolded. Those individuals who, for example, speak Swedish in Sweden and English in the United States and the UK are all majority group members. We did the same for ethnic self-identification that then allows us to compare ethnic groups that may be different “races” or other groups. Individuals in the most numerous group were coded as majority, and all others were coded as minority.7Appendix A provides the list of country (years) and the frequency count for each majority-minority measure. The correlations/tetrachoric correlations across these variables—ethnic majority-minority self-identification and majority-minority language (.153, .22), ethnic majority-minority self-identification, and immigrant status (.205, .40), and between majority-minority language and immigrant status (.031, .08)—are small, indicating that each of these variables are not the same measure.8 We also include a number of individual controls—gender (X001), age (X003), education (X025), income (V239), life satisfaction (A170), and political trust (E069_11).9 At the macro level, we measure democratization in two ways with the 10-point Economist Intelligence Unit scale (see http://www.eiu.com/home.aspx). First, the EIU codes countries on five dimensions—liberties, government functioning, political participation, political culture, and electoral process and participation—that are measured with 60 indicators (Kekic 2007). It then assigns each country an overall score of 1–10. We matched the year of the EIU scale to the specific country-year values. The exception are the 2005/2006 country years where there was no EIU data, and therefore we use 2007 (the EIU scale values change little from year to year). Second, the EIU also groups countries into one of four categories based on their overall score—Authoritarian (1–3.99), Hybrid Democracy (4–5.99), Flawed Democracy (6–7.99), Full Democracy (8–10)—and we use these categories in the latter portion of the analyses as a way to consider effects within regime types (see also Wu and Wilkes 2018).10 Results hold if the EIU scale is replaced with the Polity IV scale (with −10 denoting hereditary monarchy and +10 denoting consolidated democracy). This scale also rates countries on the basis of competitiveness of executive recruitment, openness of executive recruitment, constraints on chief executive, regulation of participation, and competitiveness of participation (Marshall et al. 2000), but we elected not to use it because it is less up to date. We also include country-specific mean education level, Gross Domestic Product (GDP), and ethnic fractionalization.11 Statistical Models Because the WVS trust measure is a binary 0–1 outcome, we use mixed-effects logistic regression models of individuals nested within country-years.12 All models also include individual controls. The elements of the models considered are expressed as follows:   log(Pij(T=1)1−Pij(T=1))=α0+α0j+α1Mij+α2Dj+α3Mij*Dj+α4Cij+α5CCj (2)where Pij represents the probability of being trusting for the ith individual in the jth country or country-year.13 The fixed effects are α0– α4, and α0j is the random intercept at the country-year/country level. Our key predictor Mij represents being a member of a minority group,14 and Dj denotes the democracy index of the jth country. We consider the interaction between ethnicity and democracy, which is denoted by Mij*Dj, and α3 is the coefficient. Cij denotes a series of individual-level control variables (age, gender, education, etc.), and α4 is a vector of corresponding coefficients. CCj denotes the list of country-year- (education, GDP per capita) and country- (ethnic fractionalization) level controls, and α5 is a vector of corresponding coefficients. Country-specific ethnic fractionalization is not expected to change significantly over a four-year period. Since we use the base specification approach to interactions with dummy variables, the interaction term refers to the difference in the slope effect of democratization between the majority and minority ethnic group rather than to the group-specific slope effect of democratization (Yip and Tsang 2007). Therefore, in a model with an ethnic minority dummy variable, democratization, and an interaction between ethnic minority and democratization, the coefficient for ethnic minority refers to the effect of being a minority overall in the less democratic countries. The coefficient for democratization refers to the overall effect of democratization. The coefficient for the interaction term captures the gaps between the majority-minority trust gaps at different levels of democracy (refer to figures 1–3—for example, the interaction term is the gap between the data points on these graphs). Figures 1–3. View largeDownload slide Plots showing country gaps in trust between majority-minority by level of democratization Note: Positive gap denotes higher majority group trust. Figures 1–3. View largeDownload slide Plots showing country gaps in trust between majority-minority by level of democratization Note: Positive gap denotes higher majority group trust. Results Descriptive We begin by presenting the bivariate correlation between the country-specific majority-minority trust gap and democratization. Since trust is coded as a 0, 1 binary, this means the gap denotes the difference between the proportion of the minority who trust and the proportion of the majority who trust. A positive gap means that the majority group trusts more, and a negative gap means that the minority group trusts more. In 2005, for example, 11 percent of WVS respondents (n = 111) in the UK self-identified as an ethnic minority while 89 percent (n = 911) did not. The proportion who trust among the minority group is therefore 25 percent (28/111), and the proportion who trust among the majority group is 31 percent (283/911)—resulting in a trust gap of 6 percent (= 0.31 – 0.25). Appendix B provides the country-specific gaps for each coding of majority-minority status (e.g., ethnic minority gap, language gap, immigrant gap). Figures 1–3 show the country-specific gaps plotted by the level of democratization. Each figure represents a different coding of ethnic minority status based on, respectively, ethnic identity, language, and immigration. Because there are so many countries, for readability, we denote countries with dots rather than their country codes. Each dot refers to a country-specific gap in trust—positive dot indicate that the gap favors the majority group, and negative dots indicate that the gap favors the minority group. The fitted values line shows an upward trend, indicating that the gap increases (to favoring the majority group) as the level of democratization increases. In order to further illustrate the pattern in figures 4–6, we take these country-specific gaps in trust and consider them based on our categorical indicator of democratization that codes countries as authoritarian, hybrid, flawed, and full democracies. Figures 4–6. View largeDownload slide Boxplots showing gaps in trust by political regime type Note: Positive gap denotes higher majority group trust. Figures 4–6. View largeDownload slide Boxplots showing gaps in trust by political regime type Note: Positive gap denotes higher majority group trust. Figure 4 shows the boxplots for ethnic minority status for each type of government. The line in the middle of the box denotes the median trust gap, the outer edges of the box show the upper and lower quartiles, and the farthest lines show the maximum and minimum trust gaps (excluding outliers, which are denoted by the black dots). Essentially what these boxplots show is a negative or non-existent gap in authoritarian, hybrid, and flawed democracies (denoting that in terms of the median there is either no gap or that ethnic minorities trust more) and a positive gap (denoting higher majority group trust) in the fully democratic societies. Figures 5 and 6 also show a similar pattern—there is either no gap or a gap in favor of minorities in non-full democracies and a gap that favors the majority group in democratic societies (e.g., majority group members are more trusting). Multivariate Next we present the results of the mixed-effects logistic regression models. Table 4 presents the coefficients (log-odds) and standard errors from these models. For each measure of ethnic majority minority (self-identified as an ethnic minority, speak non-dominant language at home = linguistic minority, and not born in the country = immigrant), we estimate three models. The first model tests the main effects of ethnic majority-minority status and democratization by including a dummy variable for minority status and the democratization scale measure. It also includes all the individual controls. This model allows us to test whether there is an ethnic majority-minority gap in trust across the board and whether trust increases with democratization. The results show that, across the board, there is a negative and statistically significant effect of ethnic minority status on trust (−.120), meaning that ethnic minority group members trust less. The coefficient for democratization is positive (.149), indicating that trust increases with democratization. Table 4. Multilevel models showing effect of ethnicity and democratization on trust Minority=  Belong to ethnic minority group  Speak non-dominant language at home  Self-identified as an immigrant    Model (1)  Model (2)  Model (3)  Model (4)  Model (5)  Model (6)  Model (7)  Model (8)  Model (9)  Ethnicity effect  Belong to minority group (0, 1)  −0.120***  −0.119***  0.473***  −0.039  −0.036  0.263***  −0.066  −0.066  0.379**  (0.023)  (0.023)  (0.077)  (0.024)  (0.024)  (0.075)  (0.048)  (0.048)  (0.139)  Democracy effect                    Democracy index (1–10)  0.149*  −0.048  −0.025  0.186***  −0.003  0.006  0.166*  −0.012  −0.006  (0.064)  (0.072)  (0.072)  (0.050)  (0.057)  (0.057)  (0.077)  (0.093)  (0.094)  Interaction effect                    Minority##Democracy      −0.093***      −0.051***      −0.070***      (0.012)      (0.012)      (0.021)  Individual characteristics  Female (0, 1)  −0.066***  −0.067***  −0.068***  −0.041*  −0.041*  −0.042*  −0.023  −0.023  −0.023  (0.019)  (0.019)  (0.019)  (0.017)  (0.017)  (0.017)  (0.026)  (0.026)  (0.026)  Age (15–99)  0.005***  0.005***  0.005***  0.003***  0.003***  0.003***  0.003***  0.003***  0.003***  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  Education (0–8)  0.070***  0.070***  0.070***  0.069***  0.069***  0.068***  0.090***  0.089***  0.090***  (0.005)  (0.005)  (0.005)  (0.004)  (0.004)  (0.004)  (0.007)  (0.007)  (0.007)  Income (1–10)  0.042***  0.042***  0.042***  0.045***  0.045***  0.045***  0.045***  0.045***  0.045***  (0.005)  (0.005)  (0.005)  (0.004)  (0.004)  (0.004)  (0.007)  (0.007)  (0.007)  Life satisfaction (1–10)  0.054***  0.054***  0.054***  0.070***  0.070***  0.070***  0.072***  0.072***  0.072***  (0.005)  (0.005)  (0.005)  (0.005)  (0.005)  (0.005)  (0.007)  (0.007)  (0.007)  Political trust (1–4)  0.241***  0.241***  0.240***  0.240***  0.240***  0.239***  0.213***  0.213***  0.213***  (0.011)  (0.011)  (0.011)  (0.010)  (0.010)  (0.010)  (0.015)  (0.015)  (0.015)  Country-year characteristics  Education (societal mean)    0.266*  0.262    0.181  0.183    0.388*  0.392*    (0.135)  (0.135)    (0.118)  (0.118)    (0.189)  (0.189)  GDP per capita (in thousand US $)    0.020*  0.019*    0.026***  0.026***    0.028*  0.028*    (0.010)  (0.010)    (0.007)  (0.007)    (0.012)  (0.012)  Fearon’s ethnic diversity index (0–1)    0.007*  0.007*    0.001  0.001    0.001  0.002    (0.003)  (0.003)    (0.003)  (0.003)    (0.004)  (0.004)  Wave 2010–2014 (Ref. 2005–2009)  0.006  −0.271  −0.264  0.002  −0.297  −0.294  (Wave 2010–2014 only)      (0.244)  (0.207)  (0.208)  (0.209)  (0.189)  (0.189)      Constant  −3.992***  −4.736***  −4.861***  −4.181***  −4.283***  −4.342***  −3.884***  −5.669***  −5.731***  (0.448)  (0.787)  (0.789)  (0.360)  (0.645)  (0.646)  (0.500)  (1.178)  (1.179)  Random effect (country-year)  Variance  0.884***  0.566***  0.569***  0.747***  0.533***  0.534***  0.712***  0.472***  0.473***  (0.163)  (0.105)  (0.106)  (0.127)  (0.091)  (0.091)  (0.206)  (0.136)  (0.137)  N (individual)  73,917  73,917  73,917  87,737  87,737  87,737  33,971  33,971  33,971  N (country-year)  60  60  60  71  71  71  25  25  25  Wald χ2  1,255  1,293  1,353  1,722  1,757  1,773  671  687  697  AIC  72,436  72,415  72,352  89,332  89,314  89,299  37,054  37,050  37,040  Minority=  Belong to ethnic minority group  Speak non-dominant language at home  Self-identified as an immigrant    Model (1)  Model (2)  Model (3)  Model (4)  Model (5)  Model (6)  Model (7)  Model (8)  Model (9)  Ethnicity effect  Belong to minority group (0, 1)  −0.120***  −0.119***  0.473***  −0.039  −0.036  0.263***  −0.066  −0.066  0.379**  (0.023)  (0.023)  (0.077)  (0.024)  (0.024)  (0.075)  (0.048)  (0.048)  (0.139)  Democracy effect                    Democracy index (1–10)  0.149*  −0.048  −0.025  0.186***  −0.003  0.006  0.166*  −0.012  −0.006  (0.064)  (0.072)  (0.072)  (0.050)  (0.057)  (0.057)  (0.077)  (0.093)  (0.094)  Interaction effect                    Minority##Democracy      −0.093***      −0.051***      −0.070***      (0.012)      (0.012)      (0.021)  Individual characteristics  Female (0, 1)  −0.066***  −0.067***  −0.068***  −0.041*  −0.041*  −0.042*  −0.023  −0.023  −0.023  (0.019)  (0.019)  (0.019)  (0.017)  (0.017)  (0.017)  (0.026)  (0.026)  (0.026)  Age (15–99)  0.005***  0.005***  0.005***  0.003***  0.003***  0.003***  0.003***  0.003***  0.003***  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  Education (0–8)  0.070***  0.070***  0.070***  0.069***  0.069***  0.068***  0.090***  0.089***  0.090***  (0.005)  (0.005)  (0.005)  (0.004)  (0.004)  (0.004)  (0.007)  (0.007)  (0.007)  Income (1–10)  0.042***  0.042***  0.042***  0.045***  0.045***  0.045***  0.045***  0.045***  0.045***  (0.005)  (0.005)  (0.005)  (0.004)  (0.004)  (0.004)  (0.007)  (0.007)  (0.007)  Life satisfaction (1–10)  0.054***  0.054***  0.054***  0.070***  0.070***  0.070***  0.072***  0.072***  0.072***  (0.005)  (0.005)  (0.005)  (0.005)  (0.005)  (0.005)  (0.007)  (0.007)  (0.007)  Political trust (1–4)  0.241***  0.241***  0.240***  0.240***  0.240***  0.239***  0.213***  0.213***  0.213***  (0.011)  (0.011)  (0.011)  (0.010)  (0.010)  (0.010)  (0.015)  (0.015)  (0.015)  Country-year characteristics  Education (societal mean)    0.266*  0.262    0.181  0.183    0.388*  0.392*    (0.135)  (0.135)    (0.118)  (0.118)    (0.189)  (0.189)  GDP per capita (in thousand US $)    0.020*  0.019*    0.026***  0.026***    0.028*  0.028*    (0.010)  (0.010)    (0.007)  (0.007)    (0.012)  (0.012)  Fearon’s ethnic diversity index (0–1)    0.007*  0.007*    0.001  0.001    0.001  0.002    (0.003)  (0.003)    (0.003)  (0.003)    (0.004)  (0.004)  Wave 2010–2014 (Ref. 2005–2009)  0.006  −0.271  −0.264  0.002  −0.297  −0.294  (Wave 2010–2014 only)      (0.244)  (0.207)  (0.208)  (0.209)  (0.189)  (0.189)      Constant  −3.992***  −4.736***  −4.861***  −4.181***  −4.283***  −4.342***  −3.884***  −5.669***  −5.731***  (0.448)  (0.787)  (0.789)  (0.360)  (0.645)  (0.646)  (0.500)  (1.178)  (1.179)  Random effect (country-year)  Variance  0.884***  0.566***  0.569***  0.747***  0.533***  0.534***  0.712***  0.472***  0.473***  (0.163)  (0.105)  (0.106)  (0.127)  (0.091)  (0.091)  (0.206)  (0.136)  (0.137)  N (individual)  73,917  73,917  73,917  87,737  87,737  87,737  33,971  33,971  33,971  N (country-year)  60  60  60  71  71  71  25  25  25  Wald χ2  1,255  1,293  1,353  1,722  1,757  1,773  671  687  697  AIC  72,436  72,415  72,352  89,332  89,314  89,299  37,054  37,050  37,040  Standard errors in parentheses,* p < 0.05 ** p < 0.01 *** p < 0.001 Table 4. Multilevel models showing effect of ethnicity and democratization on trust Minority=  Belong to ethnic minority group  Speak non-dominant language at home  Self-identified as an immigrant    Model (1)  Model (2)  Model (3)  Model (4)  Model (5)  Model (6)  Model (7)  Model (8)  Model (9)  Ethnicity effect  Belong to minority group (0, 1)  −0.120***  −0.119***  0.473***  −0.039  −0.036  0.263***  −0.066  −0.066  0.379**  (0.023)  (0.023)  (0.077)  (0.024)  (0.024)  (0.075)  (0.048)  (0.048)  (0.139)  Democracy effect                    Democracy index (1–10)  0.149*  −0.048  −0.025  0.186***  −0.003  0.006  0.166*  −0.012  −0.006  (0.064)  (0.072)  (0.072)  (0.050)  (0.057)  (0.057)  (0.077)  (0.093)  (0.094)  Interaction effect                    Minority##Democracy      −0.093***      −0.051***      −0.070***      (0.012)      (0.012)      (0.021)  Individual characteristics  Female (0, 1)  −0.066***  −0.067***  −0.068***  −0.041*  −0.041*  −0.042*  −0.023  −0.023  −0.023  (0.019)  (0.019)  (0.019)  (0.017)  (0.017)  (0.017)  (0.026)  (0.026)  (0.026)  Age (15–99)  0.005***  0.005***  0.005***  0.003***  0.003***  0.003***  0.003***  0.003***  0.003***  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  Education (0–8)  0.070***  0.070***  0.070***  0.069***  0.069***  0.068***  0.090***  0.089***  0.090***  (0.005)  (0.005)  (0.005)  (0.004)  (0.004)  (0.004)  (0.007)  (0.007)  (0.007)  Income (1–10)  0.042***  0.042***  0.042***  0.045***  0.045***  0.045***  0.045***  0.045***  0.045***  (0.005)  (0.005)  (0.005)  (0.004)  (0.004)  (0.004)  (0.007)  (0.007)  (0.007)  Life satisfaction (1–10)  0.054***  0.054***  0.054***  0.070***  0.070***  0.070***  0.072***  0.072***  0.072***  (0.005)  (0.005)  (0.005)  (0.005)  (0.005)  (0.005)  (0.007)  (0.007)  (0.007)  Political trust (1–4)  0.241***  0.241***  0.240***  0.240***  0.240***  0.239***  0.213***  0.213***  0.213***  (0.011)  (0.011)  (0.011)  (0.010)  (0.010)  (0.010)  (0.015)  (0.015)  (0.015)  Country-year characteristics  Education (societal mean)    0.266*  0.262    0.181  0.183    0.388*  0.392*    (0.135)  (0.135)    (0.118)  (0.118)    (0.189)  (0.189)  GDP per capita (in thousand US $)    0.020*  0.019*    0.026***  0.026***    0.028*  0.028*    (0.010)  (0.010)    (0.007)  (0.007)    (0.012)  (0.012)  Fearon’s ethnic diversity index (0–1)    0.007*  0.007*    0.001  0.001    0.001  0.002    (0.003)  (0.003)    (0.003)  (0.003)    (0.004)  (0.004)  Wave 2010–2014 (Ref. 2005–2009)  0.006  −0.271  −0.264  0.002  −0.297  −0.294  (Wave 2010–2014 only)      (0.244)  (0.207)  (0.208)  (0.209)  (0.189)  (0.189)      Constant  −3.992***  −4.736***  −4.861***  −4.181***  −4.283***  −4.342***  −3.884***  −5.669***  −5.731***  (0.448)  (0.787)  (0.789)  (0.360)  (0.645)  (0.646)  (0.500)  (1.178)  (1.179)  Random effect (country-year)  Variance  0.884***  0.566***  0.569***  0.747***  0.533***  0.534***  0.712***  0.472***  0.473***  (0.163)  (0.105)  (0.106)  (0.127)  (0.091)  (0.091)  (0.206)  (0.136)  (0.137)  N (individual)  73,917  73,917  73,917  87,737  87,737  87,737  33,971  33,971  33,971  N (country-year)  60  60  60  71  71  71  25  25  25  Wald χ2  1,255  1,293  1,353  1,722  1,757  1,773  671  687  697  AIC  72,436  72,415  72,352  89,332  89,314  89,299  37,054  37,050  37,040  Minority=  Belong to ethnic minority group  Speak non-dominant language at home  Self-identified as an immigrant    Model (1)  Model (2)  Model (3)  Model (4)  Model (5)  Model (6)  Model (7)  Model (8)  Model (9)  Ethnicity effect  Belong to minority group (0, 1)  −0.120***  −0.119***  0.473***  −0.039  −0.036  0.263***  −0.066  −0.066  0.379**  (0.023)  (0.023)  (0.077)  (0.024)  (0.024)  (0.075)  (0.048)  (0.048)  (0.139)  Democracy effect                    Democracy index (1–10)  0.149*  −0.048  −0.025  0.186***  −0.003  0.006  0.166*  −0.012  −0.006  (0.064)  (0.072)  (0.072)  (0.050)  (0.057)  (0.057)  (0.077)  (0.093)  (0.094)  Interaction effect                    Minority##Democracy      −0.093***      −0.051***      −0.070***      (0.012)      (0.012)      (0.021)  Individual characteristics  Female (0, 1)  −0.066***  −0.067***  −0.068***  −0.041*  −0.041*  −0.042*  −0.023  −0.023  −0.023  (0.019)  (0.019)  (0.019)  (0.017)  (0.017)  (0.017)  (0.026)  (0.026)  (0.026)  Age (15–99)  0.005***  0.005***  0.005***  0.003***  0.003***  0.003***  0.003***  0.003***  0.003***  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  Education (0–8)  0.070***  0.070***  0.070***  0.069***  0.069***  0.068***  0.090***  0.089***  0.090***  (0.005)  (0.005)  (0.005)  (0.004)  (0.004)  (0.004)  (0.007)  (0.007)  (0.007)  Income (1–10)  0.042***  0.042***  0.042***  0.045***  0.045***  0.045***  0.045***  0.045***  0.045***  (0.005)  (0.005)  (0.005)  (0.004)  (0.004)  (0.004)  (0.007)  (0.007)  (0.007)  Life satisfaction (1–10)  0.054***  0.054***  0.054***  0.070***  0.070***  0.070***  0.072***  0.072***  0.072***  (0.005)  (0.005)  (0.005)  (0.005)  (0.005)  (0.005)  (0.007)  (0.007)  (0.007)  Political trust (1–4)  0.241***  0.241***  0.240***  0.240***  0.240***  0.239***  0.213***  0.213***  0.213***  (0.011)  (0.011)  (0.011)  (0.010)  (0.010)  (0.010)  (0.015)  (0.015)  (0.015)  Country-year characteristics  Education (societal mean)    0.266*  0.262    0.181  0.183    0.388*  0.392*    (0.135)  (0.135)    (0.118)  (0.118)    (0.189)  (0.189)  GDP per capita (in thousand US $)    0.020*  0.019*    0.026***  0.026***    0.028*  0.028*    (0.010)  (0.010)    (0.007)  (0.007)    (0.012)  (0.012)  Fearon’s ethnic diversity index (0–1)    0.007*  0.007*    0.001  0.001    0.001  0.002    (0.003)  (0.003)    (0.003)  (0.003)    (0.004)  (0.004)  Wave 2010–2014 (Ref. 2005–2009)  0.006  −0.271  −0.264  0.002  −0.297  −0.294  (Wave 2010–2014 only)      (0.244)  (0.207)  (0.208)  (0.209)  (0.189)  (0.189)      Constant  −3.992***  −4.736***  −4.861***  −4.181***  −4.283***  −4.342***  −3.884***  −5.669***  −5.731***  (0.448)  (0.787)  (0.789)  (0.360)  (0.645)  (0.646)  (0.500)  (1.178)  (1.179)  Random effect (country-year)  Variance  0.884***  0.566***  0.569***  0.747***  0.533***  0.534***  0.712***  0.472***  0.473***  (0.163)  (0.105)  (0.106)  (0.127)  (0.091)  (0.091)  (0.206)  (0.136)  (0.137)  N (individual)  73,917  73,917  73,917  87,737  87,737  87,737  33,971  33,971  33,971  N (country-year)  60  60  60  71  71  71  25  25  25  Wald χ2  1,255  1,293  1,353  1,722  1,757  1,773  671  687  697  AIC  72,436  72,415  72,352  89,332  89,314  89,299  37,054  37,050  37,040  Standard errors in parentheses,* p < 0.05 ** p < 0.01 *** p < 0.001 Model 2 then adds country-specific macro-level controls for education, GDP, and ethnic diversity. As would be expected with this specification, the effect of ethnic minority continues to be negative and significant (−.119) but in this instance, the coefficient for democratization is no longer significant. Separate analyses (not shown) determined that this was the result of the inclusion of the GDP measure (see Acemoglu et al. [2014] on the relationship). Model 3 adds an interaction term between ethnic minority status and democratization, thereby allowing us to test whether the gap in trust between the majority and minority group varies based on levels of democratization. That is, this model tests whether the effect of ethnic minority is context specific and depends on the level of democratization. The results show that it does. The interaction term is negative, indicating that as the level of democratization increases, minority group members are less likely to have higher trust than majority group members. With the inclusion of the interaction term, the ethnic minority effect on trust becomes positive and significant. This suggests that it is only in democracies that ethnic minorities trust less than ethnic majority group members. In contrast, in non-democratic societies, members of ethnic minorities trust more than members of ethnic majorities. Models 4–6 replicate these findings with ethnicity referring to ethnic minority language, and models 7–9 replicate these findings with ethnicity referring to immigrant status. The results are very similar. The only difference is that the initial coefficient for ethnic minority is not significant, indicating that across countries there is no aggregate ethnic trust gap. Instead, the ethnic trust gap is only for the most and least on the democratic scale. Figures 7–9 provide a graphical representation of the relationship between ethnic majority and minority status for each of the 10 levels of democratization. These figures clearly show that, irrespective of the components of ethnicity (ethnic minority, minority language, immigration), there is a positive overall effect of democratization for both majority and minority groups, but that the relative gap between groups varies according to the political situation. In authoritarian countries, minority groups tend to trust less than majority groups, and as societies get more democratic, majority group members trust more than minority group members. Figures 7–9. View largeDownload slide Plots of the interaction between ethnicity and democratization Note: Yes denotes minority group, No denotes majority group, democratization from low (1) to high (10). Figures 7–9. View largeDownload slide Plots of the interaction between ethnicity and democratization Note: Yes denotes minority group, No denotes majority group, democratization from low (1) to high (10). Next, in table 5 we consider the effect of ethnic minority status within the four EIU regime types—authoritarian, hybrid democracies, flawed democracies, and full democracies. Recall that these regimes are distinguished based on the extent to which they facilitate or hinder, among others, basic political and civic freedoms, elections, open media, and political participation. For each regime type, we estimate three models—one for each type of ethnic minority status. Model 1 for authoritarian regimes shows that ethnic minority groups in authoritarian groups tend to trust more than majority group members. These results hold when ethnicity is coded as ethnic minority group (.200) (model 2) and based on minority language (.301) (model 3). The pattern for flawed democracies is more mixed and variable, depending on the coding of ethnic minority status. When it refers to ethnic minority, the effect is negative (−.272), but for minority language it is non-significant and for immigrant it is positive (.343). For flawed democracies, there is only a negative effect for immigrant minorities (−.375). Finally, for the full democracies we see that the effect is negative across all three codings of ethnic minority status (ethnic minority, −.452; linguistic minority, −.549; and immigrant, −.298).15 Table 5. Logistic regression models showing effect of ethnicity on trust within regime types Regime type:  Authoritarian  Hybrid democracy  Flawed democracy  Full democracy    Model (1)  Model (2)  Model (3)  Model (4)  Model (5)  Model (6)  Model (7)  Model (8)  Model (9)  Model (10)  Model (11)  Model (12)  Ethnicity effect                          Belong to minority group (0, 1)  0.200***      −0.272***      −0.000      −0.452***      (0.055)      (0.045)      (0.035)      (0.046)      Speak non-dominant language at home (0, 1)    0.301***      −0.081      0.036      −0.549***      (0.049)      (0.042)      (0.035)      (0.070)    Self-identified as an immigrant (0, 1)      −0.022      0.343***      −0.375**      −0.298***      (0.090)      (0.100)      (0.125)      (0.081)  Individual characteristics  Female (0, 1)  −0.041  0.027  0.057  −0.076*  −0.096**  −0.144**  −0.202***  −0.078*  −0.031  −0.011  −0.015  −0.021  (0.047)  (0.036)  (0.048)  (0.037)  (0.034)  (0.055)  (0.032)  (0.030)  (0.053)  (0.034)  (0.031)  (0.049)  Age (15–99)  0.007***  0.009***  0.002  0.002*  0.005***  0.002  0.005***  0.006***  −0.006***  0.011***  0.008***  0.011***  (0.002)  (0.001)  (0.002)  (0.001)  (0.001)  (0.002)  (0.001)  (0.001)  (0.002)  (0.001)  (0.001)  (0.001)  Education (0–8)  0.028*  0.035***  0.020  −0.019*  −0.015  0.038**  0.084***  0.070***  0.103***  0.185***  0.180***  0.193***  (0.012)  (0.010)  (0.013)  (0.009)  (0.008)  (0.014)  (0.008)  (0.008)  (0.016)  (0.010)  (0.008)  (0.014)  Income (1–10)  0.052***  0.039***  0.025  0.030**  0.072***  0.049**  −0.015*  0.003  −0.020  0.082***  0.075***  0.074***  (0.013)  (0.010)  (0.014)  (0.010)  (0.008)  (0.016)  (0.008)  (0.007)  (0.015)  (0.008)  (0.007)  (0.013)  Life satisfaction (1–10)  0.043***  0.072***  0.058***  0.066***  0.053***  0.104***  −0.029***  0.004  0.011  0.105***  0.129***  0.161***  (0.012)  (0.009)  (0.013)  (0.009)  (0.008)  (0.014)  (0.008)  (0.007)  (0.015)  (0.011)  (0.010)  (0.016)  Political trust (1–4)  0.295***  0.218***  0.090***  0.189***  0.166***  0.151***  0.214***  0.198***  0.129***  0.376***  0.421***  0.430***  (0.027)  (0.020)  (0.026)  (0.020)  (0.018)  (0.031)  (0.017)  (0.017)  (0.029)  (0.022)  (0.021)  (0.034)  Country-year characteristics  Education (societal mean)  −0.183***  0.088**  0.356***  0.320***  0.192***  1.929***  0.449***  0.313***  1.162***  0.036  0.045*  1.300***  (0.049)  (0.032)  (0.041)  (0.031)  (0.027)  (0.156)  (0.028)  (0.030)  (0.072)  (0.025)  (0.022)  (0.086)  GDP per capita (in thousand US $)  0.199***  0.037***  0.027***  0.013***  0.014***  0.030***  −0.001  0.025***  −0.065***  0.027***  0.030***  −0.014***  (0.010)  (0.002)  (0.005)  (0.001)  (0.001)  (0.002)  (0.003)  (0.002)  (0.006)  (0.002)  (0.001)  (0.004)  Fearon’s ethnic diversity index (0–1)  0.020***  0.013***  −0.006***  −0.002***  −0.007***  −0.002*  −0.001  −0.005***  0.009***  0.011***  0.010***  0.063***  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.000)  (0.001)  (0.001)  (0.001)  (0.003)  Wave 2010–2014 (Ref. 2005–2009)  −0.373***  −0.654***  NA  −0.754***  −0.292***  NA  −0.019  −0.227***  NA  −0.245***  −0.127***  NA  (0.069)  (0.051)  (0.052)  (0.045)  (0.043)  (0.037)  (0.041)  (0.035)  Constant  −4.408***  −4.465***  −3.623***  −2.829***  −2.402***  −13.844***  −3.948***  −3.588***  −6.511***  −6.596***  −6.970***  −18.588***  (0.262)  (0.174)  (0.279)  (0.148)  (0.140)  (0.938)  (0.163)  (0.164)  (0.377)  (0.190)  (0.170)  (0.745)  N  11,508  18,101  9,229  15,791  20,808  8,094  30,257  27,885  8,492  16,361  20,943  8,156  Regime type:  Authoritarian  Hybrid democracy  Flawed democracy  Full democracy    Model (1)  Model (2)  Model (3)  Model (4)  Model (5)  Model (6)  Model (7)  Model (8)  Model (9)  Model (10)  Model (11)  Model (12)  Ethnicity effect                          Belong to minority group (0, 1)  0.200***      −0.272***      −0.000      −0.452***      (0.055)      (0.045)      (0.035)      (0.046)      Speak non-dominant language at home (0, 1)    0.301***      −0.081      0.036      −0.549***      (0.049)      (0.042)      (0.035)      (0.070)    Self-identified as an immigrant (0, 1)      −0.022      0.343***      −0.375**      −0.298***      (0.090)      (0.100)      (0.125)      (0.081)  Individual characteristics  Female (0, 1)  −0.041  0.027  0.057  −0.076*  −0.096**  −0.144**  −0.202***  −0.078*  −0.031  −0.011  −0.015  −0.021  (0.047)  (0.036)  (0.048)  (0.037)  (0.034)  (0.055)  (0.032)  (0.030)  (0.053)  (0.034)  (0.031)  (0.049)  Age (15–99)  0.007***  0.009***  0.002  0.002*  0.005***  0.002  0.005***  0.006***  −0.006***  0.011***  0.008***  0.011***  (0.002)  (0.001)  (0.002)  (0.001)  (0.001)  (0.002)  (0.001)  (0.001)  (0.002)  (0.001)  (0.001)  (0.001)  Education (0–8)  0.028*  0.035***  0.020  −0.019*  −0.015  0.038**  0.084***  0.070***  0.103***  0.185***  0.180***  0.193***  (0.012)  (0.010)  (0.013)  (0.009)  (0.008)  (0.014)  (0.008)  (0.008)  (0.016)  (0.010)  (0.008)  (0.014)  Income (1–10)  0.052***  0.039***  0.025  0.030**  0.072***  0.049**  −0.015*  0.003  −0.020  0.082***  0.075***  0.074***  (0.013)  (0.010)  (0.014)  (0.010)  (0.008)  (0.016)  (0.008)  (0.007)  (0.015)  (0.008)  (0.007)  (0.013)  Life satisfaction (1–10)  0.043***  0.072***  0.058***  0.066***  0.053***  0.104***  −0.029***  0.004  0.011  0.105***  0.129***  0.161***  (0.012)  (0.009)  (0.013)  (0.009)  (0.008)  (0.014)  (0.008)  (0.007)  (0.015)  (0.011)  (0.010)  (0.016)  Political trust (1–4)  0.295***  0.218***  0.090***  0.189***  0.166***  0.151***  0.214***  0.198***  0.129***  0.376***  0.421***  0.430***  (0.027)  (0.020)  (0.026)  (0.020)  (0.018)  (0.031)  (0.017)  (0.017)  (0.029)  (0.022)  (0.021)  (0.034)  Country-year characteristics  Education (societal mean)  −0.183***  0.088**  0.356***  0.320***  0.192***  1.929***  0.449***  0.313***  1.162***  0.036  0.045*  1.300***  (0.049)  (0.032)  (0.041)  (0.031)  (0.027)  (0.156)  (0.028)  (0.030)  (0.072)  (0.025)  (0.022)  (0.086)  GDP per capita (in thousand US $)  0.199***  0.037***  0.027***  0.013***  0.014***  0.030***  −0.001  0.025***  −0.065***  0.027***  0.030***  −0.014***  (0.010)  (0.002)  (0.005)  (0.001)  (0.001)  (0.002)  (0.003)  (0.002)  (0.006)  (0.002)  (0.001)  (0.004)  Fearon’s ethnic diversity index (0–1)  0.020***  0.013***  −0.006***  −0.002***  −0.007***  −0.002*  −0.001  −0.005***  0.009***  0.011***  0.010***  0.063***  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.000)  (0.001)  (0.001)  (0.001)  (0.003)  Wave 2010–2014 (Ref. 2005–2009)  −0.373***  −0.654***  NA  −0.754***  −0.292***  NA  −0.019  −0.227***  NA  −0.245***  −0.127***  NA  (0.069)  (0.051)  (0.052)  (0.045)  (0.043)  (0.037)  (0.041)  (0.035)  Constant  −4.408***  −4.465***  −3.623***  −2.829***  −2.402***  −13.844***  −3.948***  −3.588***  −6.511***  −6.596***  −6.970***  −18.588***  (0.262)  (0.174)  (0.279)  (0.148)  (0.140)  (0.938)  (0.163)  (0.164)  (0.377)  (0.190)  (0.170)  (0.745)  N  11,508  18,101  9,229  15,791  20,808  8,094  30,257  27,885  8,492  16,361  20,943  8,156  Standard errors in parentheses,* p < 0.05 ** p < 0.01 *** p < 0.001 Table 5. Logistic regression models showing effect of ethnicity on trust within regime types Regime type:  Authoritarian  Hybrid democracy  Flawed democracy  Full democracy    Model (1)  Model (2)  Model (3)  Model (4)  Model (5)  Model (6)  Model (7)  Model (8)  Model (9)  Model (10)  Model (11)  Model (12)  Ethnicity effect                          Belong to minority group (0, 1)  0.200***      −0.272***      −0.000      −0.452***      (0.055)      (0.045)      (0.035)      (0.046)      Speak non-dominant language at home (0, 1)    0.301***      −0.081      0.036      −0.549***      (0.049)      (0.042)      (0.035)      (0.070)    Self-identified as an immigrant (0, 1)      −0.022      0.343***      −0.375**      −0.298***      (0.090)      (0.100)      (0.125)      (0.081)  Individual characteristics  Female (0, 1)  −0.041  0.027  0.057  −0.076*  −0.096**  −0.144**  −0.202***  −0.078*  −0.031  −0.011  −0.015  −0.021  (0.047)  (0.036)  (0.048)  (0.037)  (0.034)  (0.055)  (0.032)  (0.030)  (0.053)  (0.034)  (0.031)  (0.049)  Age (15–99)  0.007***  0.009***  0.002  0.002*  0.005***  0.002  0.005***  0.006***  −0.006***  0.011***  0.008***  0.011***  (0.002)  (0.001)  (0.002)  (0.001)  (0.001)  (0.002)  (0.001)  (0.001)  (0.002)  (0.001)  (0.001)  (0.001)  Education (0–8)  0.028*  0.035***  0.020  −0.019*  −0.015  0.038**  0.084***  0.070***  0.103***  0.185***  0.180***  0.193***  (0.012)  (0.010)  (0.013)  (0.009)  (0.008)  (0.014)  (0.008)  (0.008)  (0.016)  (0.010)  (0.008)  (0.014)  Income (1–10)  0.052***  0.039***  0.025  0.030**  0.072***  0.049**  −0.015*  0.003  −0.020  0.082***  0.075***  0.074***  (0.013)  (0.010)  (0.014)  (0.010)  (0.008)  (0.016)  (0.008)  (0.007)  (0.015)  (0.008)  (0.007)  (0.013)  Life satisfaction (1–10)  0.043***  0.072***  0.058***  0.066***  0.053***  0.104***  −0.029***  0.004  0.011  0.105***  0.129***  0.161***  (0.012)  (0.009)  (0.013)  (0.009)  (0.008)  (0.014)  (0.008)  (0.007)  (0.015)  (0.011)  (0.010)  (0.016)  Political trust (1–4)  0.295***  0.218***  0.090***  0.189***  0.166***  0.151***  0.214***  0.198***  0.129***  0.376***  0.421***  0.430***  (0.027)  (0.020)  (0.026)  (0.020)  (0.018)  (0.031)  (0.017)  (0.017)  (0.029)  (0.022)  (0.021)  (0.034)  Country-year characteristics  Education (societal mean)  −0.183***  0.088**  0.356***  0.320***  0.192***  1.929***  0.449***  0.313***  1.162***  0.036  0.045*  1.300***  (0.049)  (0.032)  (0.041)  (0.031)  (0.027)  (0.156)  (0.028)  (0.030)  (0.072)  (0.025)  (0.022)  (0.086)  GDP per capita (in thousand US $)  0.199***  0.037***  0.027***  0.013***  0.014***  0.030***  −0.001  0.025***  −0.065***  0.027***  0.030***  −0.014***  (0.010)  (0.002)  (0.005)  (0.001)  (0.001)  (0.002)  (0.003)  (0.002)  (0.006)  (0.002)  (0.001)  (0.004)  Fearon’s ethnic diversity index (0–1)  0.020***  0.013***  −0.006***  −0.002***  −0.007***  −0.002*  −0.001  −0.005***  0.009***  0.011***  0.010***  0.063***  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.000)  (0.001)  (0.001)  (0.001)  (0.003)  Wave 2010–2014 (Ref. 2005–2009)  −0.373***  −0.654***  NA  −0.754***  −0.292***  NA  −0.019  −0.227***  NA  −0.245***  −0.127***  NA  (0.069)  (0.051)  (0.052)  (0.045)  (0.043)  (0.037)  (0.041)  (0.035)  Constant  −4.408***  −4.465***  −3.623***  −2.829***  −2.402***  −13.844***  −3.948***  −3.588***  −6.511***  −6.596***  −6.970***  −18.588***  (0.262)  (0.174)  (0.279)  (0.148)  (0.140)  (0.938)  (0.163)  (0.164)  (0.377)  (0.190)  (0.170)  (0.745)  N  11,508  18,101  9,229  15,791  20,808  8,094  30,257  27,885  8,492  16,361  20,943  8,156  Regime type:  Authoritarian  Hybrid democracy  Flawed democracy  Full democracy    Model (1)  Model (2)  Model (3)  Model (4)  Model (5)  Model (6)  Model (7)  Model (8)  Model (9)  Model (10)  Model (11)  Model (12)  Ethnicity effect                          Belong to minority group (0, 1)  0.200***      −0.272***      −0.000      −0.452***      (0.055)      (0.045)      (0.035)      (0.046)      Speak non-dominant language at home (0, 1)    0.301***      −0.081      0.036      −0.549***      (0.049)      (0.042)      (0.035)      (0.070)    Self-identified as an immigrant (0, 1)      −0.022      0.343***      −0.375**      −0.298***      (0.090)      (0.100)      (0.125)      (0.081)  Individual characteristics  Female (0, 1)  −0.041  0.027  0.057  −0.076*  −0.096**  −0.144**  −0.202***  −0.078*  −0.031  −0.011  −0.015  −0.021  (0.047)  (0.036)  (0.048)  (0.037)  (0.034)  (0.055)  (0.032)  (0.030)  (0.053)  (0.034)  (0.031)  (0.049)  Age (15–99)  0.007***  0.009***  0.002  0.002*  0.005***  0.002  0.005***  0.006***  −0.006***  0.011***  0.008***  0.011***  (0.002)  (0.001)  (0.002)  (0.001)  (0.001)  (0.002)  (0.001)  (0.001)  (0.002)  (0.001)  (0.001)  (0.001)  Education (0–8)  0.028*  0.035***  0.020  −0.019*  −0.015  0.038**  0.084***  0.070***  0.103***  0.185***  0.180***  0.193***  (0.012)  (0.010)  (0.013)  (0.009)  (0.008)  (0.014)  (0.008)  (0.008)  (0.016)  (0.010)  (0.008)  (0.014)  Income (1–10)  0.052***  0.039***  0.025  0.030**  0.072***  0.049**  −0.015*  0.003  −0.020  0.082***  0.075***  0.074***  (0.013)  (0.010)  (0.014)  (0.010)  (0.008)  (0.016)  (0.008)  (0.007)  (0.015)  (0.008)  (0.007)  (0.013)  Life satisfaction (1–10)  0.043***  0.072***  0.058***  0.066***  0.053***  0.104***  −0.029***  0.004  0.011  0.105***  0.129***  0.161***  (0.012)  (0.009)  (0.013)  (0.009)  (0.008)  (0.014)  (0.008)  (0.007)  (0.015)  (0.011)  (0.010)  (0.016)  Political trust (1–4)  0.295***  0.218***  0.090***  0.189***  0.166***  0.151***  0.214***  0.198***  0.129***  0.376***  0.421***  0.430***  (0.027)  (0.020)  (0.026)  (0.020)  (0.018)  (0.031)  (0.017)  (0.017)  (0.029)  (0.022)  (0.021)  (0.034)  Country-year characteristics  Education (societal mean)  −0.183***  0.088**  0.356***  0.320***  0.192***  1.929***  0.449***  0.313***  1.162***  0.036  0.045*  1.300***  (0.049)  (0.032)  (0.041)  (0.031)  (0.027)  (0.156)  (0.028)  (0.030)  (0.072)  (0.025)  (0.022)  (0.086)  GDP per capita (in thousand US $)  0.199***  0.037***  0.027***  0.013***  0.014***  0.030***  −0.001  0.025***  −0.065***  0.027***  0.030***  −0.014***  (0.010)  (0.002)  (0.005)  (0.001)  (0.001)  (0.002)  (0.003)  (0.002)  (0.006)  (0.002)  (0.001)  (0.004)  Fearon’s ethnic diversity index (0–1)  0.020***  0.013***  −0.006***  −0.002***  −0.007***  −0.002*  −0.001  −0.005***  0.009***  0.011***  0.010***  0.063***  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.000)  (0.001)  (0.001)  (0.001)  (0.003)  Wave 2010–2014 (Ref. 2005–2009)  −0.373***  −0.654***  NA  −0.754***  −0.292***  NA  −0.019  −0.227***  NA  −0.245***  −0.127***  NA  (0.069)  (0.051)  (0.052)  (0.045)  (0.043)  (0.037)  (0.041)  (0.035)  Constant  −4.408***  −4.465***  −3.623***  −2.829***  −2.402***  −13.844***  −3.948***  −3.588***  −6.511***  −6.596***  −6.970***  −18.588***  (0.262)  (0.174)  (0.279)  (0.148)  (0.140)  (0.938)  (0.163)  (0.164)  (0.377)  (0.190)  (0.170)  (0.745)  N  11,508  18,101  9,229  15,791  20,808  8,094  30,257  27,885  8,492  16,361  20,943  8,156  Standard errors in parentheses,* p < 0.05 ** p < 0.01 *** p < 0.001 Figure 10 provides a visual illustration of the odds ratios for how each of the three ethnic minority measures affects trust in four types of regimes. Odds ratios are obtained from each of the models in table 5. Here it is worth focusing on the specific type of regime. The evidence for full democracies is robust—the effect of ethnic majority minority on trust is negative and therefore in the anticipated direction, and this effect occurs for, respectively, all types of ethnicity. The effect of majority-minority for the authoritarian countries is also robust—it is positive and occurs for two out of three types of ethnic majority-minority status. The effect in the flawed democracies is in line with expectations, but because it is only for immigrant minority it is slightly less robust. Finally, the effect for the hybrid democracies is somewhat of a puzzle. Whereas immigrant minorities trust more, those who belong to an ethnic minority group trust less. One possibility is that in comparison to linguistic and ethnic minority groups, the meaning of immigrant status is more variable across different types of societies. That is, it could be that there are some kind of immigrant self-selection processes occurring to different kinds of societies—with one kind of individual selecting to more democratic states and another to less democratic states. Given that these effects of immigrant minority status on trust remain—even after controls for some of the characteristics that we might expect to be associated with selection such as education and income are included—it would seem that some other process must be in play. Nevertheless, the fact that the gap in trust across regime types remains even when ethnicity refers to ethnic and linguistic minority rather than immigrant minority clearly shows that selection is not the primary cause. Figure 10. View largeDownload slide Odds ratios showing how three types of ethnicity affect trust in four regime types. Note: Above 1 denotes minority more trusting, below 1 majority more trusting. Figure 10. View largeDownload slide Odds ratios showing how three types of ethnicity affect trust in four regime types. Note: Above 1 denotes minority more trusting, below 1 majority more trusting. Finally, in order to test the mechanism underlying this gap, we considered whether self-placement on the class scale (X045) and perception of racist behavior in the neighborhood (H002_04) operate as mediators. This entailed estimating the same multilevel models with these potential mechanistic outcomes as dependent variables (see, e.g., Hanushek and Kimko 2000). The idea is that, due to their feelings of relative powerlessness compared to the majority group, minority group members (of all kinds) might be more sensitive to racist behavior in the neighborhood. The same logic holds for self-placement on the class scale. Of course, it should be noted that racist behavior in the neighborhood could be occurring even if it is targeting a different group, and that this is not a direct measure of individual-level discrimination. Therefore, because these are indirect and external correlates of the experience of relative powerlessness at best because there might be many other mechanisms at play, and because the WVS data does not contain any other measure that we could use to capture this concept more directly, these results should be understood as exploratory rather than confirmatory. The results are presented in Appendix C. For each outcome, we estimate three models: one for each type of ethnicity—ethnic minority, linguistic minority, and immigrant. The interaction between ethnic minority and democratization again captures the gap between the majority and minority group. The coefficient is positive, indicating that results showing that compared to less democratic societies, minority groups in democracies are more likely than the majority to see racist behavior in the neighborhood and to place themselves on the low end of the class scale than majority group members. That is, as would be expected under a situation of relative powerlessness, the gap between the majority and minority group increases in more democratic societies. Conclusion Research shows that when asked whether “most people can be trusted,” ethnic minorities tend to trust less (De Vroome, Hooghe, and Marien 2013; Smith 2010; Stolle and Harell 2013; Wilkes and Wu 2017). Not only could lower trust be an issue for individual members of ethnic minority groups, but the gap in trust could also matter more broadly for the growing number of countries seeing a rise in ethnic and racial diversity (e.g., see Bove and Elia 2017). The problem is that the vast majority of the evidence about low trust among ethnic minorities comes from studies conducted within democratic societies, and it has been well established that democratization increases trust (Delhey, Newton, and Welzel 2011; Inglehart 1999). How, then, can theoretical arguments about the benefits of democracy for trust be reconciled with the empirical evidence showing that there are many ethnic gaps in trust in democratic societies? Studies to date have been unable to answer this question because they have largely considered the relationship between ethnicity and trust in democratic countries. The research presented in this paper addresses this gap by considering the relationship between ethnicity and trust across political regime types. In so doing, we made two overall contributions to the study of the relationship between ethnicity and trust. The first contribution has been the development of a majority-minority argument that states that any effect of ethnicity is reflective of a relative power positioning within society as denoted by majority-minority status. That is, what matters is not someone’s specific language, religion, or other cultural marker but rather the relative positioning that these different groupings denote. This relative positioning becomes evident by considering the relationship between ethnic majority-minority status and trust under different macro-political contexts tied to democratization. Of relevance for ethnic gaps in trust is that by its very nature, democracy provides the potential for the tyranny of the majority (Brittain 1996; Gamble 1997). Thus, even if the structure governing overall inequality is fairer in democracies, democratic institutions are at times used to prevent the integration and success of minority groups (Bjørnskov 2008, 274–75). Democratic freedoms also increase the saliency and awareness of inequality (ibid.), thereby creating the conditions under which ethnic minority groups come to understand and link their status to differential treatment (Kymlicka 2002; Ziller 2017). In contrast, ethnic minority groups often dominate under authoritarian regimes. The majority-minority argument thus puts ethnic gaps in trust into global perspective. It allows for the separation of ethnic effects from political context effects and draws attention to the fact that ethnic gaps in trust might look very different under different types of political contexts. Ethnic differences in trust are not only reflective of migration of less trusting individuals from less trusting societies. And, while the argument clearly suggests that the gap is related to some kind of power imbalance possibly via discrimination, it does not assume that ethnic majorities are de facto more trusting. As the case of authoritarian societies illustrates, in some cases ethnic minorities are the beneficiaries of a power imbalance, and this shows in their higher relative trust levels. The second contribution has been the development of new micro-level majority-minority measures of ethnicity for use with global cross-national datasets such as the World Values Survey. Currently the only comparable cross-national individual measure of ethnicity in such datasets is immigration status (e.g., see WVS and Gallup World Polls; Helliwell, Wang, and Xu [2014]). Yet ethnicity, particularly in non-democratic countries, typically refers to many other groupings, including those based on race, cultural identification, and language. The challenge is that as it currently stands, the individual micro-level ethnicity measures in the WVS refer to different ethnic groups and different languages within countries. The majority-minority measures provided in this paper surmount this challenge. Rather than focusing on a particular categorical content such as White race or English language, which may be common to some but not all countries, we focus on the fact that individuals within groups occupy a relative position. What the development of this micro-level measure now means is that by including individual controls for factors such as education and income, it is now possible not only to control for group differences in cultural and economic resources such as income (e.g., see Baldwin and Huber 2010) but also to conduct global comparison on the basis of ethnicity. We then used these new measures to test the majority-minority argument with the World Values Survey. The results demonstrate that while democracy increases generalized trust across the board, the effect of ethnic majority-minority status depends on the level of democratization. In less democratic societies, there is either no trust gap, or at the authoritarian extreme, a situation where ethnic minority groups have higher trust than ethnic majority groups. In contrast, as the level of democratization increases, ethnic majority groups gain more trust relative to ethnic minority groups. The net result is that while democratization raises everyone’s trust, it also leads to a gap in trust that favors the majority group. Because this result holds irrespective of whether ethnic majority-minority status is defined on the basis of ethnic self-identification, language, or immigration, we therefore conclude that these gaps in trust are reflective of relative power positioning within societies. That being said, there are a number of avenues for further research. First, while this study considered far more countries than previous work, an important limitation is that there are still many countries without ethnicity data. As more countries in cross-national datasets such as the WVS collect information about ethnicity, the associations considered here should be revisited. Second is that the systematic and comprehensive testing of mechanisms will be a challenge for further research. Clearly, it is going to be difficult to identify singular mechanisms that can fully account for ethnic gaps in trust, particularly across societies. Certainly while the focus in this paper was on why ethnic minority groups trust more or less relative to ethnic majority groups, attention ought also to be paid to mechanisms that can explain the reverse. That is, the fact that ethnic majority groups might trust more or less in particular contexts (or at the very least acknowledging this possibility) should be considered. Finally, while the focus of this study was on generalized social trust, the mechanisms that explain the relationship between ethnicity and political and institutional trust should also be considered as, in this latter case, trust among minorities is often higher (Maxwell 2010). Footnotes 1 Our immigration measure is not new (see, e.g., Dinesen and Sønderskov 2015; Helliwell, Wang, and Xu 2014). 2 Dinesen and Sønderskov (2015) and Helliwell, Wang, and Xu (2014) have a measure of whether the respondent is an immigrant. Still, it remains the case that all ethnic minorities are not immigrants. 3 Ziller (2017) is an exception. He considers the relationship between government effectiveness and ethnic trust gaps within the European context (20 countries). His measure of government effectiveness is highly correlated with democratization insofar as the 11 countries in his study with the highest government effectiveness scores are all full democracies (e.g., Sweden, Switzerland, etc.) and the nine countries at the bottom(e.g., Hungary, Slovenia, etc.) are all flawed democracies (if one uses the EIU democratization categories—see the Methods section of this paper). His study shows that increases in government effectiveness are associated with increases in ethnic trust gaps in Europe. 4 There is a high rate of missing data on ethnicity (e.g., minority, immigrant) because the question used to define ethnicity was not asked in some countries or has responses indicating 100 percent in the majority group (see Appendix A). We removed these countries because ascribed status is difficult to predict using other variables. Therefore, there is no data for these countries rather than a variable that has missing data for some respondents within these countries. 5 Some countries use race and others use national origin; in some countries, the categories changed over time; others are missing or have no variation (e.g., 100 percent one group) or have too few cases (fewer than 30) for reliable estimates—there is no question that this measure is not without difficulties. Nevertheless, it is currently the only global measure of its kind available. The inclusion of multiple measures of majority-minority group status not subject to these limitations helps ensure that the overall conclusions of the study are robust. 6 The majority is not a statistical majority in some countries. 7 In most countries, because there are multiple minority groups, this means that numerically dominant minority groups will end up driving the results for the minority category. For example, the values in table 3 show that for the United States, Black African and Spanish/Hispanic are the predominant ethnic minority groups. Both would be coded as minority, and since they are about the same numerical size, the results would reflect both. This is an accurate reflection of reality, since previous research shows that both of these groups have lower trust than Whites. There is no particular minority group driving the results for Sweden, and the results for Taiwan would mainly reflect the Hakka group. Still, the process of collapsing across categories does mean that divergent patterns among some smaller groups could potentially be “washed out.” This is an issue with all categorical measures, particularly those of a binary nature. 8 There are two reasons why the linguistic minority and immigrant measure are not highly correlated overall. First, the immigrant status question was not asked in as many countries as the language question. Second, while in some countries immigrant and linguistic minority are highly correlated, this is not the case in others. That is, typically in many (but not all) immigrant-receiving democratic countries, immigrants are also minority language speakers. The numbers in Appendix A, for example, show that in the United States (2011) there are 247 immigrants and 214 minority language speakers and the tetrachoric correlation is .76. In contrast, in many of the non-democratic countries there are still a host of individuals who identify as linguistic minorities (e.g., Uzbekistan 2011) but very few immigrants. The terachoric correlation within Uzbekistan is .43. 9 We do not have a measure of country of origin of immigrants, and so the argument that immigrants coming from less trusting non-OECD countries (e.g., see Dinesen 2012, 2013) might explain some of the gap cannot be entirely discounted. 10 According to the EIU (2007–2014), the features of each regime type are as follows: “Full democracies: Countries in which basic political freedoms and civil liberties are respected. Media are independent and diverse. There is an effective system of checks and balances. The judiciary is independent and judicial decisions are enforced. Flawed democracies: These countries also have free and fair elections and basic civil liberties are respected. However, there are significant weaknesses in other aspects of democracy, including problems in governance, an underdeveloped political culture and low levels of political participation. Hybrid regimes: Elections have substantial irregularities that often prevent them from being both free and fair. Government pressure on opposition parties and candidates may be common. Corruption tends to be widespread and the rule of law is weak. Typically, there is harassment of and pressure on journalists, and the judiciary is not independent. Authoritarian regimes: Many countries in this category are outright dictatorships. Elections, if they do occur, are not free and fair. Media are typically state-owned or controlled by groups connected to the ruling regime. There is no independent judiciary” (abridged from 45–46). 11 Details about the EIU and Polity IV indices can be obtained at http://www.eiu.com/ and http://www.systemicpeace.org/polity/polity4.htm. The data on GDP was obtained from the World Bank data matched to the country-year and can be obtained at http://data.worldbank.org/indicator/NY.GDP.MKTP.CD. The data on country-specific ethnic diversity is from Fearon’s (2003) ethnic fractionalization index. 12 Because each country in the WVS has an N that is generalizable to that country and because the influence of country Ns is small, we do not weight the WVS data. Further, we do not also include year and country as random (see Schmidt-Catran and Fairbrother 2015) because only some countries have more than one year. Thus, our models are cross-sectional rather than longitudinal. As Fairbrother and Martin (2013) point out in their analysis of economic inequality and trust in US states, there could be relationships that exist across (cross-sectional) but not within (longitudinal) geographic units. The net result is that our models provide an estimation of the difference between countries at varying levels of democratization but they do not test whether the process of becoming more or less democratic will lead to rising or falling trust. 13 Some countries only have one wave of data included in our analysis. 14 Democracy is a country-year-level variable. The purpose of the interaction between ethnicity and democracy is to show that ethnicity has different effects (slopes) on trust across country-years with different levels of democratization. Therefore, we do not include a random slope for ethnicity in the models. This said, the results do not change when we run additional estimations of models with a random slope (not shown). 15 As there are a number of countries in the WVS dataset whereby the percentage ethnic minority shifted too much over time to be the result of “real” change (e.g., Cyprus, Malaysia, Morocco, Russia, Thailand, and Ukraine), we reestimated all models without these countries in order to ensure the robustness of the results. The results hold. About the Authors Rima Wilkes is Professor of Sociology at the University of British Columbia and an affiliate of the Laboratory for Comparative Research at the Higher School of Economics in Russia. Her research interests include race, ethnicity and Indigeneity, and political sociology. Recent publications have appeared in International Migration Review, Canadian Review of Sociology, Social Science and Medicine, and PNAS. Cary Wu is a PhD candidate in Sociology at the University of British Columbia, Canada. His research focuses on political sociology, urban sociology, and migration studies. His work has appeared in International Journal of Comparative Sociology, International Political Science Review, Geoforum, American Sociologist, and China Review. Supplementary Material Supplementary material is available at Social Forces online. References Acemoglu, D., S. Naidu, P. Restrepo, and J. A. Robinson. 2014. “Democracy Does Cause Growth.” Working Paper No. w20004, National Bureau of Economic Research. Alesina, A., and E. La Ferrara. 2002. “ Who Trusts Others?” Journal of Public Economics  85: 207– 34. Google Scholar CrossRef Search ADS   Bachmann, R. 2001. “ Trust, Power and Control in Trans-Organizational Relations.” Organization Studies  22( 2): 337– 65. Google Scholar CrossRef Search ADS   Bahry, D., M. Kosolapov, P. Kozyreva, and R. K. Wilson. 2005. “ Ethnicity and Trust: Evidence from Russia.” American Political Science Review  99( 4): 521– 32. Google Scholar CrossRef Search ADS   Baldwin, K., and J. D. Huber. 2010. “ Economic versus Cultural Differences: Forms of Ethnic Diversity and Public Goods Provision.” American Political Science Review  104( 4): 644– 62. Google Scholar CrossRef Search ADS   Bezemer, D., and R. Jong-A-Pin. 2013. “ Democracy, Globalization and Ethnic Violence.” Journal of Comparative Economics  41( 1): 108– 25. Google Scholar CrossRef Search ADS   Bienen, H. 2015. Kenya: The Politics of Participation and Control . Princeton, NJ: Princeton University Press. Google Scholar CrossRef Search ADS   Bjørnskov, C. 2008. “ Social Trust and Fractionalization: A Possible Reinterpretation.” European Sociological Review  24: 271– 83. Google Scholar CrossRef Search ADS   Bove, Vincenzo, and Leandro Elia. 2017. “ Migration, Diversity, and Economic Growth.” World Development  89: 227– 39. Google Scholar CrossRef Search ADS   Brittain, John C. 1996. “ Direct Democracy by the Majority Can Jeopardize the Civil Rights of Minority or Other Powerless Groups.” Annual Survey of American Literature  1: 69– 77. Catala, A. 2015. “ Democracy, Trust, and Epistemic Justice.” The Monist  98( 4): 424– 40. Google Scholar CrossRef Search ADS   Chua, A. 2004. World on Fire: How Exporting Free Market Democracy Breeds Ethnic Hatred and Global Instability . New York: Doubleday. Cook, K. S., and T. Yamagishi. 1992. “ Power in Exchange Networks: A Power-Dependence Formulation.” Social Networks  14( 3–4): 245– 65. Google Scholar CrossRef Search ADS   Cook, K. S., T. Yamagishi, C. Cheshire, R. Cooper, M. Matsuda, and R. Mashima. 2005. “ Trust Building via Risk Taking: A Cross-Societal Experiment.” Social Psychology Quarterly  68( 2): 121– 42. Google Scholar CrossRef Search ADS   Davis, D. R., and W. H. Moore. 1997. “ Ethnicity Matters: Transnational Ethnic Alliances and Foreign Policy Behavior.” International Studies Quarterly  41( 1): 171– 84. Google Scholar CrossRef Search ADS   De Vroome, T., M. Hooghe, and S. Marien. 2013. “ The Origins of Generalized and Political Trust among Immigrant Minorities and the Majority Population in the Netherlands.” European Sociological Review  29( 6): 1336– 50. Google Scholar CrossRef Search ADS   Delhey, J., and K. Newton. 2005. “ Predicting Cross-National Levels of Social Trust: Global Pattern or Nordic Exceptionalism?” European Sociological Review  21: 311– 27. Google Scholar CrossRef Search ADS   Delhey, J., K. Newton, and C. Welzel. 2011. “ How General Is Trust in ‘Most People’? Solving the Radius of Trust Problem.” American Sociological Review  76: 786– 807. Google Scholar CrossRef Search ADS   Dinesen, P. T. 2012. “ Parental Transmission of Trust or Perceptions of Institutional Fairness: Generalized Trust of Non-Western Immigrants in a High-Trust Society.” Comparative Politics  44: 273– 89. Google Scholar CrossRef Search ADS   ———. 2013. “ Where You Come From or Where You Live? Examining the Cultural and Institutional Explanation of Generalized Trust Using Migration as a Natural Experiment.” European Sociological Review  29: 114– 28. Google Scholar CrossRef Search ADS   Dinesen, P. T., and K. M. Sønderskov. 2015. “ Ethnic Diversity and Social Trust: Evidence from the Micro-Context.” American Sociological Review  80( 3): 550– 73. Google Scholar CrossRef Search ADS   Doerschler, P., and P. I. Jackson. 2012. “ Do Muslims in Germany Really Fail to Integrate? Muslim Integration and Trust in Public Institutions.” Journal of International Migration and Integration  13: 503– 23. Economist Intelligence Unit. 2007–2014. “Democracy Index 2007–2014.” Retrived May 20, 2016, from http://www.eiu.com/. Fairbrother, M. 2014. “ Two Multilevel Modeling Techniques for Analyzing Comparative Longitudinal Survey Datasets.” Political Science Research and Methods  2( 1): 119– 40. Google Scholar CrossRef Search ADS   Fairbrother, M., and I. W. Martin. 2013. “ Does Inequality Erode Social Trust? Results from Multilevel Models of US States and Counties.” Social Science Research  42( 2): 347– 60. Google Scholar CrossRef Search ADS   Farrell, H. 2004. “Trust, Distrust, and Power.” In Distrust , edited by Russell Hardin, 85– 105. New York: Russell Sage Foundation. Fearon, J. D. 2003. “ Ethnic and Cultural Diversity by Country.” Journal of Economic Growth  8( 2): 195– 222. Google Scholar CrossRef Search ADS   Gamble, B. S. 1997. “ Putting Civil Rights to a Popular Vote.” American Journal of Political Science  ( January) 41: 245– 69. Google Scholar CrossRef Search ADS   Grabb, E., R. Andersen, M. Hwang, and S. Milligan. 2009. “ Confidence in Political Institutions in Canada and the United States: Assessing the Interactive Role of Region and Race.” American Review of Canadian Studies  39( 4): 379– 97. Google Scholar CrossRef Search ADS   Guseva, A., and A. Rona-Tas. 2001. “ Uncertainty, Risk, and Trust: Russian and American Credit Card Markets Compared.” American Sociological Review  66: 623– 46. Google Scholar CrossRef Search ADS   Håkansson, P., and F. Sjöholm. 2007. “ Who Do You Trust? Ethnicity and Trust in Bosnia and Herzegovina.” Europe-Asia Studies  59: 961– 76. Google Scholar CrossRef Search ADS   Hanushek, E. A., and D. D. Kimko. 2000. “ Schooling, Labor-Force Quality, and the Growth of Nations.” American Economic Review  ( December) 90: 1184– 1208. Google Scholar CrossRef Search ADS   Helliwell, J. F., and S. Wang. 2010. “Trust and Well-Being.” Working Paper No. w15911, National Bureau of Economic Research. Helliwell, J. F., S. Wang, and J. Xu. 2014. “How Durable Are Social Norms? Immigrant Trust and Generosity in 132 Countries.” Working Paper No. w19855, National Bureau of Economic Research. Hetherington, M. J., and S. Globetti. 2002. “ Political Trust and Racial Policy Preferences.” American Journal of Political Science  46: 253– 75. Google Scholar CrossRef Search ADS   Hooghe, M., T. Reeskens, D. Stolle, and A. Trappers 2009. “ Ethnic Diversity and Generalized Trust in Europe: A Cross-National Multilevel Study.” Comparative Political Studies  42: 198– 223. Google Scholar CrossRef Search ADS   Horowitz, D. L. 1995. “ Democracy in Divided Societies.” Journal of Democracy  4: 18– 38. Google Scholar CrossRef Search ADS   Hwang, M. 2013. “In Canada We Trust? Understanding Ethno-Racial Variations in Social and Political Trust.” Doctoral dissertation, University of British Columbia. Inglehart, R. 1999. “Trust, Well-Being and Democracy.” In Democracy and Trust , edited by Mark Warren, 88– 120. Cambridge: Cambridge University Press. Google Scholar CrossRef Search ADS   Johnson, N. D., and A. Mislin. 2012. “ How Much Should We Trust the World Values Survey Trust Question?” Economics Letters  116: 210– 12. Google Scholar CrossRef Search ADS   Kekic, L. 2007. “ The Economist Intelligence Unit’s Index of Democracy.” Economist  21: 1– 11. Kesler, C., and I. Bloemraad. 2010. “ Does Immigration Erode Social Capital? The Conditional Effects of Immigration-Generated Diversity on Trust, Membership, and Participation across 19 Countries, 1981–2000.” Canadian Journal of Political Science/Revue canadienne de science politique  43( 2): 319– 47. Google Scholar CrossRef Search ADS   Knack, S., and P. Keefer. 1997. “ Does Social Capital Have an Economic Payoff? A Cross-Country Investigation.” Quarterly Journal of Economics  112( 4): 1251– 88. Google Scholar CrossRef Search ADS   Kotzian, P. 2011. “ Public Support for Liberal Democracy.” International Political Science Review  32: 23– 41. Google Scholar CrossRef Search ADS   Kymlicka, W. 2002. Contemporary Political Philosophy: An Introduction . Oxford: Oxford University Press. Lévi-Strauss, C. 1969. The Elementary Structures of Kinship  (No. 340). Boston: Beacon Press. Lewis, J. David, and A. J. Weigert. 1985. “ Trust as a Social Reality.” Social Forces  63( 4): 967– 85. Google Scholar CrossRef Search ADS   ———. 2012. “ The Social Dynamics of Trust: Theoretical and Empirical Research, 1985–2012.” Social Forces  91( 1): 25– 31. Google Scholar CrossRef Search ADS   Lewis, D., 2013. Direct democracy and minority rights: A critical assessment of the tyranny of the majority in the American states . New York: Routledge. Ljunge, M. 2014. “ Trust Issues: Evidence on the Intergenerational Trust Transmission among Children of Immigrants.” Journal of Economic Behavior & Organization  106: 175– 96. Google Scholar CrossRef Search ADS   Marshall, M.G., & Jaggers, K. ( 2000). Polity IV project: Political regime characteristics and transitions, 1800-1999. Unpublished manuscript, University of Maryland, Center for International Development and Conflict Management. Google Scholar Maxwell, R. 2010. “ Evaluating Migrant Integration: Political Attitudes Across Generations in Europe.” International Migration Review  44: 25– 52. Google Scholar CrossRef Search ADS   Mewes, J. 2012. “Inequality and Trust Re-Visited.” Working Paper, accessed June 16, 2016, at http://www.soc.umu.se/digitalAssets/103/103912_11_2012_mewes.pdf. ———. 2014. “ Gen(d)eralized Trust: Women, Work, and Trust in Strangers.” European Sociological Review  30: 373– 86. Google Scholar CrossRef Search ADS   Molm, L. D., M. M. Whitham, and D. Melamed. 2012. “ Forms of Exchange and Integrative Bonds: Effects of History and Embeddedness.” American Sociological Review  77( 1): 141– 65. Google Scholar CrossRef Search ADS   Nannestad, P. 2008. “ What Have We Learned about Generalized Trust, if Anything?” Annual Review of Political Science  11: 413– 36. Google Scholar CrossRef Search ADS   Nannestad, P., G. T. Svendsen, P. T. Dinesen, and K. M. Sønderskov. 2014. “ Do Institutions or Culture Determine the Level of Social Trust? The Natural Experiment of Migration from Non-Western to Western Countries.” Journal of Ethnic and Migration Studies  40( 4): 544– 65. Google Scholar CrossRef Search ADS   Nunkoo, R., and H. Ramkissoon. 2012. “ Power, Trust, Social Exchange and Community Support.” Annals of Tourism Research  39( 2): 997– 1023. Google Scholar CrossRef Search ADS   Okamoto, D., and G. C. Mora. 2014. “ Panethnicity.” Annual Review of Sociology  40: 219– 39. Google Scholar CrossRef Search ADS   Oskarsson, S., Svensson, T. and Öberg, P., 2009. Power, trust, and institutional constraints: Individual level evidence. Rationality and Society , 21( 2), 171– 195. Google Scholar CrossRef Search ADS   Öuberg, P., and T. Svensson. 2010. “ Does Power Drive Out Trust? Relations between Labour Market Actors in Sweden.” Political Studies  58( 1): 143– 66. Google Scholar CrossRef Search ADS   Patsiurko, N., J. L. Campbell, and J. A. Hall. 2012. “ Measuring Cultural Diversity: Ethnic, Linguistic and Religious Fractionalization in the OECD.” Ethnic and Racial Studies  35: 195– 217. Pokharel, R., 2013. Politics and problematics of the definition and categorization of ethnicity in Nepal. Bodhi: An Interdisciplinary Journal , 5( 1), 1– 15. Google Scholar CrossRef Search ADS   Prempeh, H. K. 2013. “ Constitutionalism, Ethnicity and Minority Rights in Africa: A Reply to Jeremie Gilbert.” International Journal of Constitutional Law  11( 2): 438– 43. Google Scholar CrossRef Search ADS   Putnam, R. D. 1993. “ What Makes Democracy Work?” National Civic Review  82: 101– 7. Google Scholar CrossRef Search ADS   ———. 2007. “ E Pluribus Unum: Diversity and Community in the Twenty-First Century: The 2006 Johan Skytte Prize Lecture.” Scandinavian Political Studies  30( 2): 137– 74. Google Scholar CrossRef Search ADS   Ridgeway, C. L., and L. Smith-Lovin. 1999. “ The Gender System and Interaction.” Annual Review of Sociology  25: 191– 216. Google Scholar CrossRef Search ADS   Roscigno, V. J. 2011. “ Power, Revisited.” Social Forces  90( 2): 349– 74. Google Scholar CrossRef Search ADS   Rothstein, B. 2011. The Quality of Government: Corruption, Social Trust, and Inequality in International Perspective . Chicago: University of Chicago Press. Google Scholar CrossRef Search ADS   Rothstein, B. and Stolle, D., 2001, September. Social capital and street-level bureaucracy: An institutional theory of generalized trust. In ESF Conference Social Capital: Interdisciplinary Perspectives, Exeter, UK September 15-20. Prepared for the “Trust in Government Conference” at the Center for the Study of Democratic Rothstein, B. and D. Stolle. 2008. “ The State and Social Capital: An Institutional Theory of Generalized Trust.” Comparative Politics  7: 441– 59. Google Scholar CrossRef Search ADS   Rothstein, B. and E. M. Uslaner. 2005. “ All for All: Equality, Corruption, and Social Trust.” World Politics  58: 41– 72. Google Scholar CrossRef Search ADS   Rotter, J. B. 1971. “ Generalized Expectancies for Interpersonal Trust.” American Psychologist  26: 443– 52. Google Scholar CrossRef Search ADS   Schilke, O., M. Reimann, and K. S. Cook. 2015. “ Power Decreases Trust in Social Exchange.” Proceedings of the National Academy of Sciences  112: 12950– 55. Google Scholar CrossRef Search ADS   Schmidt-Catran, A. W., and M. Fairbrother. 2015. “ The Random Effects in Multilevel Models: Getting Them Wrong and Getting Them Right.” European Sociological Review  32( 1): 23– 38. Google Scholar CrossRef Search ADS   Smith, S. S. 2010. “ Race and Trust.” Annual Review of Sociology  36: 453– 75. Google Scholar CrossRef Search ADS   Soroka, S., J. F. Helliwell, and R. Johnston. 2003. “Measuring and Modelling Trust.” In Diversity, Social Capital and the Welfare State , edited by F. Kay and D. Johnston, 279– 303 Vancouver: UBC Press. Stolle, D., and A. Harell. 2013. “ Social Capital and Ethno‐Racial Diversity: Learning to Trust in an Immigrant Society.” Political Studies  61: 42– 66. Google Scholar CrossRef Search ADS   Sturgis, P., I. Brunton-Smith, S. Read, and N. Allum. 2011. “ Does Ethnic Diversity Erode Trust? Putnam’s ‘Hunkering Down’ Thesis Reconsidered.” British Journal of Political Science  41( 1): 57– 82. Google Scholar CrossRef Search ADS   Tilly, C. 1998. Durable Inequalities . Berkeley: University of California Press. Tilly, C., 2001. Mechanisms in political processes. Annual review of political science , 4( 1), 21– 41. Google Scholar CrossRef Search ADS   Tsai, M.C., Laczko, L. and Bjørnskov, C., 2011. Social diversity, institutions and trust: A cross-national analysis. Social Indicators Research , 101( 3), 305– 322. Google Scholar CrossRef Search ADS   Uslaner, E. M. 2008. “ Where You Stand Depends upon Where Your Grandparents Sat: The Inheritability of Generalized Trust.” Public Opinion Quarterly  72: 725– 40. Google Scholar CrossRef Search ADS   Uslaner, E. M., and M. Brown. 2005. “ Inequality, Trust, and Civic Engagement.” American Politics Research  33: 868– 94. Google Scholar CrossRef Search ADS   Wilkes, R. 2011. “ Re-Thinking the Decline in Trust: A Comparison of Black and White Americans.” Social Science Research  40( 6): 1596– 1610. Google Scholar CrossRef Search ADS   Wilkes, R., and C. Wu. 2017. “Minorities and Trust.” In Oxford Handbook of Social and Political Trust , edited by Eric Uslaner, 231– 49. New York: Oxford University Press. World Values Survey, Wave 5. 2005–2009. Official Aggregate v.20140429. World Values Survey Association, www.worldvaluessurvey.org. Aggregate File Producer: Asep/JDS, Madrid, Spain. World Values Survey, Wave 6. 2010–2014. Official Aggregate v.20150418. World Values Survey Association, www.worldvaluessurvey.org. Aggregate File Producer: Asep/JDS, Madrid, Spain. Wu, C., and R. Wilkes. 2016. “ Durable Power and Generalized Trust in Everyday Social Exchange.” Proceedings of the National Academy of Sciences USA  113( 11): E1417. Google Scholar CrossRef Search ADS   ———. 2017. “The Stats Bear It Out: In Quebec, Trust Is Low.” Maclean’s. http://www.macleans.ca/news/canada/the-stats-bear-it-out-in-quebec-trust-is-low/. ———. 2018. “ Finding Critical Trusters: A Response Pattern Model of Political Trust.” International Journal of Comparative Sociology . doi:10.1177/0020715218761520 Yamagishi, T. 2011. Trust: The Evolutionary Game of Mind and Society . Tokyo: Springer. Yip, P. S., and E. W. Tsang. 2007. “ Interpreting Dummy Variables and Their Interaction Effects in Strategy Research.” Strategic Organization  5: 13– 30. Google Scholar CrossRef Search ADS   Ziller, C. 2015. “ Ethnic Diversity, Economic and Cultural Contexts, and Social Trust: Cross-Sectional and Longitudinal Evidence from European Regions, 2002–2010.” Social Forces  93( 3): 1211– 40. Google Scholar CrossRef Search ADS   ———. 2017. “ Equal Treatment Regulations and Ethnic Minority Social Trust.” European Sociological Review  33( 4): 563– 75. Google Scholar CrossRef Search ADS   Author notes We thank Sylvia Fuller, Howard Ramos, and Rengin Firat for helpful comments on previous versions of this paper. We also thank the anonymous reviewers and Social Forces editor Arne Kalleberg. The article was prepared within the framework of a subsidy granted to the HSE by the Government of the Russian Federation for the implementation of the Russian Academic Excellence Project “5–100.” Support was also provided by Riksbankens Jubileumsfonds: the Swedish Foundation for Humanities and Social Sciences, project no. NHS14-2035:1, and by a grant from the Social Sciences and Humanities Research Council of Canada. Address correspondence to Rima Wilkes, University of British Columbia and Laboratory for Comparative Social Research, National Research University Higher School of Economics, Russian Federation, Department of Sociology, University of British Columbia, Vancouver, BC, Canada; phone: +604 822-6855; e-mail: wilkesr@mail.ubc.ca. © The Author(s) 2018. Published by Oxford University Press on behalf of the University of North Carolina at Chapel Hill. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Social Forces Oxford University Press

Ethnicity, Democracy, Trust: A Majority-Minority Approach

Social Forces , Volume Advance Article – Apr 13, 2018

Loading next page...
 
/lp/ou_press/ethnicity-democracy-trust-a-majority-minority-approach-BYOIjhow4J
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the University of North Carolina at Chapel Hill. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
0037-7732
eISSN
1534-7605
D.O.I.
10.1093/sf/soy027
Publisher site
See Article on Publisher Site

Abstract

Abstract Why do ethnic and racialized minorities have lower trust? While previous research emphasizes individual factors such as the national and cultural origins of ethnic groups, this paper draws attention to the ethnic majority-minority relationship. We argue that ethnic differences in trust are a function of the power dynamics underlying this relationship and that these dynamics are particularly salient in democratic political systems. To test this argument, we develop new measures of ethnic majority-minority status, which for the first time allows for global cross-national comparison of heterogeneous ethnic groups at the micro level. Using the World Values Survey, we test the majority-minority argument, showing that, while democracy increases generalized trust across the board, it also leads to a gap in trust that favors the majority group. This gap remains even after the inclusion of controls for country differences in factors such as ethnic diversity and GDP. Introduction Recent tensions with Muslims in France, ongoing Eastern European resistance to Syrian and North African refugees, and racially motivated crimes in Scandinavia remind us that in many countries ethnicity and race continue to be a major axis of social, economic, and political inequality. These challenges are often accompanied by a crisis of trust—seen through a significant difference in the trust in generalized others shown by ethnic minority and majority populations. Indeed, studies of trust in European countries as well as Canada and the United States clearly show that multiple measures of ethnicity (including race, national origin, and language) predict generalized trust, that is, the belief that most people can be trusted. One explanation of these differences focuses on the fact that minority groups are often immigrants. Differences in trust stem from the fact that immigrants have lower levels of economic and social capital and because they often come from less democratic countries, where trust tends to be scarce (De Vroome, Hooghe, and Marien 2013; Dinesen 2012, 2013; Uslaner 2008). However, ethnic gaps in trust cannot simply be attributed to immigration from less democratic to more democratic societies. As illustrated by the cases of Black Americans in the United States and people of color and Quebecers in Canada, members of minority groups born in democratic societies also have lower trust (Smith 2010; Stolle and Harell 2013; Wilkes 2011). This presents a puzzle. Democracy is widely held to be beneficial for trust across the board, including for ethnic minorities (Rothstein 2011; Rothstein and Uslaner 2005). How, then, can theoretical arguments about the benefits of democracy for trust be reconciled with the evidence of ethnic gaps in trust in democratic societies? The argument presented in this paper is that, in addition to the role of individual characteristics and place of origin, the power denoted in majority-minority group position needs consideration. That is, while ethnicity denotes particular socio-cultural and/or racial categories, these categories also reflect the experience of political, economic, and social power. Power is an important precursor to trust (Catala 2015; Yamagishi 2011). We further posit that this experience of majority-minority position is likely to be more salient in democratic societies than in non-democratic societies. By definition, the majority group has more potential for power, especially political power, relative to the minority in democratic systems. In non-democratic societies, as there is no principle of majority rule, either the power gap between the majority and minority is less significant or else the situation is even reversed—the minority group holds political and economic power. A test of this argument requires considering ethnic groups in both democratic and non-democratic societies. Studies to date have yet to do so, focusing mostly on comparisons among largely democratic societies (Kotzian 2011; Mewes 2012; Ziller 2017). The lack of attention to non-democratic societies might be because of a theoretical interest in European immigration and race relations in Western societies. This lack of attention is compounded by a major empirical challenge to doing so. Globally, different countries use different categories to denote ethnicity. The United States, for example, tends to use race-based categories of identification such as White, Black, and Asian, whereas in Ghana the major ethnic groups include Ga Afangbe, Akan, Ewe, Dagbani, and Guan. These differences mean there is little way to compare ethnic groups across countries except in terms of macro-level diversity. Here we present new micro-level measures of majority-minority status based on differences stemming from ethnic self-identification, language, and immigration that allow us to surmount this challenge.1 We provide a detailed description of how to create these new measures with the World Values Survey data. The use of these measures means that this paper provides the first global cross-national comparison of ethnic groups at the micro level. We then use the new majority-minority measures to analyze the relationship between ethnicity, trust, and democratization. The results show that democracy does condition the effect of ethnic minority-majority position on trust. Although everyone trusts more in democracies, the majority-minority gap in trust is also larger. These results hold irrespective of whether majority-minority group status is defined on the basis of ethnic majority-minority self-identification, majority-minority language, or immigration status. Trust, Ethnicity, and Democratization Trust is often understood as “a generalized expectancy held by an individual or a group that the word, promise, verbal, or written statement of another individual or group can be relied on” (Rotter 1971, 444). Trust in unknown others of this kind is of vital importance to the interactions between citizens, their leaders, and their institutions. It is seen as foundational to all aspects of society, operating as a form of social glue (Lewis and Weigert 1985, 2012; see also Wu and Wilkes 2016) and therefore leads to important societal and political benefits, among them personal and social well-being (Helliwell and Wang 2010), policy support (Hetherington and Globetti 2002), and political participation (Uslaner and Brown 2005). For this reason, trust is one of the most central concerns across the social sciences (Nannestad 2008; Smith 2010). Survey research shows that in Europe responses to the widely used question indicating generalized trust—whether “most people can be trusted”—often vary on the basis of immigration status, with immigrants less trusting than the native born (Doerschler and Jackson 2012; Hooghe et al. 2009). Cross-European studies show a trust gap between immigrants and non-immigrant groups more generally (Kotzian 2011; Mewes 2012, 2014), as do country-specific studies of Denmark (Bjørnskov 2008) and the Netherlands (De Vroome, Hooghe, and Marien 2013) (see also Dinesen [2012, 2013]; Ljunge [2014]; Nannestad et al. [2014]; Ziller [2015]). Several studies have attributed the lower trust of immigrant groups, especially those coming to Europe, to the fact that they came from less democratic societies that engender lower human capital as well as a culture of distrust (Dinesen 2012, 2013; De Vroome, Hooghe, and Marien 2013; Ljunge 2014; Ziller 2015). In democratic systems, laws are put in place to ensure that different groups of citizens are treated relatively equally, that there is a distribution of resources subject to “binding consultation,” and that there is protection “against arbitrary action by government agents” (Tilly 2001, 31). This is why factors such as the length of democracy, institutional effectiveness, and institutional impartiality all increase trust (Rothstein and Stolle 2008; Rothstein and Uslaner 2005). In contrast, in less democratic societies corruption and nepotism rule the day (ibid.). This explains why generalized trust is, on average, higher in more democratic countries, with the Scandinavian countries—Sweden, Norway, Denmark—having the highest mean global trust levels (Delhey and Newton 2005; Rothstein and Stolle 2001). Writing about ethnicity, Tilly (1998, 243) says that democracy “attenuates the performance effects of previously categorically differentiated experiences and reduces the transmission of categorically organized advantages from generation to generation.” The scope of who can be trusted is expanded and, hence, ethnic outgroup trust tends to be much higher in democratic societies (Delhey, Newton, and Welzel 2011). The problem is that we also see ethnic gaps in trust in democratic and non-democratic societies that cannot be attributed to migration from less democratic societies (Putnam 1993; Smith 2010; Uslaner 2008). Indeed, in democratic societies these ethnic gaps in trust exist even though democratic institutions are supposed to ensure fairness and equality. Blacks in the United States trust less than White Americans (Smith 2010). Indigenous people and visible minorities have lower trust than non-Indigenous people and non-visible minorities (Hwang 2013; Stolle and Harell 2013). Quebecers have lower trust than the rest of Canada (Grabb et al. 2009; Wu and Wilkes 2017). Elsewhere, the research also shows that Croats are more trusting than Bosniaks in Bosnia and Herzegovina (Håkansson and Sjöholm 2007) and that, in the Russian Federation, Russians have higher trust in most people than do Tatars or Yakuts (Bahry et al. 2005). Our argument is that, in addition to considering the role of democracy, the relative power denoted in majority-minority group position needs consideration. Indeed, since trust is an interpersonal construct and a social relationship (Rotter 1971; Yamagishi 2011), power is one of its important precursors (e.g., Bachmann 2001; Cook et al. 2005; Farrell 2004; Nunkoo and Ramkissoon 2012; Schilke, Reimann, and Cook 2015; Yamagishi 2011, 26). Power matters because it regulates risk and uncertainty (Guseva and Rona-Tas 2001; Lévi-Strauss 1969; Molm, Whitham, and Melamed 2012). More powerful people tend to trust more because they are more capable of handling the negative consequences should the trust relationship falter (Nunkoo and Ramkissoon 2012; Oskarsson et al. 2009; Öuberg and Svensson 2010; Wu and Wilkes 2016; Yamagishi 2011). As Cook et al. (2005, 140–42) write, power disparities create “fertile ground for distrust” and “commonly block the possibility of trust’” (see also Nunkoo and Ramkissoon 2012, 1005; Öuberg and Svensson [2010, 145]). While much of the social exchange literature on power and trust treats power as a characteristic of a specific and relatively short-term exchange relationship (Cook and Yamagishi 1992; Molm, Whitham, and Melamed 2012; Schilke, Reimann, and Cook 2015), the dynamics associated with ethnicity are a more enduring form of inequality (Tilly 1998). That is, while ethnicity denotes socio-cultural and/or racial categories of identification, these categories also reflect the experience of political, economic, and social power (Okamoto and Mora 2014; Pokharel 2013, 1) and the “valuations of the actors involved” (Ridgeway and Smith-Lovin 1999; Roscigno 2011, 352). We further argue that the majority-minority relationship is unique in democracies because, by definition, the majority group in democracies is likely to have more power relative to the minority group. This is not to say that actual powerlessness on the part of minorities is greater in democracies. Rather, the argument is that while democracy does increase individual political rights, democratic systems must also contend with the “tyranny of the majority” whereby the democratic process enables the majority to dominate the minority (Brittain 1996; Gamble 1997). Although there are clearly checks and balances in place (Lewis 2013), its potential may nevertheless be a major reason why minority groups in democracies have lower social trust than the majority group. At the group level, cases in democratic societies such as Black Pete in the Netherlands, the Washington Redskins in the United States, and Tintin (Belgium) with its depiction of Congolese people lead to “a deep divide between majority and minority groups” in terms of how they understand social practice and this, in turn, could affect trust (Catala 2015, 424). This relative imbalance is accompanied and reflective of heightened expectations around equality that come with democracies. As Kymlicka writes, in today’s democracies, members of historically subordinated groups demand equality, and demand it as a right. They “believe they are entitled to equality, and entitled to it now, not in some indefinite or millenarian future” (2002, 8). Therefore, while rights are accorded on an individual basis, the ideology of equality may also create a heightened expectation of fair treatment on the part of minority groups (Ziller 2017). In contrast, in less democratic systems the power manifested in the ethnic majority-minority relationships is often quite different. Expectations around fairness and equality may not exist in less democratic societies, where policies are top down and do little to mitigate the advantage held by some ethnic minority groups (Chua 2004; Horowitz 1995). The majority ethnic group does not de facto have more power. The Chinese in Malaysia, for example, have been economically advantaged, and the Sunnis in Lebanon have occupied positions of political privilege (Davis and Moore 1997). Certainly in many African countries “minority-group elites, with the backing of their ethnic kin or of ethnic coalitions in which their own group was dominant, have often been disproportionately represented or influential players in many of Africa’s authoritarian regimes” (Prempeh 2013, 441). Thus it is that advantaged minorities such as the Kikuyu in Kenya, despite comprising a relatively small proportion of the overall population, have produced three out of four presidents and have dominated the local economy (Bezemer and Jong-A-Pin 2013; Bienen 2015). Taken together, these points suggest that the nature of the majority-minority relationship is likely to be very different in democratic societies—although everyone will trust more in democratic societies, the gap in trust may also be larger. A test of this argument requires comparing individual members of majority and minority ethnic groups in democracies with individual members of majority and minority ethnic groups in other kinds of political systems. Therefore, we need to be able to consider both individuals as well as political systems. Table 1 illustrates three general frameworks—macro; micro; and micro-macro—currently used to study ethnicity and trust. We use table 1 to illustrate that with only a few exceptions, current research has yet to consider individual ethnic minority groups across political systems. Table 1. Frameworks for the study of the relationship between ethnicity, democracy, and trust, showing limited number of studies in fourth column Framework  Macro  Micro  Micro-macro  Sample research questions  Is average trust higher in more democratic countries?  Does ethnicity affect trust?  Is there an effect of ethnicity on trust controlling for state/country-level factors?  Does the effect of ethnicity on trust vary in democratic and non-democratic societies?  Dependent variable           Trust  yes  yes  yes  yes  Independent variables           Ethnicity            Individual-level ethnicity  X  yes  Xa  yes    Regional/country-level ethnic diversity  yes  X  Xa  yes   Democracy  yes  X  Xa  yes  Comparative dimension  yes  X  yes  yes  Sample studies using framework  Knack and Keefer (1997); Bjørnskov (2008); Tsai et al. (2011); Ziller (2015)  Bahry et al. (2005); Håkansson and Sjöholm (2007); Wilkes (2011); Dinesen (2012)  Putnam (2007); Uslaner (2008); Sturgis et al. (2011); Dinesen and Sønderskov (2015)  Ziller (2017) (European countries)  Framework  Macro  Micro  Micro-macro  Sample research questions  Is average trust higher in more democratic countries?  Does ethnicity affect trust?  Is there an effect of ethnicity on trust controlling for state/country-level factors?  Does the effect of ethnicity on trust vary in democratic and non-democratic societies?  Dependent variable           Trust  yes  yes  yes  yes  Independent variables           Ethnicity            Individual-level ethnicity  X  yes  Xa  yes    Regional/country-level ethnic diversity  yes  X  Xa  yes   Democracy  yes  X  Xa  yes  Comparative dimension  yes  X  yes  yes  Sample studies using framework  Knack and Keefer (1997); Bjørnskov (2008); Tsai et al. (2011); Ziller (2015)  Bahry et al. (2005); Håkansson and Sjöholm (2007); Wilkes (2011); Dinesen (2012)  Putnam (2007); Uslaner (2008); Sturgis et al. (2011); Dinesen and Sønderskov (2015)  Ziller (2017) (European countries)  aFor those studies that include both individual and macro-level variables, the limitation is that if there is a measure of individual-level ethnicity, then it is typically a within-country study, whereas if there are multiple countries, then the individual ethnicity measure is either absent or limited. Table 1. Frameworks for the study of the relationship between ethnicity, democracy, and trust, showing limited number of studies in fourth column Framework  Macro  Micro  Micro-macro  Sample research questions  Is average trust higher in more democratic countries?  Does ethnicity affect trust?  Is there an effect of ethnicity on trust controlling for state/country-level factors?  Does the effect of ethnicity on trust vary in democratic and non-democratic societies?  Dependent variable           Trust  yes  yes  yes  yes  Independent variables           Ethnicity            Individual-level ethnicity  X  yes  Xa  yes    Regional/country-level ethnic diversity  yes  X  Xa  yes   Democracy  yes  X  Xa  yes  Comparative dimension  yes  X  yes  yes  Sample studies using framework  Knack and Keefer (1997); Bjørnskov (2008); Tsai et al. (2011); Ziller (2015)  Bahry et al. (2005); Håkansson and Sjöholm (2007); Wilkes (2011); Dinesen (2012)  Putnam (2007); Uslaner (2008); Sturgis et al. (2011); Dinesen and Sønderskov (2015)  Ziller (2017) (European countries)  Framework  Macro  Micro  Micro-macro  Sample research questions  Is average trust higher in more democratic countries?  Does ethnicity affect trust?  Is there an effect of ethnicity on trust controlling for state/country-level factors?  Does the effect of ethnicity on trust vary in democratic and non-democratic societies?  Dependent variable           Trust  yes  yes  yes  yes  Independent variables           Ethnicity            Individual-level ethnicity  X  yes  Xa  yes    Regional/country-level ethnic diversity  yes  X  Xa  yes   Democracy  yes  X  Xa  yes  Comparative dimension  yes  X  yes  yes  Sample studies using framework  Knack and Keefer (1997); Bjørnskov (2008); Tsai et al. (2011); Ziller (2015)  Bahry et al. (2005); Håkansson and Sjöholm (2007); Wilkes (2011); Dinesen (2012)  Putnam (2007); Uslaner (2008); Sturgis et al. (2011); Dinesen and Sønderskov (2015)  Ziller (2017) (European countries)  aFor those studies that include both individual and macro-level variables, the limitation is that if there is a measure of individual-level ethnicity, then it is typically a within-country study, whereas if there are multiple countries, then the individual ethnicity measure is either absent or limited. The first column shows the macro-level framework. A typical research question for this framework would be “Is average trust higher in more democratic societies?” To answer this question requires comparing the average levels of trust across places—these could be national and/or regional—and assessing whether places with greater democratization have higher or lower trust (e.g., see Knack and Keefer 1997; Delhey and Newton 2005; Bjørnskov 2008; Ziller 2015). This approach can consider the impact of place-based ethnic diversity via ethnic fractionalization measures (Alesina and La Ferrara 2002). Macro-level measures of ethnic fractionalization can only denote the level of ethnic diversity within societies. They contain no information about economic or cultural differences between groups (Baldwin and Huber 2010). The macro-level measures of ethnic BGI (between group (income) inequality) and CLF (linguistic cultural fractionalization) developed by Baldwin and Huber are a highly innovative solution to this problem, though at present these remain limited to 46 countries. What the macro-level approach cannot do, however, is consider the individual-level impact of ethnicity on trust. Such a task has, however, been taken up by micro-level research, the framework of which is illustrated in the second column. Here the focus is on a host of individual-level variables and how they affect trust, and key among these is ethnicity. This research clearly documents ethnic gaps in trust within societies (e.g., see Bahry et al. 2005; Dinesen 2012; Håkansson and Sjöholm 2007; Wilkes 2011). Information as to the individual-level effects of ethnicity is limited in that these studies focus on a single country—therefore, we have little sense of how country-specific factors may or may not matter. The third column shows the micro-macro framework. In this framework there is a focus on both individual micro-level characteristics such as age and education as well as macro-level characteristics such as geographically based ethnic diversity. Here a typical question might be “Do individuals show differences in trust even when they live in different contexts?” or “Does place-based ethnic diversity affect individual-level trust?” (see, e.g., Dinesen and Sønderskov 2015; Putnam 2007; Sturgis et al. 2011; Uslaner 2008). While this approach typically has a measure of individual ethnicity and/or of democracy and diversity, only one study in the European context (Ziller 2017) has a measure of both. That is, as illustrated in the fourth column, while there are a few studies that consider whether ethnic diversity conditions the effect of ethnicity (e.g., Soroka, Helliwell, and Johnston [2003] consider the interaction between the level of diversity in a census tract and visible minority status in Canada) by and large these remain constrained to single-country studies. Those that have more country-level information (e.g., Kesler and Bloemraad 2010) typically do not have an individual-level measure of ethnicity.2 Ultimately, table 1 shows that there is little research that has simultaneously considered ethnicity, democratization, and trust.3 These concepts—and associated variables—are needed in order to consider the majority-minority argument. One reason why no study has done so to date might be that such an endeavor requires an individual-level measure of ethnicity that is comparable across countries so that country-level differences in democratization can also be included. This remains a major challenge because currently there is no way to measure ethnicity cross-nationally with categories that are commensurable across countries—at least not globally. That is, in many cases the standard markers of ethnicity—for example ethnic self-identification or language—are incommensurable across most countries. In, for example, the World Values Survey data, the ethnic self-identification categories for the United States include White/Caucasian, White Hispanic, Black African, Other, and Mixed. In Mexico the groups include colored (light), colored (dark), and white. Singapore’s categories include Chinese, South Asian, Malay, and Eurasian. Kyrgyzstan’s categories include Kyrgyz, Russian, Uzbek, and Turkish. While these ethnic categories are reflective of the classifications that countries believe are important (Patsiurko, Campbell, and Hall 2012), they cannot be matched across countries. One solution to this problem is to make use of a measure such as visible minority status or foreign born. This solution might only be relevant to certain national contexts. Dinesen and Sønderskov (2015) (Denmark) and Helliwell, Wang, and Xu (2014) (global) have a measure of whether the respondent is an immigrant. Still, it remains the case that ethnic minorities, particularly in non-democratic societies, are not all immigrants. In the following discussion of the empirical application of the majority-minority approach, we provide a solution to this problem. Data and Variables The primary focal measures are trust, ethnicity (majority-minority), and democratization. Most of these measures are from the most recent two waves of the World Values Survey (2005–2009, 2010–2014)—variables from other sources are appended. Table 2 provides the documentation, coding, and summary statistics for all variables used in the analyses.4 Table 2. Measures of key variables in analysis Variable  Survey question  Coding  Mean (%)  Individual level         Trust  WVS (A165): Most people can be trusted or you can’t be too careful  Binary (0, 1)  25.89   Majority-minority  WVS (X051): Belong to the non-dominant group in society  Binary (0, 1)  19.34  WVS (G016): Speak non-dominant language at home  Binary (0, 1)  16.36  WVS (G027A): Self-identified as an immigrant  Binary (0, 1)  4.55   Age  WVS (X003): in years  in years, 15–99  41.78   Gender  WVS (X001): 1 = female, 0 = male  Binary (0, 1)  52.18   Income  WVS (X047): Scale of income  Scale (1–10)  4.74   Life satisfaction  WVS (A170): Satisfaction with your life  Scale (1–10)  6.77   Education  WVS (X025): Highest educational level attained  Scale (1–8)  4.81   Political trust  WVS (E069_11) Confidence in the government  Scale (1–4)  2.442  Macro level         Democratization  Economist Intelligence Unit Democracy Index; Authoritarian: (1–3.99), Hybrid democracy: (4–5.99), Flawed democracy: 6–7.99), Full democracy: (8–10)  Index (1–10)  6.08   Ethnic fractionalization  Fearon (2003) ethnic fractionalization index  Index (0–1)  0.426   Gross Domestic Product  World Bank GDP per capita  Per capita constant dollars ($245–$83,556)  $16,330  Variable  Survey question  Coding  Mean (%)  Individual level         Trust  WVS (A165): Most people can be trusted or you can’t be too careful  Binary (0, 1)  25.89   Majority-minority  WVS (X051): Belong to the non-dominant group in society  Binary (0, 1)  19.34  WVS (G016): Speak non-dominant language at home  Binary (0, 1)  16.36  WVS (G027A): Self-identified as an immigrant  Binary (0, 1)  4.55   Age  WVS (X003): in years  in years, 15–99  41.78   Gender  WVS (X001): 1 = female, 0 = male  Binary (0, 1)  52.18   Income  WVS (X047): Scale of income  Scale (1–10)  4.74   Life satisfaction  WVS (A170): Satisfaction with your life  Scale (1–10)  6.77   Education  WVS (X025): Highest educational level attained  Scale (1–8)  4.81   Political trust  WVS (E069_11) Confidence in the government  Scale (1–4)  2.442  Macro level         Democratization  Economist Intelligence Unit Democracy Index; Authoritarian: (1–3.99), Hybrid democracy: (4–5.99), Flawed democracy: 6–7.99), Full democracy: (8–10)  Index (1–10)  6.08   Ethnic fractionalization  Fearon (2003) ethnic fractionalization index  Index (0–1)  0.426   Gross Domestic Product  World Bank GDP per capita  Per capita constant dollars ($245–$83,556)  $16,330  Note: WVS, World Values Survey (2005–2008, 2010–2014). Table 2. Measures of key variables in analysis Variable  Survey question  Coding  Mean (%)  Individual level         Trust  WVS (A165): Most people can be trusted or you can’t be too careful  Binary (0, 1)  25.89   Majority-minority  WVS (X051): Belong to the non-dominant group in society  Binary (0, 1)  19.34  WVS (G016): Speak non-dominant language at home  Binary (0, 1)  16.36  WVS (G027A): Self-identified as an immigrant  Binary (0, 1)  4.55   Age  WVS (X003): in years  in years, 15–99  41.78   Gender  WVS (X001): 1 = female, 0 = male  Binary (0, 1)  52.18   Income  WVS (X047): Scale of income  Scale (1–10)  4.74   Life satisfaction  WVS (A170): Satisfaction with your life  Scale (1–10)  6.77   Education  WVS (X025): Highest educational level attained  Scale (1–8)  4.81   Political trust  WVS (E069_11) Confidence in the government  Scale (1–4)  2.442  Macro level         Democratization  Economist Intelligence Unit Democracy Index; Authoritarian: (1–3.99), Hybrid democracy: (4–5.99), Flawed democracy: 6–7.99), Full democracy: (8–10)  Index (1–10)  6.08   Ethnic fractionalization  Fearon (2003) ethnic fractionalization index  Index (0–1)  0.426   Gross Domestic Product  World Bank GDP per capita  Per capita constant dollars ($245–$83,556)  $16,330  Variable  Survey question  Coding  Mean (%)  Individual level         Trust  WVS (A165): Most people can be trusted or you can’t be too careful  Binary (0, 1)  25.89   Majority-minority  WVS (X051): Belong to the non-dominant group in society  Binary (0, 1)  19.34  WVS (G016): Speak non-dominant language at home  Binary (0, 1)  16.36  WVS (G027A): Self-identified as an immigrant  Binary (0, 1)  4.55   Age  WVS (X003): in years  in years, 15–99  41.78   Gender  WVS (X001): 1 = female, 0 = male  Binary (0, 1)  52.18   Income  WVS (X047): Scale of income  Scale (1–10)  4.74   Life satisfaction  WVS (A170): Satisfaction with your life  Scale (1–10)  6.77   Education  WVS (X025): Highest educational level attained  Scale (1–8)  4.81   Political trust  WVS (E069_11) Confidence in the government  Scale (1–4)  2.442  Macro level         Democratization  Economist Intelligence Unit Democracy Index; Authoritarian: (1–3.99), Hybrid democracy: (4–5.99), Flawed democracy: 6–7.99), Full democracy: (8–10)  Index (1–10)  6.08   Ethnic fractionalization  Fearon (2003) ethnic fractionalization index  Index (0–1)  0.426   Gross Domestic Product  World Bank GDP per capita  Per capita constant dollars ($245–$83,556)  $16,330  Note: WVS, World Values Survey (2005–2008, 2010–2014). Trust is measured with the question “Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people?” coded on a 0–1 scale (1 denotes trust) for the WVS data (see, e.g., Fairbrother [2014]; Hooghe et al. [2009]; and Johnson and Mislin [2012] on the use of this measure). Ethnic majority-minority group status is coded in three ways: based on ethnic self-identification (X025), language use (language spoken at home, G016), and immigrant group status (self-identification as an immigrant, G027A). While self-identification as an immigrant is comparable across countries, this is not the case with either the ethnic self-identification or language spoken at home measures. Different countries have different sets of categories of ethnic self-identification and home language.5 There is no consistent overlap across countries for these categories. Using multiple different measures of ethnic majority-minority status further ensures the robustness of our argument. Table 3 illustrates this problem using the example of the United States, Sweden, and Taiwan. For each country, the table shows the WVS country categories for ethnic group and language at home. The US ethnic categories include Black/African, Spanish/Hispanic, White/Caucasian, Other, and Mixed Races; Sweden’s include White/Caucasian and Asian- Central (Arabic), Asian-East (Chinese, Japanese), Asian-South (Indian, Hindu, Pakistani), and Black African; and Taiwan’s categories includes Minnanese from Taiwan, Hakka from Taiwan, China, and Aborigine. It might be argued that White/Caucasian is common, but this is only trust for the United States and Sweden, not for Taiwan. The language variable also shows little overlap: the United States has six languages, Sweden has seven, and Taiwan has five. Further is that while English (spoken in the United States and Sweden) might be one of the world’s global languages, not only is it irrelevant in the Taiwanese context, but the meaning of speaking English is clearly different in Sweden than it is in the United States. Table 3. Example of the majority-minority coding strategy (WVS 2010–2014), majority in bold United States  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  Black African  127  10.17  No answer  33  1.48  Spanish; Hispanic  124  9.93  Chinese  7  0.31  White/Caucasian White  922  73.82  English  2,016  90.32  Other  36  2.88  French  2  0.09  Mixed races  40  3.2  Japanese  3  0.13        Spanish; Castilian  145  6.5        Other  26  1.16  Total  1,249  100  Total  2,232  100  Sweden  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  Missing; Not specified  7  0.58  Missing  1  0.08  Don´t know  16  1.33  No answer  2  0.17  Asian-Central (Arabic)  28  2.32  Chinese  1  0.08  Asian-East (Chinese, Japanese)  7  0.58  English  13  1.08  Asian-South (Indian, Hindu, Pakistani)  30  2.49  Finnish  3  0.25  Black African  8  0.66  French  2  0.17  White/Caucasian White  1,091  90.46  Spanish; Castilian  9  0.75  Other  19  1.58  Swedish  1,114  92.37        Other  61  5.06  Total  1,206  100  Total  1,206  100  Taiwan  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  No answer  7  0.57  No answer  2  0.16  Don´t know  10  0.81  Don´t know  1  0.08  Other  39  3.15  Mandarin  528  42.65  TW: Hakka from Taiwan  110  8.89  Other  51  4.12  Minnanese from Taiwan  987  79.73  TW: Taiwanese (Minnanese)  628  50.73  TW: China  73  5.9  TW: Hakka  27  2.18  TW: Aborigine  12  0.97  TW: Aborigine language  1  0.08  Total  1,238  100  Total  1,238  100  United States  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  Black African  127  10.17  No answer  33  1.48  Spanish; Hispanic  124  9.93  Chinese  7  0.31  White/Caucasian White  922  73.82  English  2,016  90.32  Other  36  2.88  French  2  0.09  Mixed races  40  3.2  Japanese  3  0.13        Spanish; Castilian  145  6.5        Other  26  1.16  Total  1,249  100  Total  2,232  100  Sweden  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  Missing; Not specified  7  0.58  Missing  1  0.08  Don´t know  16  1.33  No answer  2  0.17  Asian-Central (Arabic)  28  2.32  Chinese  1  0.08  Asian-East (Chinese, Japanese)  7  0.58  English  13  1.08  Asian-South (Indian, Hindu, Pakistani)  30  2.49  Finnish  3  0.25  Black African  8  0.66  French  2  0.17  White/Caucasian White  1,091  90.46  Spanish; Castilian  9  0.75  Other  19  1.58  Swedish  1,114  92.37        Other  61  5.06  Total  1,206  100  Total  1,206  100  Taiwan  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  No answer  7  0.57  No answer  2  0.16  Don´t know  10  0.81  Don´t know  1  0.08  Other  39  3.15  Mandarin  528  42.65  TW: Hakka from Taiwan  110  8.89  Other  51  4.12  Minnanese from Taiwan  987  79.73  TW: Taiwanese (Minnanese)  628  50.73  TW: China  73  5.9  TW: Hakka  27  2.18  TW: Aborigine  12  0.97  TW: Aborigine language  1  0.08  Total  1,238  100  Total  1,238  100  Table 3. Example of the majority-minority coding strategy (WVS 2010–2014), majority in bold United States  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  Black African  127  10.17  No answer  33  1.48  Spanish; Hispanic  124  9.93  Chinese  7  0.31  White/Caucasian White  922  73.82  English  2,016  90.32  Other  36  2.88  French  2  0.09  Mixed races  40  3.2  Japanese  3  0.13        Spanish; Castilian  145  6.5        Other  26  1.16  Total  1,249  100  Total  2,232  100  Sweden  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  Missing; Not specified  7  0.58  Missing  1  0.08  Don´t know  16  1.33  No answer  2  0.17  Asian-Central (Arabic)  28  2.32  Chinese  1  0.08  Asian-East (Chinese, Japanese)  7  0.58  English  13  1.08  Asian-South (Indian, Hindu, Pakistani)  30  2.49  Finnish  3  0.25  Black African  8  0.66  French  2  0.17  White/Caucasian White  1,091  90.46  Spanish; Castilian  9  0.75  Other  19  1.58  Swedish  1,114  92.37        Other  61  5.06  Total  1,206  100  Total  1,206  100  Taiwan  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  No answer  7  0.57  No answer  2  0.16  Don´t know  10  0.81  Don´t know  1  0.08  Other  39  3.15  Mandarin  528  42.65  TW: Hakka from Taiwan  110  8.89  Other  51  4.12  Minnanese from Taiwan  987  79.73  TW: Taiwanese (Minnanese)  628  50.73  TW: China  73  5.9  TW: Hakka  27  2.18  TW: Aborigine  12  0.97  TW: Aborigine language  1  0.08  Total  1,238  100  Total  1,238  100  United States  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  Black African  127  10.17  No answer  33  1.48  Spanish; Hispanic  124  9.93  Chinese  7  0.31  White/Caucasian White  922  73.82  English  2,016  90.32  Other  36  2.88  French  2  0.09  Mixed races  40  3.2  Japanese  3  0.13        Spanish; Castilian  145  6.5        Other  26  1.16  Total  1,249  100  Total  2,232  100  Sweden  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  Missing; Not specified  7  0.58  Missing  1  0.08  Don´t know  16  1.33  No answer  2  0.17  Asian-Central (Arabic)  28  2.32  Chinese  1  0.08  Asian-East (Chinese, Japanese)  7  0.58  English  13  1.08  Asian-South (Indian, Hindu, Pakistani)  30  2.49  Finnish  3  0.25  Black African  8  0.66  French  2  0.17  White/Caucasian White  1,091  90.46  Spanish; Castilian  9  0.75  Other  19  1.58  Swedish  1,114  92.37        Other  61  5.06  Total  1,206  100  Total  1,206  100  Taiwan  Ethnic group  Freq.  Percent  Language at home  Freq.  Percent  No answer  7  0.57  No answer  2  0.16  Don´t know  10  0.81  Don´t know  1  0.08  Other  39  3.15  Mandarin  528  42.65  TW: Hakka from Taiwan  110  8.89  Other  51  4.12  Minnanese from Taiwan  987  79.73  TW: Taiwanese (Minnanese)  628  50.73  TW: China  73  5.9  TW: Hakka  27  2.18  TW: Aborigine  12  0.97  TW: Aborigine language  1  0.08  Total  1,238  100  Total  1,238  100  However, if we consider whether these categories denote ethnic majority-minority status, they can be compared in a meaningful way. We aggregated each variable to the country level and then used the country percentages to identify the numeric majority group and language.6 In table 3, the majority group is bolded. Those individuals who, for example, speak Swedish in Sweden and English in the United States and the UK are all majority group members. We did the same for ethnic self-identification that then allows us to compare ethnic groups that may be different “races” or other groups. Individuals in the most numerous group were coded as majority, and all others were coded as minority.7Appendix A provides the list of country (years) and the frequency count for each majority-minority measure. The correlations/tetrachoric correlations across these variables—ethnic majority-minority self-identification and majority-minority language (.153, .22), ethnic majority-minority self-identification, and immigrant status (.205, .40), and between majority-minority language and immigrant status (.031, .08)—are small, indicating that each of these variables are not the same measure.8 We also include a number of individual controls—gender (X001), age (X003), education (X025), income (V239), life satisfaction (A170), and political trust (E069_11).9 At the macro level, we measure democratization in two ways with the 10-point Economist Intelligence Unit scale (see http://www.eiu.com/home.aspx). First, the EIU codes countries on five dimensions—liberties, government functioning, political participation, political culture, and electoral process and participation—that are measured with 60 indicators (Kekic 2007). It then assigns each country an overall score of 1–10. We matched the year of the EIU scale to the specific country-year values. The exception are the 2005/2006 country years where there was no EIU data, and therefore we use 2007 (the EIU scale values change little from year to year). Second, the EIU also groups countries into one of four categories based on their overall score—Authoritarian (1–3.99), Hybrid Democracy (4–5.99), Flawed Democracy (6–7.99), Full Democracy (8–10)—and we use these categories in the latter portion of the analyses as a way to consider effects within regime types (see also Wu and Wilkes 2018).10 Results hold if the EIU scale is replaced with the Polity IV scale (with −10 denoting hereditary monarchy and +10 denoting consolidated democracy). This scale also rates countries on the basis of competitiveness of executive recruitment, openness of executive recruitment, constraints on chief executive, regulation of participation, and competitiveness of participation (Marshall et al. 2000), but we elected not to use it because it is less up to date. We also include country-specific mean education level, Gross Domestic Product (GDP), and ethnic fractionalization.11 Statistical Models Because the WVS trust measure is a binary 0–1 outcome, we use mixed-effects logistic regression models of individuals nested within country-years.12 All models also include individual controls. The elements of the models considered are expressed as follows:   log(Pij(T=1)1−Pij(T=1))=α0+α0j+α1Mij+α2Dj+α3Mij*Dj+α4Cij+α5CCj (2)where Pij represents the probability of being trusting for the ith individual in the jth country or country-year.13 The fixed effects are α0– α4, and α0j is the random intercept at the country-year/country level. Our key predictor Mij represents being a member of a minority group,14 and Dj denotes the democracy index of the jth country. We consider the interaction between ethnicity and democracy, which is denoted by Mij*Dj, and α3 is the coefficient. Cij denotes a series of individual-level control variables (age, gender, education, etc.), and α4 is a vector of corresponding coefficients. CCj denotes the list of country-year- (education, GDP per capita) and country- (ethnic fractionalization) level controls, and α5 is a vector of corresponding coefficients. Country-specific ethnic fractionalization is not expected to change significantly over a four-year period. Since we use the base specification approach to interactions with dummy variables, the interaction term refers to the difference in the slope effect of democratization between the majority and minority ethnic group rather than to the group-specific slope effect of democratization (Yip and Tsang 2007). Therefore, in a model with an ethnic minority dummy variable, democratization, and an interaction between ethnic minority and democratization, the coefficient for ethnic minority refers to the effect of being a minority overall in the less democratic countries. The coefficient for democratization refers to the overall effect of democratization. The coefficient for the interaction term captures the gaps between the majority-minority trust gaps at different levels of democracy (refer to figures 1–3—for example, the interaction term is the gap between the data points on these graphs). Figures 1–3. View largeDownload slide Plots showing country gaps in trust between majority-minority by level of democratization Note: Positive gap denotes higher majority group trust. Figures 1–3. View largeDownload slide Plots showing country gaps in trust between majority-minority by level of democratization Note: Positive gap denotes higher majority group trust. Results Descriptive We begin by presenting the bivariate correlation between the country-specific majority-minority trust gap and democratization. Since trust is coded as a 0, 1 binary, this means the gap denotes the difference between the proportion of the minority who trust and the proportion of the majority who trust. A positive gap means that the majority group trusts more, and a negative gap means that the minority group trusts more. In 2005, for example, 11 percent of WVS respondents (n = 111) in the UK self-identified as an ethnic minority while 89 percent (n = 911) did not. The proportion who trust among the minority group is therefore 25 percent (28/111), and the proportion who trust among the majority group is 31 percent (283/911)—resulting in a trust gap of 6 percent (= 0.31 – 0.25). Appendix B provides the country-specific gaps for each coding of majority-minority status (e.g., ethnic minority gap, language gap, immigrant gap). Figures 1–3 show the country-specific gaps plotted by the level of democratization. Each figure represents a different coding of ethnic minority status based on, respectively, ethnic identity, language, and immigration. Because there are so many countries, for readability, we denote countries with dots rather than their country codes. Each dot refers to a country-specific gap in trust—positive dot indicate that the gap favors the majority group, and negative dots indicate that the gap favors the minority group. The fitted values line shows an upward trend, indicating that the gap increases (to favoring the majority group) as the level of democratization increases. In order to further illustrate the pattern in figures 4–6, we take these country-specific gaps in trust and consider them based on our categorical indicator of democratization that codes countries as authoritarian, hybrid, flawed, and full democracies. Figures 4–6. View largeDownload slide Boxplots showing gaps in trust by political regime type Note: Positive gap denotes higher majority group trust. Figures 4–6. View largeDownload slide Boxplots showing gaps in trust by political regime type Note: Positive gap denotes higher majority group trust. Figure 4 shows the boxplots for ethnic minority status for each type of government. The line in the middle of the box denotes the median trust gap, the outer edges of the box show the upper and lower quartiles, and the farthest lines show the maximum and minimum trust gaps (excluding outliers, which are denoted by the black dots). Essentially what these boxplots show is a negative or non-existent gap in authoritarian, hybrid, and flawed democracies (denoting that in terms of the median there is either no gap or that ethnic minorities trust more) and a positive gap (denoting higher majority group trust) in the fully democratic societies. Figures 5 and 6 also show a similar pattern—there is either no gap or a gap in favor of minorities in non-full democracies and a gap that favors the majority group in democratic societies (e.g., majority group members are more trusting). Multivariate Next we present the results of the mixed-effects logistic regression models. Table 4 presents the coefficients (log-odds) and standard errors from these models. For each measure of ethnic majority minority (self-identified as an ethnic minority, speak non-dominant language at home = linguistic minority, and not born in the country = immigrant), we estimate three models. The first model tests the main effects of ethnic majority-minority status and democratization by including a dummy variable for minority status and the democratization scale measure. It also includes all the individual controls. This model allows us to test whether there is an ethnic majority-minority gap in trust across the board and whether trust increases with democratization. The results show that, across the board, there is a negative and statistically significant effect of ethnic minority status on trust (−.120), meaning that ethnic minority group members trust less. The coefficient for democratization is positive (.149), indicating that trust increases with democratization. Table 4. Multilevel models showing effect of ethnicity and democratization on trust Minority=  Belong to ethnic minority group  Speak non-dominant language at home  Self-identified as an immigrant    Model (1)  Model (2)  Model (3)  Model (4)  Model (5)  Model (6)  Model (7)  Model (8)  Model (9)  Ethnicity effect  Belong to minority group (0, 1)  −0.120***  −0.119***  0.473***  −0.039  −0.036  0.263***  −0.066  −0.066  0.379**  (0.023)  (0.023)  (0.077)  (0.024)  (0.024)  (0.075)  (0.048)  (0.048)  (0.139)  Democracy effect                    Democracy index (1–10)  0.149*  −0.048  −0.025  0.186***  −0.003  0.006  0.166*  −0.012  −0.006  (0.064)  (0.072)  (0.072)  (0.050)  (0.057)  (0.057)  (0.077)  (0.093)  (0.094)  Interaction effect                    Minority##Democracy      −0.093***      −0.051***      −0.070***      (0.012)      (0.012)      (0.021)  Individual characteristics  Female (0, 1)  −0.066***  −0.067***  −0.068***  −0.041*  −0.041*  −0.042*  −0.023  −0.023  −0.023  (0.019)  (0.019)  (0.019)  (0.017)  (0.017)  (0.017)  (0.026)  (0.026)  (0.026)  Age (15–99)  0.005***  0.005***  0.005***  0.003***  0.003***  0.003***  0.003***  0.003***  0.003***  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  Education (0–8)  0.070***  0.070***  0.070***  0.069***  0.069***  0.068***  0.090***  0.089***  0.090***  (0.005)  (0.005)  (0.005)  (0.004)  (0.004)  (0.004)  (0.007)  (0.007)  (0.007)  Income (1–10)  0.042***  0.042***  0.042***  0.045***  0.045***  0.045***  0.045***  0.045***  0.045***  (0.005)  (0.005)  (0.005)  (0.004)  (0.004)  (0.004)  (0.007)  (0.007)  (0.007)  Life satisfaction (1–10)  0.054***  0.054***  0.054***  0.070***  0.070***  0.070***  0.072***  0.072***  0.072***  (0.005)  (0.005)  (0.005)  (0.005)  (0.005)  (0.005)  (0.007)  (0.007)  (0.007)  Political trust (1–4)  0.241***  0.241***  0.240***  0.240***  0.240***  0.239***  0.213***  0.213***  0.213***  (0.011)  (0.011)  (0.011)  (0.010)  (0.010)  (0.010)  (0.015)  (0.015)  (0.015)  Country-year characteristics  Education (societal mean)    0.266*  0.262    0.181  0.183    0.388*  0.392*    (0.135)  (0.135)    (0.118)  (0.118)    (0.189)  (0.189)  GDP per capita (in thousand US $)    0.020*  0.019*    0.026***  0.026***    0.028*  0.028*    (0.010)  (0.010)    (0.007)  (0.007)    (0.012)  (0.012)  Fearon’s ethnic diversity index (0–1)    0.007*  0.007*    0.001  0.001    0.001  0.002    (0.003)  (0.003)    (0.003)  (0.003)    (0.004)  (0.004)  Wave 2010–2014 (Ref. 2005–2009)  0.006  −0.271  −0.264  0.002  −0.297  −0.294  (Wave 2010–2014 only)      (0.244)  (0.207)  (0.208)  (0.209)  (0.189)  (0.189)      Constant  −3.992***  −4.736***  −4.861***  −4.181***  −4.283***  −4.342***  −3.884***  −5.669***  −5.731***  (0.448)  (0.787)  (0.789)  (0.360)  (0.645)  (0.646)  (0.500)  (1.178)  (1.179)  Random effect (country-year)  Variance  0.884***  0.566***  0.569***  0.747***  0.533***  0.534***  0.712***  0.472***  0.473***  (0.163)  (0.105)  (0.106)  (0.127)  (0.091)  (0.091)  (0.206)  (0.136)  (0.137)  N (individual)  73,917  73,917  73,917  87,737  87,737  87,737  33,971  33,971  33,971  N (country-year)  60  60  60  71  71  71  25  25  25  Wald χ2  1,255  1,293  1,353  1,722  1,757  1,773  671  687  697  AIC  72,436  72,415  72,352  89,332  89,314  89,299  37,054  37,050  37,040  Minority=  Belong to ethnic minority group  Speak non-dominant language at home  Self-identified as an immigrant    Model (1)  Model (2)  Model (3)  Model (4)  Model (5)  Model (6)  Model (7)  Model (8)  Model (9)  Ethnicity effect  Belong to minority group (0, 1)  −0.120***  −0.119***  0.473***  −0.039  −0.036  0.263***  −0.066  −0.066  0.379**  (0.023)  (0.023)  (0.077)  (0.024)  (0.024)  (0.075)  (0.048)  (0.048)  (0.139)  Democracy effect                    Democracy index (1–10)  0.149*  −0.048  −0.025  0.186***  −0.003  0.006  0.166*  −0.012  −0.006  (0.064)  (0.072)  (0.072)  (0.050)  (0.057)  (0.057)  (0.077)  (0.093)  (0.094)  Interaction effect                    Minority##Democracy      −0.093***      −0.051***      −0.070***      (0.012)      (0.012)      (0.021)  Individual characteristics  Female (0, 1)  −0.066***  −0.067***  −0.068***  −0.041*  −0.041*  −0.042*  −0.023  −0.023  −0.023  (0.019)  (0.019)  (0.019)  (0.017)  (0.017)  (0.017)  (0.026)  (0.026)  (0.026)  Age (15–99)  0.005***  0.005***  0.005***  0.003***  0.003***  0.003***  0.003***  0.003***  0.003***  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  Education (0–8)  0.070***  0.070***  0.070***  0.069***  0.069***  0.068***  0.090***  0.089***  0.090***  (0.005)  (0.005)  (0.005)  (0.004)  (0.004)  (0.004)  (0.007)  (0.007)  (0.007)  Income (1–10)  0.042***  0.042***  0.042***  0.045***  0.045***  0.045***  0.045***  0.045***  0.045***  (0.005)  (0.005)  (0.005)  (0.004)  (0.004)  (0.004)  (0.007)  (0.007)  (0.007)  Life satisfaction (1–10)  0.054***  0.054***  0.054***  0.070***  0.070***  0.070***  0.072***  0.072***  0.072***  (0.005)  (0.005)  (0.005)  (0.005)  (0.005)  (0.005)  (0.007)  (0.007)  (0.007)  Political trust (1–4)  0.241***  0.241***  0.240***  0.240***  0.240***  0.239***  0.213***  0.213***  0.213***  (0.011)  (0.011)  (0.011)  (0.010)  (0.010)  (0.010)  (0.015)  (0.015)  (0.015)  Country-year characteristics  Education (societal mean)    0.266*  0.262    0.181  0.183    0.388*  0.392*    (0.135)  (0.135)    (0.118)  (0.118)    (0.189)  (0.189)  GDP per capita (in thousand US $)    0.020*  0.019*    0.026***  0.026***    0.028*  0.028*    (0.010)  (0.010)    (0.007)  (0.007)    (0.012)  (0.012)  Fearon’s ethnic diversity index (0–1)    0.007*  0.007*    0.001  0.001    0.001  0.002    (0.003)  (0.003)    (0.003)  (0.003)    (0.004)  (0.004)  Wave 2010–2014 (Ref. 2005–2009)  0.006  −0.271  −0.264  0.002  −0.297  −0.294  (Wave 2010–2014 only)      (0.244)  (0.207)  (0.208)  (0.209)  (0.189)  (0.189)      Constant  −3.992***  −4.736***  −4.861***  −4.181***  −4.283***  −4.342***  −3.884***  −5.669***  −5.731***  (0.448)  (0.787)  (0.789)  (0.360)  (0.645)  (0.646)  (0.500)  (1.178)  (1.179)  Random effect (country-year)  Variance  0.884***  0.566***  0.569***  0.747***  0.533***  0.534***  0.712***  0.472***  0.473***  (0.163)  (0.105)  (0.106)  (0.127)  (0.091)  (0.091)  (0.206)  (0.136)  (0.137)  N (individual)  73,917  73,917  73,917  87,737  87,737  87,737  33,971  33,971  33,971  N (country-year)  60  60  60  71  71  71  25  25  25  Wald χ2  1,255  1,293  1,353  1,722  1,757  1,773  671  687  697  AIC  72,436  72,415  72,352  89,332  89,314  89,299  37,054  37,050  37,040  Standard errors in parentheses,* p < 0.05 ** p < 0.01 *** p < 0.001 Table 4. Multilevel models showing effect of ethnicity and democratization on trust Minority=  Belong to ethnic minority group  Speak non-dominant language at home  Self-identified as an immigrant    Model (1)  Model (2)  Model (3)  Model (4)  Model (5)  Model (6)  Model (7)  Model (8)  Model (9)  Ethnicity effect  Belong to minority group (0, 1)  −0.120***  −0.119***  0.473***  −0.039  −0.036  0.263***  −0.066  −0.066  0.379**  (0.023)  (0.023)  (0.077)  (0.024)  (0.024)  (0.075)  (0.048)  (0.048)  (0.139)  Democracy effect                    Democracy index (1–10)  0.149*  −0.048  −0.025  0.186***  −0.003  0.006  0.166*  −0.012  −0.006  (0.064)  (0.072)  (0.072)  (0.050)  (0.057)  (0.057)  (0.077)  (0.093)  (0.094)  Interaction effect                    Minority##Democracy      −0.093***      −0.051***      −0.070***      (0.012)      (0.012)      (0.021)  Individual characteristics  Female (0, 1)  −0.066***  −0.067***  −0.068***  −0.041*  −0.041*  −0.042*  −0.023  −0.023  −0.023  (0.019)  (0.019)  (0.019)  (0.017)  (0.017)  (0.017)  (0.026)  (0.026)  (0.026)  Age (15–99)  0.005***  0.005***  0.005***  0.003***  0.003***  0.003***  0.003***  0.003***  0.003***  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  Education (0–8)  0.070***  0.070***  0.070***  0.069***  0.069***  0.068***  0.090***  0.089***  0.090***  (0.005)  (0.005)  (0.005)  (0.004)  (0.004)  (0.004)  (0.007)  (0.007)  (0.007)  Income (1–10)  0.042***  0.042***  0.042***  0.045***  0.045***  0.045***  0.045***  0.045***  0.045***  (0.005)  (0.005)  (0.005)  (0.004)  (0.004)  (0.004)  (0.007)  (0.007)  (0.007)  Life satisfaction (1–10)  0.054***  0.054***  0.054***  0.070***  0.070***  0.070***  0.072***  0.072***  0.072***  (0.005)  (0.005)  (0.005)  (0.005)  (0.005)  (0.005)  (0.007)  (0.007)  (0.007)  Political trust (1–4)  0.241***  0.241***  0.240***  0.240***  0.240***  0.239***  0.213***  0.213***  0.213***  (0.011)  (0.011)  (0.011)  (0.010)  (0.010)  (0.010)  (0.015)  (0.015)  (0.015)  Country-year characteristics  Education (societal mean)    0.266*  0.262    0.181  0.183    0.388*  0.392*    (0.135)  (0.135)    (0.118)  (0.118)    (0.189)  (0.189)  GDP per capita (in thousand US $)    0.020*  0.019*    0.026***  0.026***    0.028*  0.028*    (0.010)  (0.010)    (0.007)  (0.007)    (0.012)  (0.012)  Fearon’s ethnic diversity index (0–1)    0.007*  0.007*    0.001  0.001    0.001  0.002    (0.003)  (0.003)    (0.003)  (0.003)    (0.004)  (0.004)  Wave 2010–2014 (Ref. 2005–2009)  0.006  −0.271  −0.264  0.002  −0.297  −0.294  (Wave 2010–2014 only)      (0.244)  (0.207)  (0.208)  (0.209)  (0.189)  (0.189)      Constant  −3.992***  −4.736***  −4.861***  −4.181***  −4.283***  −4.342***  −3.884***  −5.669***  −5.731***  (0.448)  (0.787)  (0.789)  (0.360)  (0.645)  (0.646)  (0.500)  (1.178)  (1.179)  Random effect (country-year)  Variance  0.884***  0.566***  0.569***  0.747***  0.533***  0.534***  0.712***  0.472***  0.473***  (0.163)  (0.105)  (0.106)  (0.127)  (0.091)  (0.091)  (0.206)  (0.136)  (0.137)  N (individual)  73,917  73,917  73,917  87,737  87,737  87,737  33,971  33,971  33,971  N (country-year)  60  60  60  71  71  71  25  25  25  Wald χ2  1,255  1,293  1,353  1,722  1,757  1,773  671  687  697  AIC  72,436  72,415  72,352  89,332  89,314  89,299  37,054  37,050  37,040  Minority=  Belong to ethnic minority group  Speak non-dominant language at home  Self-identified as an immigrant    Model (1)  Model (2)  Model (3)  Model (4)  Model (5)  Model (6)  Model (7)  Model (8)  Model (9)  Ethnicity effect  Belong to minority group (0, 1)  −0.120***  −0.119***  0.473***  −0.039  −0.036  0.263***  −0.066  −0.066  0.379**  (0.023)  (0.023)  (0.077)  (0.024)  (0.024)  (0.075)  (0.048)  (0.048)  (0.139)  Democracy effect                    Democracy index (1–10)  0.149*  −0.048  −0.025  0.186***  −0.003  0.006  0.166*  −0.012  −0.006  (0.064)  (0.072)  (0.072)  (0.050)  (0.057)  (0.057)  (0.077)  (0.093)  (0.094)  Interaction effect                    Minority##Democracy      −0.093***      −0.051***      −0.070***      (0.012)      (0.012)      (0.021)  Individual characteristics  Female (0, 1)  −0.066***  −0.067***  −0.068***  −0.041*  −0.041*  −0.042*  −0.023  −0.023  −0.023  (0.019)  (0.019)  (0.019)  (0.017)  (0.017)  (0.017)  (0.026)  (0.026)  (0.026)  Age (15–99)  0.005***  0.005***  0.005***  0.003***  0.003***  0.003***  0.003***  0.003***  0.003***  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  Education (0–8)  0.070***  0.070***  0.070***  0.069***  0.069***  0.068***  0.090***  0.089***  0.090***  (0.005)  (0.005)  (0.005)  (0.004)  (0.004)  (0.004)  (0.007)  (0.007)  (0.007)  Income (1–10)  0.042***  0.042***  0.042***  0.045***  0.045***  0.045***  0.045***  0.045***  0.045***  (0.005)  (0.005)  (0.005)  (0.004)  (0.004)  (0.004)  (0.007)  (0.007)  (0.007)  Life satisfaction (1–10)  0.054***  0.054***  0.054***  0.070***  0.070***  0.070***  0.072***  0.072***  0.072***  (0.005)  (0.005)  (0.005)  (0.005)  (0.005)  (0.005)  (0.007)  (0.007)  (0.007)  Political trust (1–4)  0.241***  0.241***  0.240***  0.240***  0.240***  0.239***  0.213***  0.213***  0.213***  (0.011)  (0.011)  (0.011)  (0.010)  (0.010)  (0.010)  (0.015)  (0.015)  (0.015)  Country-year characteristics  Education (societal mean)    0.266*  0.262    0.181  0.183    0.388*  0.392*    (0.135)  (0.135)    (0.118)  (0.118)    (0.189)  (0.189)  GDP per capita (in thousand US $)    0.020*  0.019*    0.026***  0.026***    0.028*  0.028*    (0.010)  (0.010)    (0.007)  (0.007)    (0.012)  (0.012)  Fearon’s ethnic diversity index (0–1)    0.007*  0.007*    0.001  0.001    0.001  0.002    (0.003)  (0.003)    (0.003)  (0.003)    (0.004)  (0.004)  Wave 2010–2014 (Ref. 2005–2009)  0.006  −0.271  −0.264  0.002  −0.297  −0.294  (Wave 2010–2014 only)      (0.244)  (0.207)  (0.208)  (0.209)  (0.189)  (0.189)      Constant  −3.992***  −4.736***  −4.861***  −4.181***  −4.283***  −4.342***  −3.884***  −5.669***  −5.731***  (0.448)  (0.787)  (0.789)  (0.360)  (0.645)  (0.646)  (0.500)  (1.178)  (1.179)  Random effect (country-year)  Variance  0.884***  0.566***  0.569***  0.747***  0.533***  0.534***  0.712***  0.472***  0.473***  (0.163)  (0.105)  (0.106)  (0.127)  (0.091)  (0.091)  (0.206)  (0.136)  (0.137)  N (individual)  73,917  73,917  73,917  87,737  87,737  87,737  33,971  33,971  33,971  N (country-year)  60  60  60  71  71  71  25  25  25  Wald χ2  1,255  1,293  1,353  1,722  1,757  1,773  671  687  697  AIC  72,436  72,415  72,352  89,332  89,314  89,299  37,054  37,050  37,040  Standard errors in parentheses,* p < 0.05 ** p < 0.01 *** p < 0.001 Model 2 then adds country-specific macro-level controls for education, GDP, and ethnic diversity. As would be expected with this specification, the effect of ethnic minority continues to be negative and significant (−.119) but in this instance, the coefficient for democratization is no longer significant. Separate analyses (not shown) determined that this was the result of the inclusion of the GDP measure (see Acemoglu et al. [2014] on the relationship). Model 3 adds an interaction term between ethnic minority status and democratization, thereby allowing us to test whether the gap in trust between the majority and minority group varies based on levels of democratization. That is, this model tests whether the effect of ethnic minority is context specific and depends on the level of democratization. The results show that it does. The interaction term is negative, indicating that as the level of democratization increases, minority group members are less likely to have higher trust than majority group members. With the inclusion of the interaction term, the ethnic minority effect on trust becomes positive and significant. This suggests that it is only in democracies that ethnic minorities trust less than ethnic majority group members. In contrast, in non-democratic societies, members of ethnic minorities trust more than members of ethnic majorities. Models 4–6 replicate these findings with ethnicity referring to ethnic minority language, and models 7–9 replicate these findings with ethnicity referring to immigrant status. The results are very similar. The only difference is that the initial coefficient for ethnic minority is not significant, indicating that across countries there is no aggregate ethnic trust gap. Instead, the ethnic trust gap is only for the most and least on the democratic scale. Figures 7–9 provide a graphical representation of the relationship between ethnic majority and minority status for each of the 10 levels of democratization. These figures clearly show that, irrespective of the components of ethnicity (ethnic minority, minority language, immigration), there is a positive overall effect of democratization for both majority and minority groups, but that the relative gap between groups varies according to the political situation. In authoritarian countries, minority groups tend to trust less than majority groups, and as societies get more democratic, majority group members trust more than minority group members. Figures 7–9. View largeDownload slide Plots of the interaction between ethnicity and democratization Note: Yes denotes minority group, No denotes majority group, democratization from low (1) to high (10). Figures 7–9. View largeDownload slide Plots of the interaction between ethnicity and democratization Note: Yes denotes minority group, No denotes majority group, democratization from low (1) to high (10). Next, in table 5 we consider the effect of ethnic minority status within the four EIU regime types—authoritarian, hybrid democracies, flawed democracies, and full democracies. Recall that these regimes are distinguished based on the extent to which they facilitate or hinder, among others, basic political and civic freedoms, elections, open media, and political participation. For each regime type, we estimate three models—one for each type of ethnic minority status. Model 1 for authoritarian regimes shows that ethnic minority groups in authoritarian groups tend to trust more than majority group members. These results hold when ethnicity is coded as ethnic minority group (.200) (model 2) and based on minority language (.301) (model 3). The pattern for flawed democracies is more mixed and variable, depending on the coding of ethnic minority status. When it refers to ethnic minority, the effect is negative (−.272), but for minority language it is non-significant and for immigrant it is positive (.343). For flawed democracies, there is only a negative effect for immigrant minorities (−.375). Finally, for the full democracies we see that the effect is negative across all three codings of ethnic minority status (ethnic minority, −.452; linguistic minority, −.549; and immigrant, −.298).15 Table 5. Logistic regression models showing effect of ethnicity on trust within regime types Regime type:  Authoritarian  Hybrid democracy  Flawed democracy  Full democracy    Model (1)  Model (2)  Model (3)  Model (4)  Model (5)  Model (6)  Model (7)  Model (8)  Model (9)  Model (10)  Model (11)  Model (12)  Ethnicity effect                          Belong to minority group (0, 1)  0.200***      −0.272***      −0.000      −0.452***      (0.055)      (0.045)      (0.035)      (0.046)      Speak non-dominant language at home (0, 1)    0.301***      −0.081      0.036      −0.549***      (0.049)      (0.042)      (0.035)      (0.070)    Self-identified as an immigrant (0, 1)      −0.022      0.343***      −0.375**      −0.298***      (0.090)      (0.100)      (0.125)      (0.081)  Individual characteristics  Female (0, 1)  −0.041  0.027  0.057  −0.076*  −0.096**  −0.144**  −0.202***  −0.078*  −0.031  −0.011  −0.015  −0.021  (0.047)  (0.036)  (0.048)  (0.037)  (0.034)  (0.055)  (0.032)  (0.030)  (0.053)  (0.034)  (0.031)  (0.049)  Age (15–99)  0.007***  0.009***  0.002  0.002*  0.005***  0.002  0.005***  0.006***  −0.006***  0.011***  0.008***  0.011***  (0.002)  (0.001)  (0.002)  (0.001)  (0.001)  (0.002)  (0.001)  (0.001)  (0.002)  (0.001)  (0.001)  (0.001)  Education (0–8)  0.028*  0.035***  0.020  −0.019*  −0.015  0.038**  0.084***  0.070***  0.103***  0.185***  0.180***  0.193***  (0.012)  (0.010)  (0.013)  (0.009)  (0.008)  (0.014)  (0.008)  (0.008)  (0.016)  (0.010)  (0.008)  (0.014)  Income (1–10)  0.052***  0.039***  0.025  0.030**  0.072***  0.049**  −0.015*  0.003  −0.020  0.082***  0.075***  0.074***  (0.013)  (0.010)  (0.014)  (0.010)  (0.008)  (0.016)  (0.008)  (0.007)  (0.015)  (0.008)  (0.007)  (0.013)  Life satisfaction (1–10)  0.043***  0.072***  0.058***  0.066***  0.053***  0.104***  −0.029***  0.004  0.011  0.105***  0.129***  0.161***  (0.012)  (0.009)  (0.013)  (0.009)  (0.008)  (0.014)  (0.008)  (0.007)  (0.015)  (0.011)  (0.010)  (0.016)  Political trust (1–4)  0.295***  0.218***  0.090***  0.189***  0.166***  0.151***  0.214***  0.198***  0.129***  0.376***  0.421***  0.430***  (0.027)  (0.020)  (0.026)  (0.020)  (0.018)  (0.031)  (0.017)  (0.017)  (0.029)  (0.022)  (0.021)  (0.034)  Country-year characteristics  Education (societal mean)  −0.183***  0.088**  0.356***  0.320***  0.192***  1.929***  0.449***  0.313***  1.162***  0.036  0.045*  1.300***  (0.049)  (0.032)  (0.041)  (0.031)  (0.027)  (0.156)  (0.028)  (0.030)  (0.072)  (0.025)  (0.022)  (0.086)  GDP per capita (in thousand US $)  0.199***  0.037***  0.027***  0.013***  0.014***  0.030***  −0.001  0.025***  −0.065***  0.027***  0.030***  −0.014***  (0.010)  (0.002)  (0.005)  (0.001)  (0.001)  (0.002)  (0.003)  (0.002)  (0.006)  (0.002)  (0.001)  (0.004)  Fearon’s ethnic diversity index (0–1)  0.020***  0.013***  −0.006***  −0.002***  −0.007***  −0.002*  −0.001  −0.005***  0.009***  0.011***  0.010***  0.063***  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.000)  (0.001)  (0.001)  (0.001)  (0.003)  Wave 2010–2014 (Ref. 2005–2009)  −0.373***  −0.654***  NA  −0.754***  −0.292***  NA  −0.019  −0.227***  NA  −0.245***  −0.127***  NA  (0.069)  (0.051)  (0.052)  (0.045)  (0.043)  (0.037)  (0.041)  (0.035)  Constant  −4.408***  −4.465***  −3.623***  −2.829***  −2.402***  −13.844***  −3.948***  −3.588***  −6.511***  −6.596***  −6.970***  −18.588***  (0.262)  (0.174)  (0.279)  (0.148)  (0.140)  (0.938)  (0.163)  (0.164)  (0.377)  (0.190)  (0.170)  (0.745)  N  11,508  18,101  9,229  15,791  20,808  8,094  30,257  27,885  8,492  16,361  20,943  8,156  Regime type:  Authoritarian  Hybrid democracy  Flawed democracy  Full democracy    Model (1)  Model (2)  Model (3)  Model (4)  Model (5)  Model (6)  Model (7)  Model (8)  Model (9)  Model (10)  Model (11)  Model (12)  Ethnicity effect                          Belong to minority group (0, 1)  0.200***      −0.272***      −0.000      −0.452***      (0.055)      (0.045)      (0.035)      (0.046)      Speak non-dominant language at home (0, 1)    0.301***      −0.081      0.036      −0.549***      (0.049)      (0.042)      (0.035)      (0.070)    Self-identified as an immigrant (0, 1)      −0.022      0.343***      −0.375**      −0.298***      (0.090)      (0.100)      (0.125)      (0.081)  Individual characteristics  Female (0, 1)  −0.041  0.027  0.057  −0.076*  −0.096**  −0.144**  −0.202***  −0.078*  −0.031  −0.011  −0.015  −0.021  (0.047)  (0.036)  (0.048)  (0.037)  (0.034)  (0.055)  (0.032)  (0.030)  (0.053)  (0.034)  (0.031)  (0.049)  Age (15–99)  0.007***  0.009***  0.002  0.002*  0.005***  0.002  0.005***  0.006***  −0.006***  0.011***  0.008***  0.011***  (0.002)  (0.001)  (0.002)  (0.001)  (0.001)  (0.002)  (0.001)  (0.001)  (0.002)  (0.001)  (0.001)  (0.001)  Education (0–8)  0.028*  0.035***  0.020  −0.019*  −0.015  0.038**  0.084***  0.070***  0.103***  0.185***  0.180***  0.193***  (0.012)  (0.010)  (0.013)  (0.009)  (0.008)  (0.014)  (0.008)  (0.008)  (0.016)  (0.010)  (0.008)  (0.014)  Income (1–10)  0.052***  0.039***  0.025  0.030**  0.072***  0.049**  −0.015*  0.003  −0.020  0.082***  0.075***  0.074***  (0.013)  (0.010)  (0.014)  (0.010)  (0.008)  (0.016)  (0.008)  (0.007)  (0.015)  (0.008)  (0.007)  (0.013)  Life satisfaction (1–10)  0.043***  0.072***  0.058***  0.066***  0.053***  0.104***  −0.029***  0.004  0.011  0.105***  0.129***  0.161***  (0.012)  (0.009)  (0.013)  (0.009)  (0.008)  (0.014)  (0.008)  (0.007)  (0.015)  (0.011)  (0.010)  (0.016)  Political trust (1–4)  0.295***  0.218***  0.090***  0.189***  0.166***  0.151***  0.214***  0.198***  0.129***  0.376***  0.421***  0.430***  (0.027)  (0.020)  (0.026)  (0.020)  (0.018)  (0.031)  (0.017)  (0.017)  (0.029)  (0.022)  (0.021)  (0.034)  Country-year characteristics  Education (societal mean)  −0.183***  0.088**  0.356***  0.320***  0.192***  1.929***  0.449***  0.313***  1.162***  0.036  0.045*  1.300***  (0.049)  (0.032)  (0.041)  (0.031)  (0.027)  (0.156)  (0.028)  (0.030)  (0.072)  (0.025)  (0.022)  (0.086)  GDP per capita (in thousand US $)  0.199***  0.037***  0.027***  0.013***  0.014***  0.030***  −0.001  0.025***  −0.065***  0.027***  0.030***  −0.014***  (0.010)  (0.002)  (0.005)  (0.001)  (0.001)  (0.002)  (0.003)  (0.002)  (0.006)  (0.002)  (0.001)  (0.004)  Fearon’s ethnic diversity index (0–1)  0.020***  0.013***  −0.006***  −0.002***  −0.007***  −0.002*  −0.001  −0.005***  0.009***  0.011***  0.010***  0.063***  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.000)  (0.001)  (0.001)  (0.001)  (0.003)  Wave 2010–2014 (Ref. 2005–2009)  −0.373***  −0.654***  NA  −0.754***  −0.292***  NA  −0.019  −0.227***  NA  −0.245***  −0.127***  NA  (0.069)  (0.051)  (0.052)  (0.045)  (0.043)  (0.037)  (0.041)  (0.035)  Constant  −4.408***  −4.465***  −3.623***  −2.829***  −2.402***  −13.844***  −3.948***  −3.588***  −6.511***  −6.596***  −6.970***  −18.588***  (0.262)  (0.174)  (0.279)  (0.148)  (0.140)  (0.938)  (0.163)  (0.164)  (0.377)  (0.190)  (0.170)  (0.745)  N  11,508  18,101  9,229  15,791  20,808  8,094  30,257  27,885  8,492  16,361  20,943  8,156  Standard errors in parentheses,* p < 0.05 ** p < 0.01 *** p < 0.001 Table 5. Logistic regression models showing effect of ethnicity on trust within regime types Regime type:  Authoritarian  Hybrid democracy  Flawed democracy  Full democracy    Model (1)  Model (2)  Model (3)  Model (4)  Model (5)  Model (6)  Model (7)  Model (8)  Model (9)  Model (10)  Model (11)  Model (12)  Ethnicity effect                          Belong to minority group (0, 1)  0.200***      −0.272***      −0.000      −0.452***      (0.055)      (0.045)      (0.035)      (0.046)      Speak non-dominant language at home (0, 1)    0.301***      −0.081      0.036      −0.549***      (0.049)      (0.042)      (0.035)      (0.070)    Self-identified as an immigrant (0, 1)      −0.022      0.343***      −0.375**      −0.298***      (0.090)      (0.100)      (0.125)      (0.081)  Individual characteristics  Female (0, 1)  −0.041  0.027  0.057  −0.076*  −0.096**  −0.144**  −0.202***  −0.078*  −0.031  −0.011  −0.015  −0.021  (0.047)  (0.036)  (0.048)  (0.037)  (0.034)  (0.055)  (0.032)  (0.030)  (0.053)  (0.034)  (0.031)  (0.049)  Age (15–99)  0.007***  0.009***  0.002  0.002*  0.005***  0.002  0.005***  0.006***  −0.006***  0.011***  0.008***  0.011***  (0.002)  (0.001)  (0.002)  (0.001)  (0.001)  (0.002)  (0.001)  (0.001)  (0.002)  (0.001)  (0.001)  (0.001)  Education (0–8)  0.028*  0.035***  0.020  −0.019*  −0.015  0.038**  0.084***  0.070***  0.103***  0.185***  0.180***  0.193***  (0.012)  (0.010)  (0.013)  (0.009)  (0.008)  (0.014)  (0.008)  (0.008)  (0.016)  (0.010)  (0.008)  (0.014)  Income (1–10)  0.052***  0.039***  0.025  0.030**  0.072***  0.049**  −0.015*  0.003  −0.020  0.082***  0.075***  0.074***  (0.013)  (0.010)  (0.014)  (0.010)  (0.008)  (0.016)  (0.008)  (0.007)  (0.015)  (0.008)  (0.007)  (0.013)  Life satisfaction (1–10)  0.043***  0.072***  0.058***  0.066***  0.053***  0.104***  −0.029***  0.004  0.011  0.105***  0.129***  0.161***  (0.012)  (0.009)  (0.013)  (0.009)  (0.008)  (0.014)  (0.008)  (0.007)  (0.015)  (0.011)  (0.010)  (0.016)  Political trust (1–4)  0.295***  0.218***  0.090***  0.189***  0.166***  0.151***  0.214***  0.198***  0.129***  0.376***  0.421***  0.430***  (0.027)  (0.020)  (0.026)  (0.020)  (0.018)  (0.031)  (0.017)  (0.017)  (0.029)  (0.022)  (0.021)  (0.034)  Country-year characteristics  Education (societal mean)  −0.183***  0.088**  0.356***  0.320***  0.192***  1.929***  0.449***  0.313***  1.162***  0.036  0.045*  1.300***  (0.049)  (0.032)  (0.041)  (0.031)  (0.027)  (0.156)  (0.028)  (0.030)  (0.072)  (0.025)  (0.022)  (0.086)  GDP per capita (in thousand US $)  0.199***  0.037***  0.027***  0.013***  0.014***  0.030***  −0.001  0.025***  −0.065***  0.027***  0.030***  −0.014***  (0.010)  (0.002)  (0.005)  (0.001)  (0.001)  (0.002)  (0.003)  (0.002)  (0.006)  (0.002)  (0.001)  (0.004)  Fearon’s ethnic diversity index (0–1)  0.020***  0.013***  −0.006***  −0.002***  −0.007***  −0.002*  −0.001  −0.005***  0.009***  0.011***  0.010***  0.063***  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.000)  (0.001)  (0.001)  (0.001)  (0.003)  Wave 2010–2014 (Ref. 2005–2009)  −0.373***  −0.654***  NA  −0.754***  −0.292***  NA  −0.019  −0.227***  NA  −0.245***  −0.127***  NA  (0.069)  (0.051)  (0.052)  (0.045)  (0.043)  (0.037)  (0.041)  (0.035)  Constant  −4.408***  −4.465***  −3.623***  −2.829***  −2.402***  −13.844***  −3.948***  −3.588***  −6.511***  −6.596***  −6.970***  −18.588***  (0.262)  (0.174)  (0.279)  (0.148)  (0.140)  (0.938)  (0.163)  (0.164)  (0.377)  (0.190)  (0.170)  (0.745)  N  11,508  18,101  9,229  15,791  20,808  8,094  30,257  27,885  8,492  16,361  20,943  8,156  Regime type:  Authoritarian  Hybrid democracy  Flawed democracy  Full democracy    Model (1)  Model (2)  Model (3)  Model (4)  Model (5)  Model (6)  Model (7)  Model (8)  Model (9)  Model (10)  Model (11)  Model (12)  Ethnicity effect                          Belong to minority group (0, 1)  0.200***      −0.272***      −0.000      −0.452***      (0.055)      (0.045)      (0.035)      (0.046)      Speak non-dominant language at home (0, 1)    0.301***      −0.081      0.036      −0.549***      (0.049)      (0.042)      (0.035)      (0.070)    Self-identified as an immigrant (0, 1)      −0.022      0.343***      −0.375**      −0.298***      (0.090)      (0.100)      (0.125)      (0.081)  Individual characteristics  Female (0, 1)  −0.041  0.027  0.057  −0.076*  −0.096**  −0.144**  −0.202***  −0.078*  −0.031  −0.011  −0.015  −0.021  (0.047)  (0.036)  (0.048)  (0.037)  (0.034)  (0.055)  (0.032)  (0.030)  (0.053)  (0.034)  (0.031)  (0.049)  Age (15–99)  0.007***  0.009***  0.002  0.002*  0.005***  0.002  0.005***  0.006***  −0.006***  0.011***  0.008***  0.011***  (0.002)  (0.001)  (0.002)  (0.001)  (0.001)  (0.002)  (0.001)  (0.001)  (0.002)  (0.001)  (0.001)  (0.001)  Education (0–8)  0.028*  0.035***  0.020  −0.019*  −0.015  0.038**  0.084***  0.070***  0.103***  0.185***  0.180***  0.193***  (0.012)  (0.010)  (0.013)  (0.009)  (0.008)  (0.014)  (0.008)  (0.008)  (0.016)  (0.010)  (0.008)  (0.014)  Income (1–10)  0.052***  0.039***  0.025  0.030**  0.072***  0.049**  −0.015*  0.003  −0.020  0.082***  0.075***  0.074***  (0.013)  (0.010)  (0.014)  (0.010)  (0.008)  (0.016)  (0.008)  (0.007)  (0.015)  (0.008)  (0.007)  (0.013)  Life satisfaction (1–10)  0.043***  0.072***  0.058***  0.066***  0.053***  0.104***  −0.029***  0.004  0.011  0.105***  0.129***  0.161***  (0.012)  (0.009)  (0.013)  (0.009)  (0.008)  (0.014)  (0.008)  (0.007)  (0.015)  (0.011)  (0.010)  (0.016)  Political trust (1–4)  0.295***  0.218***  0.090***  0.189***  0.166***  0.151***  0.214***  0.198***  0.129***  0.376***  0.421***  0.430***  (0.027)  (0.020)  (0.026)  (0.020)  (0.018)  (0.031)  (0.017)  (0.017)  (0.029)  (0.022)  (0.021)  (0.034)  Country-year characteristics  Education (societal mean)  −0.183***  0.088**  0.356***  0.320***  0.192***  1.929***  0.449***  0.313***  1.162***  0.036  0.045*  1.300***  (0.049)  (0.032)  (0.041)  (0.031)  (0.027)  (0.156)  (0.028)  (0.030)  (0.072)  (0.025)  (0.022)  (0.086)  GDP per capita (in thousand US $)  0.199***  0.037***  0.027***  0.013***  0.014***  0.030***  −0.001  0.025***  −0.065***  0.027***  0.030***  −0.014***  (0.010)  (0.002)  (0.005)  (0.001)  (0.001)  (0.002)  (0.003)  (0.002)  (0.006)  (0.002)  (0.001)  (0.004)  Fearon’s ethnic diversity index (0–1)  0.020***  0.013***  −0.006***  −0.002***  −0.007***  −0.002*  −0.001  −0.005***  0.009***  0.011***  0.010***  0.063***  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.000)  (0.001)  (0.001)  (0.001)  (0.003)  Wave 2010–2014 (Ref. 2005–2009)  −0.373***  −0.654***  NA  −0.754***  −0.292***  NA  −0.019  −0.227***  NA  −0.245***  −0.127***  NA  (0.069)  (0.051)  (0.052)  (0.045)  (0.043)  (0.037)  (0.041)  (0.035)  Constant  −4.408***  −4.465***  −3.623***  −2.829***  −2.402***  −13.844***  −3.948***  −3.588***  −6.511***  −6.596***  −6.970***  −18.588***  (0.262)  (0.174)  (0.279)  (0.148)  (0.140)  (0.938)  (0.163)  (0.164)  (0.377)  (0.190)  (0.170)  (0.745)  N  11,508  18,101  9,229  15,791  20,808  8,094  30,257  27,885  8,492  16,361  20,943  8,156  Standard errors in parentheses,* p < 0.05 ** p < 0.01 *** p < 0.001 Figure 10 provides a visual illustration of the odds ratios for how each of the three ethnic minority measures affects trust in four types of regimes. Odds ratios are obtained from each of the models in table 5. Here it is worth focusing on the specific type of regime. The evidence for full democracies is robust—the effect of ethnic majority minority on trust is negative and therefore in the anticipated direction, and this effect occurs for, respectively, all types of ethnicity. The effect of majority-minority for the authoritarian countries is also robust—it is positive and occurs for two out of three types of ethnic majority-minority status. The effect in the flawed democracies is in line with expectations, but because it is only for immigrant minority it is slightly less robust. Finally, the effect for the hybrid democracies is somewhat of a puzzle. Whereas immigrant minorities trust more, those who belong to an ethnic minority group trust less. One possibility is that in comparison to linguistic and ethnic minority groups, the meaning of immigrant status is more variable across different types of societies. That is, it could be that there are some kind of immigrant self-selection processes occurring to different kinds of societies—with one kind of individual selecting to more democratic states and another to less democratic states. Given that these effects of immigrant minority status on trust remain—even after controls for some of the characteristics that we might expect to be associated with selection such as education and income are included—it would seem that some other process must be in play. Nevertheless, the fact that the gap in trust across regime types remains even when ethnicity refers to ethnic and linguistic minority rather than immigrant minority clearly shows that selection is not the primary cause. Figure 10. View largeDownload slide Odds ratios showing how three types of ethnicity affect trust in four regime types. Note: Above 1 denotes minority more trusting, below 1 majority more trusting. Figure 10. View largeDownload slide Odds ratios showing how three types of ethnicity affect trust in four regime types. Note: Above 1 denotes minority more trusting, below 1 majority more trusting. Finally, in order to test the mechanism underlying this gap, we considered whether self-placement on the class scale (X045) and perception of racist behavior in the neighborhood (H002_04) operate as mediators. This entailed estimating the same multilevel models with these potential mechanistic outcomes as dependent variables (see, e.g., Hanushek and Kimko 2000). The idea is that, due to their feelings of relative powerlessness compared to the majority group, minority group members (of all kinds) might be more sensitive to racist behavior in the neighborhood. The same logic holds for self-placement on the class scale. Of course, it should be noted that racist behavior in the neighborhood could be occurring even if it is targeting a different group, and that this is not a direct measure of individual-level discrimination. Therefore, because these are indirect and external correlates of the experience of relative powerlessness at best because there might be many other mechanisms at play, and because the WVS data does not contain any other measure that we could use to capture this concept more directly, these results should be understood as exploratory rather than confirmatory. The results are presented in Appendix C. For each outcome, we estimate three models: one for each type of ethnicity—ethnic minority, linguistic minority, and immigrant. The interaction between ethnic minority and democratization again captures the gap between the majority and minority group. The coefficient is positive, indicating that results showing that compared to less democratic societies, minority groups in democracies are more likely than the majority to see racist behavior in the neighborhood and to place themselves on the low end of the class scale than majority group members. That is, as would be expected under a situation of relative powerlessness, the gap between the majority and minority group increases in more democratic societies. Conclusion Research shows that when asked whether “most people can be trusted,” ethnic minorities tend to trust less (De Vroome, Hooghe, and Marien 2013; Smith 2010; Stolle and Harell 2013; Wilkes and Wu 2017). Not only could lower trust be an issue for individual members of ethnic minority groups, but the gap in trust could also matter more broadly for the growing number of countries seeing a rise in ethnic and racial diversity (e.g., see Bove and Elia 2017). The problem is that the vast majority of the evidence about low trust among ethnic minorities comes from studies conducted within democratic societies, and it has been well established that democratization increases trust (Delhey, Newton, and Welzel 2011; Inglehart 1999). How, then, can theoretical arguments about the benefits of democracy for trust be reconciled with the empirical evidence showing that there are many ethnic gaps in trust in democratic societies? Studies to date have been unable to answer this question because they have largely considered the relationship between ethnicity and trust in democratic countries. The research presented in this paper addresses this gap by considering the relationship between ethnicity and trust across political regime types. In so doing, we made two overall contributions to the study of the relationship between ethnicity and trust. The first contribution has been the development of a majority-minority argument that states that any effect of ethnicity is reflective of a relative power positioning within society as denoted by majority-minority status. That is, what matters is not someone’s specific language, religion, or other cultural marker but rather the relative positioning that these different groupings denote. This relative positioning becomes evident by considering the relationship between ethnic majority-minority status and trust under different macro-political contexts tied to democratization. Of relevance for ethnic gaps in trust is that by its very nature, democracy provides the potential for the tyranny of the majority (Brittain 1996; Gamble 1997). Thus, even if the structure governing overall inequality is fairer in democracies, democratic institutions are at times used to prevent the integration and success of minority groups (Bjørnskov 2008, 274–75). Democratic freedoms also increase the saliency and awareness of inequality (ibid.), thereby creating the conditions under which ethnic minority groups come to understand and link their status to differential treatment (Kymlicka 2002; Ziller 2017). In contrast, ethnic minority groups often dominate under authoritarian regimes. The majority-minority argument thus puts ethnic gaps in trust into global perspective. It allows for the separation of ethnic effects from political context effects and draws attention to the fact that ethnic gaps in trust might look very different under different types of political contexts. Ethnic differences in trust are not only reflective of migration of less trusting individuals from less trusting societies. And, while the argument clearly suggests that the gap is related to some kind of power imbalance possibly via discrimination, it does not assume that ethnic majorities are de facto more trusting. As the case of authoritarian societies illustrates, in some cases ethnic minorities are the beneficiaries of a power imbalance, and this shows in their higher relative trust levels. The second contribution has been the development of new micro-level majority-minority measures of ethnicity for use with global cross-national datasets such as the World Values Survey. Currently the only comparable cross-national individual measure of ethnicity in such datasets is immigration status (e.g., see WVS and Gallup World Polls; Helliwell, Wang, and Xu [2014]). Yet ethnicity, particularly in non-democratic countries, typically refers to many other groupings, including those based on race, cultural identification, and language. The challenge is that as it currently stands, the individual micro-level ethnicity measures in the WVS refer to different ethnic groups and different languages within countries. The majority-minority measures provided in this paper surmount this challenge. Rather than focusing on a particular categorical content such as White race or English language, which may be common to some but not all countries, we focus on the fact that individuals within groups occupy a relative position. What the development of this micro-level measure now means is that by including individual controls for factors such as education and income, it is now possible not only to control for group differences in cultural and economic resources such as income (e.g., see Baldwin and Huber 2010) but also to conduct global comparison on the basis of ethnicity. We then used these new measures to test the majority-minority argument with the World Values Survey. The results demonstrate that while democracy increases generalized trust across the board, the effect of ethnic majority-minority status depends on the level of democratization. In less democratic societies, there is either no trust gap, or at the authoritarian extreme, a situation where ethnic minority groups have higher trust than ethnic majority groups. In contrast, as the level of democratization increases, ethnic majority groups gain more trust relative to ethnic minority groups. The net result is that while democratization raises everyone’s trust, it also leads to a gap in trust that favors the majority group. Because this result holds irrespective of whether ethnic majority-minority status is defined on the basis of ethnic self-identification, language, or immigration, we therefore conclude that these gaps in trust are reflective of relative power positioning within societies. That being said, there are a number of avenues for further research. First, while this study considered far more countries than previous work, an important limitation is that there are still many countries without ethnicity data. As more countries in cross-national datasets such as the WVS collect information about ethnicity, the associations considered here should be revisited. Second is that the systematic and comprehensive testing of mechanisms will be a challenge for further research. Clearly, it is going to be difficult to identify singular mechanisms that can fully account for ethnic gaps in trust, particularly across societies. Certainly while the focus in this paper was on why ethnic minority groups trust more or less relative to ethnic majority groups, attention ought also to be paid to mechanisms that can explain the reverse. That is, the fact that ethnic majority groups might trust more or less in particular contexts (or at the very least acknowledging this possibility) should be considered. Finally, while the focus of this study was on generalized social trust, the mechanisms that explain the relationship between ethnicity and political and institutional trust should also be considered as, in this latter case, trust among minorities is often higher (Maxwell 2010). Footnotes 1 Our immigration measure is not new (see, e.g., Dinesen and Sønderskov 2015; Helliwell, Wang, and Xu 2014). 2 Dinesen and Sønderskov (2015) and Helliwell, Wang, and Xu (2014) have a measure of whether the respondent is an immigrant. Still, it remains the case that all ethnic minorities are not immigrants. 3 Ziller (2017) is an exception. He considers the relationship between government effectiveness and ethnic trust gaps within the European context (20 countries). His measure of government effectiveness is highly correlated with democratization insofar as the 11 countries in his study with the highest government effectiveness scores are all full democracies (e.g., Sweden, Switzerland, etc.) and the nine countries at the bottom(e.g., Hungary, Slovenia, etc.) are all flawed democracies (if one uses the EIU democratization categories—see the Methods section of this paper). His study shows that increases in government effectiveness are associated with increases in ethnic trust gaps in Europe. 4 There is a high rate of missing data on ethnicity (e.g., minority, immigrant) because the question used to define ethnicity was not asked in some countries or has responses indicating 100 percent in the majority group (see Appendix A). We removed these countries because ascribed status is difficult to predict using other variables. Therefore, there is no data for these countries rather than a variable that has missing data for some respondents within these countries. 5 Some countries use race and others use national origin; in some countries, the categories changed over time; others are missing or have no variation (e.g., 100 percent one group) or have too few cases (fewer than 30) for reliable estimates—there is no question that this measure is not without difficulties. Nevertheless, it is currently the only global measure of its kind available. The inclusion of multiple measures of majority-minority group status not subject to these limitations helps ensure that the overall conclusions of the study are robust. 6 The majority is not a statistical majority in some countries. 7 In most countries, because there are multiple minority groups, this means that numerically dominant minority groups will end up driving the results for the minority category. For example, the values in table 3 show that for the United States, Black African and Spanish/Hispanic are the predominant ethnic minority groups. Both would be coded as minority, and since they are about the same numerical size, the results would reflect both. This is an accurate reflection of reality, since previous research shows that both of these groups have lower trust than Whites. There is no particular minority group driving the results for Sweden, and the results for Taiwan would mainly reflect the Hakka group. Still, the process of collapsing across categories does mean that divergent patterns among some smaller groups could potentially be “washed out.” This is an issue with all categorical measures, particularly those of a binary nature. 8 There are two reasons why the linguistic minority and immigrant measure are not highly correlated overall. First, the immigrant status question was not asked in as many countries as the language question. Second, while in some countries immigrant and linguistic minority are highly correlated, this is not the case in others. That is, typically in many (but not all) immigrant-receiving democratic countries, immigrants are also minority language speakers. The numbers in Appendix A, for example, show that in the United States (2011) there are 247 immigrants and 214 minority language speakers and the tetrachoric correlation is .76. In contrast, in many of the non-democratic countries there are still a host of individuals who identify as linguistic minorities (e.g., Uzbekistan 2011) but very few immigrants. The terachoric correlation within Uzbekistan is .43. 9 We do not have a measure of country of origin of immigrants, and so the argument that immigrants coming from less trusting non-OECD countries (e.g., see Dinesen 2012, 2013) might explain some of the gap cannot be entirely discounted. 10 According to the EIU (2007–2014), the features of each regime type are as follows: “Full democracies: Countries in which basic political freedoms and civil liberties are respected. Media are independent and diverse. There is an effective system of checks and balances. The judiciary is independent and judicial decisions are enforced. Flawed democracies: These countries also have free and fair elections and basic civil liberties are respected. However, there are significant weaknesses in other aspects of democracy, including problems in governance, an underdeveloped political culture and low levels of political participation. Hybrid regimes: Elections have substantial irregularities that often prevent them from being both free and fair. Government pressure on opposition parties and candidates may be common. Corruption tends to be widespread and the rule of law is weak. Typically, there is harassment of and pressure on journalists, and the judiciary is not independent. Authoritarian regimes: Many countries in this category are outright dictatorships. Elections, if they do occur, are not free and fair. Media are typically state-owned or controlled by groups connected to the ruling regime. There is no independent judiciary” (abridged from 45–46). 11 Details about the EIU and Polity IV indices can be obtained at http://www.eiu.com/ and http://www.systemicpeace.org/polity/polity4.htm. The data on GDP was obtained from the World Bank data matched to the country-year and can be obtained at http://data.worldbank.org/indicator/NY.GDP.MKTP.CD. The data on country-specific ethnic diversity is from Fearon’s (2003) ethnic fractionalization index. 12 Because each country in the WVS has an N that is generalizable to that country and because the influence of country Ns is small, we do not weight the WVS data. Further, we do not also include year and country as random (see Schmidt-Catran and Fairbrother 2015) because only some countries have more than one year. Thus, our models are cross-sectional rather than longitudinal. As Fairbrother and Martin (2013) point out in their analysis of economic inequality and trust in US states, there could be relationships that exist across (cross-sectional) but not within (longitudinal) geographic units. The net result is that our models provide an estimation of the difference between countries at varying levels of democratization but they do not test whether the process of becoming more or less democratic will lead to rising or falling trust. 13 Some countries only have one wave of data included in our analysis. 14 Democracy is a country-year-level variable. The purpose of the interaction between ethnicity and democracy is to show that ethnicity has different effects (slopes) on trust across country-years with different levels of democratization. Therefore, we do not include a random slope for ethnicity in the models. This said, the results do not change when we run additional estimations of models with a random slope (not shown). 15 As there are a number of countries in the WVS dataset whereby the percentage ethnic minority shifted too much over time to be the result of “real” change (e.g., Cyprus, Malaysia, Morocco, Russia, Thailand, and Ukraine), we reestimated all models without these countries in order to ensure the robustness of the results. The results hold. About the Authors Rima Wilkes is Professor of Sociology at the University of British Columbia and an affiliate of the Laboratory for Comparative Research at the Higher School of Economics in Russia. Her research interests include race, ethnicity and Indigeneity, and political sociology. Recent publications have appeared in International Migration Review, Canadian Review of Sociology, Social Science and Medicine, and PNAS. Cary Wu is a PhD candidate in Sociology at the University of British Columbia, Canada. His research focuses on political sociology, urban sociology, and migration studies. His work has appeared in International Journal of Comparative Sociology, International Political Science Review, Geoforum, American Sociologist, and China Review. Supplementary Material Supplementary material is available at Social Forces online. References Acemoglu, D., S. Naidu, P. Restrepo, and J. A. Robinson. 2014. “Democracy Does Cause Growth.” Working Paper No. w20004, National Bureau of Economic Research. Alesina, A., and E. La Ferrara. 2002. “ Who Trusts Others?” Journal of Public Economics  85: 207– 34. Google Scholar CrossRef Search ADS   Bachmann, R. 2001. “ Trust, Power and Control in Trans-Organizational Relations.” Organization Studies  22( 2): 337– 65. Google Scholar CrossRef Search ADS   Bahry, D., M. Kosolapov, P. Kozyreva, and R. K. Wilson. 2005. “ Ethnicity and Trust: Evidence from Russia.” American Political Science Review  99( 4): 521– 32. Google Scholar CrossRef Search ADS   Baldwin, K., and J. D. Huber. 2010. “ Economic versus Cultural Differences: Forms of Ethnic Diversity and Public Goods Provision.” American Political Science Review  104( 4): 644– 62. Google Scholar CrossRef Search ADS   Bezemer, D., and R. Jong-A-Pin. 2013. “ Democracy, Globalization and Ethnic Violence.” Journal of Comparative Economics  41( 1): 108– 25. Google Scholar CrossRef Search ADS   Bienen, H. 2015. Kenya: The Politics of Participation and Control . Princeton, NJ: Princeton University Press. Google Scholar CrossRef Search ADS   Bjørnskov, C. 2008. “ Social Trust and Fractionalization: A Possible Reinterpretation.” European Sociological Review  24: 271– 83. Google Scholar CrossRef Search ADS   Bove, Vincenzo, and Leandro Elia. 2017. “ Migration, Diversity, and Economic Growth.” World Development  89: 227– 39. Google Scholar CrossRef Search ADS   Brittain, John C. 1996. “ Direct Democracy by the Majority Can Jeopardize the Civil Rights of Minority or Other Powerless Groups.” Annual Survey of American Literature  1: 69– 77. Catala, A. 2015. “ Democracy, Trust, and Epistemic Justice.” The Monist  98( 4): 424– 40. Google Scholar CrossRef Search ADS   Chua, A. 2004. World on Fire: How Exporting Free Market Democracy Breeds Ethnic Hatred and Global Instability . New York: Doubleday. Cook, K. S., and T. Yamagishi. 1992. “ Power in Exchange Networks: A Power-Dependence Formulation.” Social Networks  14( 3–4): 245– 65. Google Scholar CrossRef Search ADS   Cook, K. S., T. Yamagishi, C. Cheshire, R. Cooper, M. Matsuda, and R. Mashima. 2005. “ Trust Building via Risk Taking: A Cross-Societal Experiment.” Social Psychology Quarterly  68( 2): 121– 42. Google Scholar CrossRef Search ADS   Davis, D. R., and W. H. Moore. 1997. “ Ethnicity Matters: Transnational Ethnic Alliances and Foreign Policy Behavior.” International Studies Quarterly  41( 1): 171– 84. Google Scholar CrossRef Search ADS   De Vroome, T., M. Hooghe, and S. Marien. 2013. “ The Origins of Generalized and Political Trust among Immigrant Minorities and the Majority Population in the Netherlands.” European Sociological Review  29( 6): 1336– 50. Google Scholar CrossRef Search ADS   Delhey, J., and K. Newton. 2005. “ Predicting Cross-National Levels of Social Trust: Global Pattern or Nordic Exceptionalism?” European Sociological Review  21: 311– 27. Google Scholar CrossRef Search ADS   Delhey, J., K. Newton, and C. Welzel. 2011. “ How General Is Trust in ‘Most People’? Solving the Radius of Trust Problem.” American Sociological Review  76: 786– 807. Google Scholar CrossRef Search ADS   Dinesen, P. T. 2012. “ Parental Transmission of Trust or Perceptions of Institutional Fairness: Generalized Trust of Non-Western Immigrants in a High-Trust Society.” Comparative Politics  44: 273– 89. Google Scholar CrossRef Search ADS   ———. 2013. “ Where You Come From or Where You Live? Examining the Cultural and Institutional Explanation of Generalized Trust Using Migration as a Natural Experiment.” European Sociological Review  29: 114– 28. Google Scholar CrossRef Search ADS   Dinesen, P. T., and K. M. Sønderskov. 2015. “ Ethnic Diversity and Social Trust: Evidence from the Micro-Context.” American Sociological Review  80( 3): 550– 73. Google Scholar CrossRef Search ADS   Doerschler, P., and P. I. Jackson. 2012. “ Do Muslims in Germany Really Fail to Integrate? Muslim Integration and Trust in Public Institutions.” Journal of International Migration and Integration  13: 503– 23. Economist Intelligence Unit. 2007–2014. “Democracy Index 2007–2014.” Retrived May 20, 2016, from http://www.eiu.com/. Fairbrother, M. 2014. “ Two Multilevel Modeling Techniques for Analyzing Comparative Longitudinal Survey Datasets.” Political Science Research and Methods  2( 1): 119– 40. Google Scholar CrossRef Search ADS   Fairbrother, M., and I. W. Martin. 2013. “ Does Inequality Erode Social Trust? Results from Multilevel Models of US States and Counties.” Social Science Research  42( 2): 347– 60. Google Scholar CrossRef Search ADS   Farrell, H. 2004. “Trust, Distrust, and Power.” In Distrust , edited by Russell Hardin, 85– 105. New York: Russell Sage Foundation. Fearon, J. D. 2003. “ Ethnic and Cultural Diversity by Country.” Journal of Economic Growth  8( 2): 195– 222. Google Scholar CrossRef Search ADS   Gamble, B. S. 1997. “ Putting Civil Rights to a Popular Vote.” American Journal of Political Science  ( January) 41: 245– 69. Google Scholar CrossRef Search ADS   Grabb, E., R. Andersen, M. Hwang, and S. Milligan. 2009. “ Confidence in Political Institutions in Canada and the United States: Assessing the Interactive Role of Region and Race.” American Review of Canadian Studies  39( 4): 379– 97. Google Scholar CrossRef Search ADS   Guseva, A., and A. Rona-Tas. 2001. “ Uncertainty, Risk, and Trust: Russian and American Credit Card Markets Compared.” American Sociological Review  66: 623– 46. Google Scholar CrossRef Search ADS   Håkansson, P., and F. Sjöholm. 2007. “ Who Do You Trust? Ethnicity and Trust in Bosnia and Herzegovina.” Europe-Asia Studies  59: 961– 76. Google Scholar CrossRef Search ADS   Hanushek, E. A., and D. D. Kimko. 2000. “ Schooling, Labor-Force Quality, and the Growth of Nations.” American Economic Review  ( December) 90: 1184– 1208. Google Scholar CrossRef Search ADS   Helliwell, J. F., and S. Wang. 2010. “Trust and Well-Being.” Working Paper No. w15911, National Bureau of Economic Research. Helliwell, J. F., S. Wang, and J. Xu. 2014. “How Durable Are Social Norms? Immigrant Trust and Generosity in 132 Countries.” Working Paper No. w19855, National Bureau of Economic Research. Hetherington, M. J., and S. Globetti. 2002. “ Political Trust and Racial Policy Preferences.” American Journal of Political Science  46: 253– 75. Google Scholar CrossRef Search ADS   Hooghe, M., T. Reeskens, D. Stolle, and A. Trappers 2009. “ Ethnic Diversity and Generalized Trust in Europe: A Cross-National Multilevel Study.” Comparative Political Studies  42: 198– 223. Google Scholar CrossRef Search ADS   Horowitz, D. L. 1995. “ Democracy in Divided Societies.” Journal of Democracy  4: 18– 38. Google Scholar CrossRef Search ADS   Hwang, M. 2013. “In Canada We Trust? Understanding Ethno-Racial Variations in Social and Political Trust.” Doctoral dissertation, University of British Columbia. Inglehart, R. 1999. “Trust, Well-Being and Democracy.” In Democracy and Trust , edited by Mark Warren, 88– 120. Cambridge: Cambridge University Press. Google Scholar CrossRef Search ADS   Johnson, N. D., and A. Mislin. 2012. “ How Much Should We Trust the World Values Survey Trust Question?” Economics Letters  116: 210– 12. Google Scholar CrossRef Search ADS   Kekic, L. 2007. “ The Economist Intelligence Unit’s Index of Democracy.” Economist  21: 1– 11. Kesler, C., and I. Bloemraad. 2010. “ Does Immigration Erode Social Capital? The Conditional Effects of Immigration-Generated Diversity on Trust, Membership, and Participation across 19 Countries, 1981–2000.” Canadian Journal of Political Science/Revue canadienne de science politique  43( 2): 319– 47. Google Scholar CrossRef Search ADS   Knack, S., and P. Keefer. 1997. “ Does Social Capital Have an Economic Payoff? A Cross-Country Investigation.” Quarterly Journal of Economics  112( 4): 1251– 88. Google Scholar CrossRef Search ADS   Kotzian, P. 2011. “ Public Support for Liberal Democracy.” International Political Science Review  32: 23– 41. Google Scholar CrossRef Search ADS   Kymlicka, W. 2002. Contemporary Political Philosophy: An Introduction . Oxford: Oxford University Press. Lévi-Strauss, C. 1969. The Elementary Structures of Kinship  (No. 340). Boston: Beacon Press. Lewis, J. David, and A. J. Weigert. 1985. “ Trust as a Social Reality.” Social Forces  63( 4): 967– 85. Google Scholar CrossRef Search ADS   ———. 2012. “ The Social Dynamics of Trust: Theoretical and Empirical Research, 1985–2012.” Social Forces  91( 1): 25– 31. Google Scholar CrossRef Search ADS   Lewis, D., 2013. Direct democracy and minority rights: A critical assessment of the tyranny of the majority in the American states . New York: Routledge. Ljunge, M. 2014. “ Trust Issues: Evidence on the Intergenerational Trust Transmission among Children of Immigrants.” Journal of Economic Behavior & Organization  106: 175– 96. Google Scholar CrossRef Search ADS   Marshall, M.G., & Jaggers, K. ( 2000). Polity IV project: Political regime characteristics and transitions, 1800-1999. Unpublished manuscript, University of Maryland, Center for International Development and Conflict Management. Google Scholar Maxwell, R. 2010. “ Evaluating Migrant Integration: Political Attitudes Across Generations in Europe.” International Migration Review  44: 25– 52. Google Scholar CrossRef Search ADS   Mewes, J. 2012. “Inequality and Trust Re-Visited.” Working Paper, accessed June 16, 2016, at http://www.soc.umu.se/digitalAssets/103/103912_11_2012_mewes.pdf. ———. 2014. “ Gen(d)eralized Trust: Women, Work, and Trust in Strangers.” European Sociological Review  30: 373– 86. Google Scholar CrossRef Search ADS   Molm, L. D., M. M. Whitham, and D. Melamed. 2012. “ Forms of Exchange and Integrative Bonds: Effects of History and Embeddedness.” American Sociological Review  77( 1): 141– 65. Google Scholar CrossRef Search ADS   Nannestad, P. 2008. “ What Have We Learned about Generalized Trust, if Anything?” Annual Review of Political Science  11: 413– 36. Google Scholar CrossRef Search ADS   Nannestad, P., G. T. Svendsen, P. T. Dinesen, and K. M. Sønderskov. 2014. “ Do Institutions or Culture Determine the Level of Social Trust? The Natural Experiment of Migration from Non-Western to Western Countries.” Journal of Ethnic and Migration Studies  40( 4): 544– 65. Google Scholar CrossRef Search ADS   Nunkoo, R., and H. Ramkissoon. 2012. “ Power, Trust, Social Exchange and Community Support.” Annals of Tourism Research  39( 2): 997– 1023. Google Scholar CrossRef Search ADS   Okamoto, D., and G. C. Mora. 2014. “ Panethnicity.” Annual Review of Sociology  40: 219– 39. Google Scholar CrossRef Search ADS   Oskarsson, S., Svensson, T. and Öberg, P., 2009. Power, trust, and institutional constraints: Individual level evidence. Rationality and Society , 21( 2), 171– 195. Google Scholar CrossRef Search ADS   Öuberg, P., and T. Svensson. 2010. “ Does Power Drive Out Trust? Relations between Labour Market Actors in Sweden.” Political Studies  58( 1): 143– 66. Google Scholar CrossRef Search ADS   Patsiurko, N., J. L. Campbell, and J. A. Hall. 2012. “ Measuring Cultural Diversity: Ethnic, Linguistic and Religious Fractionalization in the OECD.” Ethnic and Racial Studies  35: 195– 217. Pokharel, R., 2013. Politics and problematics of the definition and categorization of ethnicity in Nepal. Bodhi: An Interdisciplinary Journal , 5( 1), 1– 15. Google Scholar CrossRef Search ADS   Prempeh, H. K. 2013. “ Constitutionalism, Ethnicity and Minority Rights in Africa: A Reply to Jeremie Gilbert.” International Journal of Constitutional Law  11( 2): 438– 43. Google Scholar CrossRef Search ADS   Putnam, R. D. 1993. “ What Makes Democracy Work?” National Civic Review  82: 101– 7. Google Scholar CrossRef Search ADS   ———. 2007. “ E Pluribus Unum: Diversity and Community in the Twenty-First Century: The 2006 Johan Skytte Prize Lecture.” Scandinavian Political Studies  30( 2): 137– 74. Google Scholar CrossRef Search ADS   Ridgeway, C. L., and L. Smith-Lovin. 1999. “ The Gender System and Interaction.” Annual Review of Sociology  25: 191– 216. Google Scholar CrossRef Search ADS   Roscigno, V. J. 2011. “ Power, Revisited.” Social Forces  90( 2): 349– 74. Google Scholar CrossRef Search ADS   Rothstein, B. 2011. The Quality of Government: Corruption, Social Trust, and Inequality in International Perspective . Chicago: University of Chicago Press. Google Scholar CrossRef Search ADS   Rothstein, B. and Stolle, D., 2001, September. Social capital and street-level bureaucracy: An institutional theory of generalized trust. In ESF Conference Social Capital: Interdisciplinary Perspectives, Exeter, UK September 15-20. Prepared for the “Trust in Government Conference” at the Center for the Study of Democratic Rothstein, B. and D. Stolle. 2008. “ The State and Social Capital: An Institutional Theory of Generalized Trust.” Comparative Politics  7: 441– 59. Google Scholar CrossRef Search ADS   Rothstein, B. and E. M. Uslaner. 2005. “ All for All: Equality, Corruption, and Social Trust.” World Politics  58: 41– 72. Google Scholar CrossRef Search ADS   Rotter, J. B. 1971. “ Generalized Expectancies for Interpersonal Trust.” American Psychologist  26: 443– 52. Google Scholar CrossRef Search ADS   Schilke, O., M. Reimann, and K. S. Cook. 2015. “ Power Decreases Trust in Social Exchange.” Proceedings of the National Academy of Sciences  112: 12950– 55. Google Scholar CrossRef Search ADS   Schmidt-Catran, A. W., and M. Fairbrother. 2015. “ The Random Effects in Multilevel Models: Getting Them Wrong and Getting Them Right.” European Sociological Review  32( 1): 23– 38. Google Scholar CrossRef Search ADS   Smith, S. S. 2010. “ Race and Trust.” Annual Review of Sociology  36: 453– 75. Google Scholar CrossRef Search ADS   Soroka, S., J. F. Helliwell, and R. Johnston. 2003. “Measuring and Modelling Trust.” In Diversity, Social Capital and the Welfare State , edited by F. Kay and D. Johnston, 279– 303 Vancouver: UBC Press. Stolle, D., and A. Harell. 2013. “ Social Capital and Ethno‐Racial Diversity: Learning to Trust in an Immigrant Society.” Political Studies  61: 42– 66. Google Scholar CrossRef Search ADS   Sturgis, P., I. Brunton-Smith, S. Read, and N. Allum. 2011. “ Does Ethnic Diversity Erode Trust? Putnam’s ‘Hunkering Down’ Thesis Reconsidered.” British Journal of Political Science  41( 1): 57– 82. Google Scholar CrossRef Search ADS   Tilly, C. 1998. Durable Inequalities . Berkeley: University of California Press. Tilly, C., 2001. Mechanisms in political processes. Annual review of political science , 4( 1), 21– 41. Google Scholar CrossRef Search ADS   Tsai, M.C., Laczko, L. and Bjørnskov, C., 2011. Social diversity, institutions and trust: A cross-national analysis. Social Indicators Research , 101( 3), 305– 322. Google Scholar CrossRef Search ADS   Uslaner, E. M. 2008. “ Where You Stand Depends upon Where Your Grandparents Sat: The Inheritability of Generalized Trust.” Public Opinion Quarterly  72: 725– 40. Google Scholar CrossRef Search ADS   Uslaner, E. M., and M. Brown. 2005. “ Inequality, Trust, and Civic Engagement.” American Politics Research  33: 868– 94. Google Scholar CrossRef Search ADS   Wilkes, R. 2011. “ Re-Thinking the Decline in Trust: A Comparison of Black and White Americans.” Social Science Research  40( 6): 1596– 1610. Google Scholar CrossRef Search ADS   Wilkes, R., and C. Wu. 2017. “Minorities and Trust.” In Oxford Handbook of Social and Political Trust , edited by Eric Uslaner, 231– 49. New York: Oxford University Press. World Values Survey, Wave 5. 2005–2009. Official Aggregate v.20140429. World Values Survey Association, www.worldvaluessurvey.org. Aggregate File Producer: Asep/JDS, Madrid, Spain. World Values Survey, Wave 6. 2010–2014. Official Aggregate v.20150418. World Values Survey Association, www.worldvaluessurvey.org. Aggregate File Producer: Asep/JDS, Madrid, Spain. Wu, C., and R. Wilkes. 2016. “ Durable Power and Generalized Trust in Everyday Social Exchange.” Proceedings of the National Academy of Sciences USA  113( 11): E1417. Google Scholar CrossRef Search ADS   ———. 2017. “The Stats Bear It Out: In Quebec, Trust Is Low.” Maclean’s. http://www.macleans.ca/news/canada/the-stats-bear-it-out-in-quebec-trust-is-low/. ———. 2018. “ Finding Critical Trusters: A Response Pattern Model of Political Trust.” International Journal of Comparative Sociology . doi:10.1177/0020715218761520 Yamagishi, T. 2011. Trust: The Evolutionary Game of Mind and Society . Tokyo: Springer. Yip, P. S., and E. W. Tsang. 2007. “ Interpreting Dummy Variables and Their Interaction Effects in Strategy Research.” Strategic Organization  5: 13– 30. Google Scholar CrossRef Search ADS   Ziller, C. 2015. “ Ethnic Diversity, Economic and Cultural Contexts, and Social Trust: Cross-Sectional and Longitudinal Evidence from European Regions, 2002–2010.” Social Forces  93( 3): 1211– 40. Google Scholar CrossRef Search ADS   ———. 2017. “ Equal Treatment Regulations and Ethnic Minority Social Trust.” European Sociological Review  33( 4): 563– 75. Google Scholar CrossRef Search ADS   Author notes We thank Sylvia Fuller, Howard Ramos, and Rengin Firat for helpful comments on previous versions of this paper. We also thank the anonymous reviewers and Social Forces editor Arne Kalleberg. The article was prepared within the framework of a subsidy granted to the HSE by the Government of the Russian Federation for the implementation of the Russian Academic Excellence Project “5–100.” Support was also provided by Riksbankens Jubileumsfonds: the Swedish Foundation for Humanities and Social Sciences, project no. NHS14-2035:1, and by a grant from the Social Sciences and Humanities Research Council of Canada. Address correspondence to Rima Wilkes, University of British Columbia and Laboratory for Comparative Social Research, National Research University Higher School of Economics, Russian Federation, Department of Sociology, University of British Columbia, Vancouver, BC, Canada; phone: +604 822-6855; e-mail: wilkesr@mail.ubc.ca. © The Author(s) 2018. Published by Oxford University Press on behalf of the University of North Carolina at Chapel Hill. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Social ForcesOxford University Press

Published: Apr 13, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off