A different perspective on exogamy: Are non-migrant partners in mixed unions more liberal in their attitudes toward gender, family, and religion than other natives?

Classic assimilation theory perceives migrant-native intermarriage as both a means to and a result of immigrants’ integration processes into host societies. The literature is increasingly focusing on marital exogamy of immigrants, yet almost nothing is known about their native partners. This explorative study contributes to the literature on migrant integration and social cohesion in Europe by asking whether the native partners in exogamous unions have different attitudes toward gender equality, sexual liberalization, family solidarity, and religiosity/secularization than natives in endogamous unions. Our theoretical considerations are based on preference, social exchange, and modernization theories. We use data of the Generations and Gender Survey (GGS) of seven countries. The sample size is 38,447 natives aged 18 to 85, of whom about 4% are in a mixed union. The regression results of the study are mixed. Persons in exogamous unions show greater agreement with family solidarity, are thus less individualistic than those in endogamous couples. Yet, mixing is associated with greater openness to sexual liberalization and gender equality as well as more secular attitudes. These findings can only partially be explained by sociodemographic control variables. Hence, immigrants in exogamous unions with natives may integrate into the more liberal milieu of their host societies, in which natives continue to place a high value on providing support to family members.


Introduction
Increasing shares of migrants in European host societies have affected receiving countries by creating a change of opportunities on partner markets. It is still true that non-migrant natives are usually in unions with other natives, and migrants are generally in unions with other migrants (Schroedter/Kalter 2008). However, in European countries with larger migrant populations, there appears to be a slight increase in mating preference for members of minority groups as partners (Potârcă/Mills, 2015), and the numbers of intermarriages are on the rise (Qian/Lichter 2011). In Europe, the average crude mixed marriage rate is 0.8 per thousand persons, and about 15% of the total number of marriages in 2006 and 2007 were intermarriages (Lanzieri 2012).
Classic and new assimilation theories perceive marriages between majority and (migrant) minority groups as a means to or/and a result of immigrants' integration into their host societies (Alba/Nee 1997;Aldridge 1978;Gordon 1964). Elwert (2018) highlighted that "intermarriage lies between the poles of social openness and social restraints" (Elwert 2018: 14); it is connected with social structure (Blau 1977;Blau/Becker/Fitzpatrick 1984), and social status (Davis 1941). Numerous studies have focused on the migrant partners in mixed couples in Europe and the US, including on their determinants and trends in partner choices (e.g. Baykara-Krumme/Fuß 2009; Beck-Gernsheim 2006;Chiswick/Houseworth 2011;Chi 2015;Lichter/Qian/Tumin 2015;Pagnini/Morgan 1990). Research on the so-called intermarriage premium effect (Furtado/Song 2015;Lee/Bean 2004) works with the implicit assumption that the native partner facilitates (further) social integration of the migrant partner. However, to date almost nothing is known about the native partners in these unions (Elwert 2018).
This is somewhat surprising, as the literature promotes the perception that people in mixed marriages could be "seen as the ultimate boundary breaker" (Rodríguez-García 2015: 13). This basic assumption, however, has rarely been contested (Milewski/Gawron 2019). Our paper is of an explorative nature and aims to increase our knowledge about the native partners in such unions, who are as diverse as the immigrants themselves.
For the natives, there are different possible types of mixed unions with migrants, which coincide with different societal and legal conditions for meeting, mating, and marrying the potential partner. First, the marriage markets in older member states of the European Union include a growing percentage of descendants of the immigrants who arrived in the 1950s and 1960s (Eurostat 2014). Second, migrants from European and non-European countries today, especially women, move more often as singles than the classic work migrants who arrived in the 1960s did (Anthias/Lazaridis 2000). Some research has indicated that being in a mixed union may foster the possibilities of permanent residence in Europe for migrants from non-European countries (de Hart 2015; Koopmans/ Michalowski/Waibel 2012). Migrants from countries belonging to the European Union have the privilege of freedom of movement (Verwiebe 2008). In the member states of the European Union, the share of European migrants is still lower than the share of non-European migrants (Mau/Büttner 2011), but the number of intra-European couples has been increasing (Schroedter/de Winter/Koelet 2015;Verwiebe 2008). Third, marriage migration may not be a new phenomenon, but as a result of globalization, in which also marriage markets are globalized, it is increasing (Elwert 2018). Non-migrant men and women can meet and mate with potential spouses abroad and "import" them (Constable 2003;Kim 2010;Niedomysl/Östh/van Ham 2010). In the case of marriage migration, the native partner is the starting point of the integration of the migrant into the society at destination. In these marriage market segments, natives -like migrants -may need to overcome legal obstacles created by integration and migration policies, as well as prejudices or even sanctions imposed by their social networks, if they form a mixed marriage (Alba /Foner 2014;Carol 2013;Contucci/Sandell 2015;Rodríguez-García 2010;Schlueter/ Meuleman/Davidov 2013).
In our study, we pose the research question of whether natives in mixed unions are not only a selected group in their attitudes toward diversity (as expressed in their partner choice), but whether they also have more liberal values and attitudes in other social or cultural dimensions compared to their native counterparts in endogamous unions. Our paper contributes to the literature in two ways. First, we investigate natives in a mixed union because they are assumed in the scientific literature to play a prominent role in the integration processes of the migrant partner and the children of the mixed couple. Second, we focus on values and attitudes, which have rarely been included in previous research on exogamy (Huschek/de Valk/Liefbroer 2011), but which are essential for understanding social change. We study attitudes toward sexual liberalization, gender equality, and family solidarity as well as religiosity, because these attitudes are crucial both for individual union formation and for the self-concept of societies. Despite variations within Europe, these dimensions are intertwined because all of them are subject to modernization processes (Gerhards 2010;Inglehart 1997;Norris/Inglehart 2012).
In the second section of our study, we summarize the theoretical considerations and the previous research on partner choice and exogamy, and relate them to social attitudes. Modernization theories serve as the backdrop of our research, as opinions on immigration (Rustenbach 2010) and on diversity (Brown et al. 2018) are seen as important elements in value shifts toward individualism. A mixed couple is defined as a union (marriage or cohabitation) between a native who lives in his/her country of birth and an immigrant who is not living in the country of his/her birth or who is a child of at least one parent who was not born in this country. In the third chapter, we introduce our data, which cover seven European countries of the first wave of the Generations and Gender Survey, and our analytical strategy. In the fourth chapter, we present the results of the multivariate regression models. Finally, we discuss our results and the limitations of our analyses, as well as the implications of our findings for future research.

Mate selection and exogamy from the natives' perspective
Mate selection may follow two main paths. On the one hand, "opposites may attract each other", and on the other hand, mate selection may be homogamous in nature, "'like marries like' -with reference to a wide variety of traits" (Goode 1964: 33), and endogamous, "marriage within the group" (Rosenfeld 2008: 3) (Burges/Wallin 1943). Previous research suggests that both of these paths complement each other (Kalmijn 1998); e.g., partners may start the mate selection process based on a preference for homogamy and then "cast a wider net" to more heterogeneous options if they are not successful. Economic theories, which assume that marriage is about gaining benefits for both partners, support both types. In positive assortative mating, the traits of the partners supplement each other and increase the benefit of a marriage, whereas in negative assortative mating, partners marry when substitutes are most effective (Becker 1973).
Classic status exchange theory (Davis 1941) views intermarriages as a possibility of status exchange (e.g. Gullickson/Fu 2010). This theory assumes that a high socioeconomic status of a partner belonging to a minority group can be exchanged with the majority status of the native partner. In the end, both partners would presumably benefit from the union (Sassler/Joyner 2011). With respect to migrant status, the literature provides mixed evidence on the question of whether exogamy is also associated with heterogamy in other socio-demographic characteristics. Some quantitative studies (e.g. Glowsky 2011;González-Ferrer et al. 2018;Medrano et al. 2014;Schroedter/Rössel 2014;van Wissen/Heering 2014) have shown that exogamy is often accompanied by heterogamy in other traits, such as age, education, and cultural characteristics, and have traced this pattern back to status exchange mechanisms. Others see exogamy as having little connection to heterogamy in other characteristics, and explain intermarriage with a migrant using homogamy/preference theories (Çelikaksoy/Nielsen/Verner 2006). They suggest that resources such as education are more important for partner choice than, for example, ethnicity (Çelikaksoy/ Nekby/Rashid 2010). This pattern may be traced back to a tendency of single people to "sort along class lines" (Choi/Tienda 2017: 302). Social class may have a greater influence on meeting potential partners than ethnicity (Cohen 1977;Lampard 2007).
The mechanisms of mate selection in mixed unions are related not only to social class, but also to gender. Therborn (2004) showed how family and gender roles have changed throughout the world over the years, but pointed to the continuous differences between men and women who are "embedded in social and economic relations of inequality" (Therborn 2004: 127). Men and women in these relations have been influenced by different ideals such as romantic love and how they identify themselves through unions (Giddens 1992), and this affects how they practice gender roles (Evans 2017). Gullickson and Fu (2010) highlighted gender differences in classic status exchange theory (Merton 1941). They suggested that the socio-economic characteristics of minority-group men are more valued than such characteristics of minority-group women on partner markets. According to household economics (Becker 1973), such gender differences are the result of a gendered work division, which maximizes the productivity of the commodities of a family, i.e., through specification in household work for women and labor market participation for men.
In an empirical study, Glowsky (2011) investigated German men who marry women from poorer countries. These native men tend to be retired or unemployed, and they exchange access to citizenship for specific attributes of migrant women, such as migrant women's physical attractiveness and being of younger age than themselves. A similar finding of exchanging youth and attractiveness has been reported for Spain, but for both native men and women who are in a mixed union with a migrant (González-Ferrer et al. 2018). Findings for Germany show that men often search for their mates abroad, rather than finding their partners in the residential migrant population of the host society. By contrast, native women depend more on the opportunity structures of the local partner market and are more likely to find a migrant partner there than to search them abroad (Nauck 2009).
In addition to heterogamy in individual social characteristics and gendered patterns, intermarriages reflect the diversity and the migration history of European countries: While immigration is a relatively new issue in Eastern European countries (Salt 2011), in Western Europe different migrant generations and contexts coexist. In former colonial powers, natives may have more similarities to migrants from former colonies, due to a shared religion, and might have more differences to Eastern European migrants (Medrano et al. 2014). In the older member states of the European Union, growing diversity stems largely from the immigration of Muslims (Modood 2003;van der Noll 2014). For countries in Eastern Europe, the experience of being a host country is new, and the numbers and the origins of the immigrants in these countries differ from those in Western Europe. Immigration patterns also vary greatly by country. In the Baltic countries minority groups are the result of migration in the Soviet era (Puur et al. 2018). For Poland, changing European politics have changed the partner market: women in mixed unions tend to have partners from Western Europe, while men tend to have partners from other Eastern European countries (Slany/Żadkowska 2017). In multiethnic countries such as Russia (Soroko 2018), the role of different Christian confessions creates still other opportunities and barriers for members of majority and minority groups to mate (Werth 2008).
Whereas this literature on mixed partner choice among natives is rather scarce, there is even less literature on the scope, internal dynamics of conflict and negotiation, and social consequences of marital mixing for natives (Rodríguez-García 2015). Beck and Beck-Gernsheim (2013) suggested a shift of power inside mixed unions, which bears the potential for problems in these partnerships. Williams emphasized the exposure of persons in mixed unions to their environment, i.e. they experience scrutiny by almost every part of a society (Williams 2010). Women in a mixed union report that they experience prejudices and discrimination when their partner is a migrant more often than men do (Breger 1998;de Hart 2006).

Exogamy and social attitudes
Previous literature on partner choice has suggested that cultural aspects also matter. Based on preference theory, the partners may have similar values, opinions, and tastes (Kalmijn 1998). Values and attitudes toward culture have rarely been directly investigated in previous research on exogamy, though (Huschek/de Valk/Liefbroer 2011). Rather, empirical studies implicitly assume that native-migrant unions may be associated with cultural differences between the partners, and that the native partner (i.e. mainly referring to Europeans) may hold the "more modern" attitudes. Such stereotyping has been fueled by scientific and public discourses on immigrant assimilation in western receiving countries, which detected a "fault line" (Alba 2005:41) between majority and minority groups caused by religion and/or culture. Specifically, the attitudes toward sexual liberalization and gender equality are seen as crucial for immigrant integration in western countries (Norris/Inglehart 2012). As these dimensions are key for partner choice, we want to elaborate how exogamy and social attitudes could be associated.
According to socialization theories, value orientations evolve in early life, reflect the life conditions in preadolescent years, and are very stable over the life course (Inglehart 1997). Theories of modernization mostly work on the assumption that a shift occurred from materialist values -i.e., safety, familism, collectivism -toward postmodern values -i.e., an emphasis on a high quality of life, self-expression, individualism -in the last decades, especially in Western European countries (Inglehart 1997;Inglehart/Welzel 2010). This value shift is visible in attitudes toward religion, gender roles, or establishing a family (Inglehart/Norris 2003;Norris/Inglehart 2012). Moreover, this shift could be the source of conflicts within a society, because older birth cohorts and more "conservative" individuals may lose their familiar surroundings and must adapt to changing societies (Inglehart/Ponarin/Inglehart 2017). Modernization theories suggest that boundaries based on national or ethnic group identity may become less important if living conditions are secure and self-expression is highly valued. In such societies, individual wellbeing is a crucial factor in fostering (interpersonal) trust (Rustenbach 2010) and tolerance (Inglehart 1997). Thus, modernization processes may also imply an attitudinal shift toward greater openness to population diversity (Inglehart 2018).
What do these modernization theories and previous empirical findings imply for natives in a mixed union with a migrant? The perception of immigration and migrants is mainly influenced by economic determinants; i.e., skilled and highly educated people are more likely to have positive attitudes toward migrants (Mayda 2006). It has also been shown that exogamous unions are more common among people who are highly educated (Carol/Lesczensky 2019) or have postmodern value orientations (Khatib-Chahidi/Hill/ Paton 1998; Mays 2012). Thus, exogamy may be an expression of pro-diversity attitudes. On the other hand, mixed unions may serve as a coping strategy when value shifts have occurred in the society. A mixed union may enable native spouses to adhere to more traditional gender roles (Refsing 1998). One such example are men in the US who marry Asian wives, assuming that these women will be more conservative than those in the US (Constable 2005). A qualitative study in Germany has provided mixed evidence about children and their parents; Khounani (2000) showed that natives in endogamous couples combine aspects of "tradition" and "modernization". Compared to these natives, the partners in a mixed, intra-European couple (in his study, the partners were mainly from Scandinavia) tend to be more "liberal," and the partners in a mixed couple with a non-European partner (usually from Iran or Turkey) tend to have more "traditional" family values and attitudes toward childrearing.
In our study, we concentrate on four dimensions of attitudes which are indicative of individualism/collectivism and modernization: gender equality, sexual liberalization, secularization/religiosity (following Norris/Inglehart 2012), and family solidarity. Perry (2014Perry ( , 2016 showed that people in the US with a higher religiosity in adulthood and religious upbringing are less likely to be in a mixed union. For many natives in Europe, religion is becoming less important (Adamczyk 2013), but public discourses and contention about Islam indicate that the issue of religion -still or again -matters in European societies (Storm 2012). In religiously mixed unions, both partners have to negotiate how they practice their own religion and that of their partner in day-to-day life. They could mix the two, or one partner may give up his or her affiliations and traditions (Tseng 1977).
Religiosity, in turn, is associated with other social attitudes, most prominent among them are attitudes toward sexual liberalization and gender equality (Norris/Inglehart 2012). There is also the role of family solidarity which could account for differences between collectivistic and individualistic behavior, and in which we can differentiate between immigrants and European natives. There is evidence of a higher intergenerational solidarity in migrant families compared to natives (Carnein/Baykara-Krumme 2013) and to families in the migrants' country of origin (Baykara-Krumme 2013). The latter aspects are related to the question of modernization as well (Inglehart 1997).

Working hypotheses
To conclude our literature review and theoretical considerations about associations of mixed couples and attitudes toward culture and religion, we formulate working hypotheses to guide our empirical study. We treat the couple type (mixed vs. endogamous) as the main explanatory variable and attitudes as dependent variables. This allows us to take into account unions of short and longer durations, as well as the possibility of value changes in an existing union. Although values evolve in younger ages and seem to be relatively stable, social attitudes could change over an individual life course. This could be the case depending on aging and changes in the responsibilities for others, or as the result of a highly significant event in society people share, or as the consequence of an individual life event or specific experiences (Hofer/Reinders/Fries 2010). In mixed partnerships, individuals may change their attitudes as a reaction to experiences of discrimination, which could affect our analysis. We could, however, not take into account how attitudes had influenced the partner choice initially or whether/why they have changed in the course of the union. We aim at exploring what attitudes individuals in mixed unions have, in order to test the assumption that the native partners in mixed unions are forerunners in migrant integration and social cohesion.
Hence, our first and main hypothesis proposes an association between exogamy and social attitudes toward gender, family, and religion, with different possible scenarios. On one hand, we can consider mixed unions to be a coping strategy for natives who had been confronted with modernization and had been searching for a more conservative partner with whom they could share these values. These natives may consider religion to be an important factor in life, and sort their public and private spheres along conservative gender and family roles. Exogamy may therefore be associated with more traditional attitudes. On the other hand, we might assume that choosing a migrant partner and living in a mixed partnership requires a person to cross social boundaries and to resist discrimination against migrants in the host society. This would mean that being in a mixed union is an expression of pro-diversity attitudes, which could, in turn, be associated with more liberal attitudes (exogamy and association of attitudes -hypothesis/H1). We estimate separate models for men and women, because previous research has provided different results for men and women.
Second, we account for the socio-demographic composition of the sample. The natives, who have higher education levels, who are younger, who are in a subsequent union, and who live in a non-marital union, may have more liberal attitudes. We examine intra-European variations using country clusters, assuming there are more liberal attitudes in the older EU member states as well as in the northern countries (Gerhards 2010;Liebert 2003;Walby 2004). Controlling for socio-demographic composition of the groups in the sample and country of residence would presumably decrease the differences in attitudes between natives in mixed and endogamous marriages (if there should be any) (compositional hypothesis/ H2).
We control for heterogamy regarding other traits -i.e., comparative education -of the partners. Such heterogamy would moderate the association between exogamy and attitudes. Thus, controlling for heterogamy is expected to reduce the effect of the mixed union (heterogamy hypothesis/ H3).
In addition, we investigate the role of marriage migration in exogamous unions. Relating to discussions of previous studies of men and women in such unions, we would expect natives in a couple with a marriage migrant to be more conservative (marriage migration hypothesis/ H4).

Data and sample
We used the Generations and Gender Survey (GGS), which is an international survey coordinated by the United Nations Economic Commission for Europe (UNECE 2005) designed to yield a better understanding of family relations. It contains data collected between 2004 and 2013 1 . We chose the GGS primarily because it contains information on attitudes of the respondents (Vikat et al. 2007). In addition, it includes a sufficient number of natives in mixed couples for a multivariate analysis. However, these data did have some limitations. We used cross-sectional data, which do not allow us to draw conclusions on causality. The data only contain information on stable unions, and provide rather limited information on the partner. Therefore, we want to emphasize that our analysis has an explorative character.
The GGS data enable us to compare older member states of the European Union and destinations for migrants (Belgium) data from 2008 to 2010, France (2005) (2006), which were subject to temporary restrictions on free movement for the first years of their memberships. Our sample also contains data of the non-EU member state Russia (2004Russia ( to 2008. As our research interest is only about natives in co-residential couples, we excluded migrant respondents (of both the first and the second generation), and singles. We constructed two study samples, separated by gender. The sample for men consists of 18,361 persons; the sample for women comprises 20,086 individuals.
To address the questions regarding marriage migration, we constructed one subsample for each sex (786 women and 696 men) that consists exclusively of natives in mixed unions.

Variables
We used four dependent variables which captured attitudes toward culture and religion. These were sexual liberalization 2 , gender equality 3 , individualism (family solidarity) 4 , and secularization 5 . These items were based on Likert scales and ranged from 1 to 5. For all four dependent variables, a low score indicates a more conservative, collectivistic, or religious point of view, and a high score was connected to more liberal attitudes, individualism, or secularism.
The main independent variable captures whether a person is in a mixed union or not: we defined that those natives, born in the country of residence with both parents also born in the country of residence, are in an endogamous union if their partner is also a native (0). If the partner was not born in the country of residence, or if the partner had not been a citizen of the country of residence since birth, we identify the person as migrant and thus the couple was regarded a mixed union (1). For our subsamples of only exogamous natives and the question of whether the partner is a marriage migrant or not, we considered how much time had passed between arriving in the destination country and the start of the union. Cases with less than one year in the host country before starting a union were defined as marriage migration (1). If the migrant partner had been in the host country longer than a year, migration and union formation were presumably not correlated. Thus, we did not classify such relationships as resulting from marriage migration (0).
We use country of residence as a group variable in multilevel analyses 6 . Additional independent variables accounted for individual socio-demographics, and couple characteristics. The birth years of the respondents were grouped into ten-year intervals. To examine education, we pooled the International Standard Classification of Education (ISCED) levels into one of three groups: low (corresponding to primary education), medium (corresponding to lower and upper secondary level, as well as post-secondary level), and high educational level (corresponding to tertiary education). The education of the partner was constructed in a similar way and was used to pinpoint educational homogamy/ heterogamy between the partners.
For the order of partnership, we used the question of whether the respondent had any previous partnerships and combined the question for the number of union dissolutions with this. If there was any previous partnership, we identified the actual partnership as a subsequent and not the first one. The variable of whether the partners were married (1) or were in a non-marital relationship (0) was constructed by combining the marital status given and the question as to whether respondents were married with their current partner.
Further controls were the number of children and the union duration, which was measured by using the year of data collection and the information of the starting year of the union.

Methods
For the question of attitudes to sexual liberalization, gender equality, individualism (family solidarity) and secularization, we estimated a linear multilevel regression model for each dimension, considering the hierarchical structure between the individual and the contextual level (Olive 2017). This was necessary to account for the differences among European countries in these questions. We present a two-level random effect generalized least squares regression model with the country of residence as cluster (Rabe-Hesketh/Skrondal 2008). Results are displayed as regression coefficients. The first models contain the main explanatory variable of whether the native is in a mixed union or not and the birth cohort. The second models add the rest of the independent variables described above. Our last analysis reduces the sample to exogamous unions in order to account for marriage migration, as this is specific to mixed unions.

Descriptive findings on exogamy
In a bivariate overview of our samples (for women see Table 1.1 and for men see Table  1.2) 4% of the native women are part of a mixed couple, as are 4% of the native men. There are differences between the countries, with the highest frequencies of women in mixed couples in Russia (7%), Belgium (5%) and France (5%), and the lowest in Lithuania (3%) and Poland (1%). The highest numbers of men in mixed couples can be found in Sweden (6%) and Russia (6%), and the lowest in Lithuania (3%) and Poland (1%). Men and women in mixed couples have higher proportions of highly educated individuals; the share of highly educated natives in mixed unions is 7 percentage points higher for men and 6 percentage points higher for women compared to endogamous unions. The share of subsequent unions is higher in exogamous unions (among women 10 percentage points and among men 11 percentage points). Marriage migration is more common for native men (25%) than for women (19%).
Native women have higher scores for attitudes toward sexual liberalization and gender equality, i.e., women are more "liberal" in social attitudes, however they are more religious than men. There are hardly any differences for family solidarity (higher scores indicating a higher degree of individualism and lower scores more collectivistic/familistic attitudes) between the sexes. By country, men and women in the older member states of the European Union have higher scores for such attitudes than those natives in Eastern Europe and Russia, with the most liberal attitudes in Sweden (results not displayed). By couple type (see Table 1.3), both native men and women in mixed couples throughout Europe have slightly higher mean scores for secularization and for sexual liberalization. Native men in mixed couples also have slightly higher mean scores for attitudes toward gender equality. Again, there are no significant differences for family solidarity.

Sexual liberalization
Belgium

Multivariate results for attitudes toward culture
Results for all four attitudes are displayed in Table 2.1 for women and 2.2 for men. 7 We start by testing our main hypothesis of an exogamy effect (H1).
The first dimension is sexual liberalization. If we only control for couple type and birth cohort (W1.1 & M1.1), women and men have significantly higher scores -indicating more modern attitudes -if they are in a mixed union. This effect becomes smaller and no longer significant for native women (W1.2), if we control for individual determinants and couple characteristics. And also for men the effect gets weaker after controlling, but remains significant with a 0.1 point higher score for being in a mixed union (M1.2). Birth cohort (Ref.: 1930(Ref.: -1939(Ref.: ) 1924(Ref.: -1929  results for the mixed union as well as for the controls like for the pooled sample. But only few of the effects were significant, which is surely related to the small sample size and the large confidence intervals.  For the second question of gender equality, the first models produce a pattern similar to that for sexual liberalization, which suggests that people in mixed unions have more modern attitudes. Note that model W2.1 shows that the effect is insignificant for native women, whereas model M2.1 indicates that the effect of being in a mixed union is significant for native men, with a score that is more than 0.1 points higher. The insertion of further independent variables decreases the effect of the couple type, and, interestingly, changes its direction for women (W2.2). Hence, for native women, exogamy is shown to be associated with having more traditional attitudes toward gender equality, all other variables held constant. The third dimension is individualism based on attitudes to family solidarity. Here we find a different pattern than in the previous two dimensions. Under control of characteristics of couples and individual determinants, native women and men in mixed unions have significantly lower scores for individualism; i.e. they have more collectivistic attitudes and support family solidarity more if they are in a mixed union (W 3.2 & M 3.2).
The fourth question relates to secularization and religiosity. Similar to the attitudes toward sexual liberalization and gender equality, being in a mixed couple has a positive (and significant) effect for native men and women (W4.1 & M4.1), indicating their higher degree of modern attitudes. In terms of effect size, mixing has the greatest impact on secularization among all four dimensions tested. Controlling for characteristics of couples and individual determinants, the effects become smaller both for men and women, but the direction remains the same (W4.2 & M4.2). For native men it is still significantly elevated if they are in a mixed union, whereas we can no longer find a significant effect for native women.

Effects of controls
In line with previous literature, our findings show that for all four attitudinal dimensions, being younger is connected to more openness toward modernization, and being older is associated with having more conservative values. Having a higher level of education is linked to having more liberal views, and living in a non-marital cohabitation is associated with more openness. However, we also find that these individual determinants have only a small influence on the effect of the couple type, which only partially supports our working hypothesis regarding compositional differences between exogamous and endogamous unions (H2).
The couple type effect is moderated by other independent variables. In our exogamy hypothesis (H3), we had assumed that exogamous unions are more likely to be subsequent unions, and that such couples are also likely to be more heterogamous in terms of other traits. Indeed, the main effect of the couple type becomes weaker when controlling for the order of partnership. Thus, being in a subsequent union is connected to having more individualistic and more secular attitudes. However, the controls for educational homogany/ heterogamy are not as clear as hypothesized. Compared to women in couples with educational homogamy, women who are less educated than their partner tend to have more modern attitudes regarding sexual liberalization, gender equality, and secularization, but more traditional attitudes regarding family solidarity. By contrast, women who are more educated than their partner score higher in individualism, but have less modern attitudes regarding sexual liberalization, gender equality, and religiosity. For men, we find that being better educated than their partner is associated with lower scores and less modern attitudes, while being less educated than their partner is associated with higher scores, and, thus, with having more liberal attitudes (except regarding individualism (family solidarity)). Hence, we find a gendered pattern in which men tend to display "modern" behavior when their partner is better educated than they are. These men also have more liberal attitudes. But for women, we find that although being better educated than their partner is a rather modern pattern of partner choice, it is not associated with having more modern attitudes.
The control variable on union duration produces similar results for each of the four dimension; the longer a union lasts, the more liberal, individualistic, and secular are their attitudes. The results on the control for the number of children indicates that having a larger family is connected with more traditional attitudes toward sexual liberalization and religiosity, but with more modern/individualistic attitudes toward gender equality and intergenerational obligations. Regarding the country of residence, we find large differences between the Western and Eastern European countries for all four dimensions. People in Poland, Lithuania, and Russia are significantly more conservative and supportive of the collective and the family than the women and men in the older member states of the European Union. We also see small differences within the cluster of the older member states; we find the most liberal and the most individualistic attitudes in Sweden as compared to natives living in France, Belgium, and Germany. These results are in line with previous literature, which demonstrated differences between European countries. 8

Marriage migration
We also estimate models to get a better understanding of the role of marriage migration ( Table 3). Note that the sample size is even smaller for this subsample and the confidence intervals become rather large. Therefore, we refrain from interpreting the significance and size of the coefficients, but explore the direction of the effect only. In principle, we find a similar pattern as for the samples of all persons in unions when we compare those natives whose partner is a marriage migrant to those in a mixed union with a migrant who did not move for union formation (be it either a single first-generation migrant, or a migrant child born abroad or at destination of his/her migrant parents). Having a marriage migrant as partner is associated with higher secularization and more openness toward sexual liberalization and gender equality. Thus, our working hypothesis regarding a higher degree of traditionalism among natives in a couple with a marriage migrant has to be rejected (H4). The effects of the independent variables are similar to those described in section 4.3. Note: * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001.
Models 1 controlled for birth cohort, Models 2 additionally controlled for individual variables and characteristics of the couple.
8 Moreover, we find that exogamy is associated with almost the same results across Europe, so that the moderating effect of the country differences cannot explain our results.

Discussion and conclusion
As intermarriages reflect opportunity and social structures, as well as the host society's openness to immigrants, such unions are special cases in terms of mate selection. Moreover, intermarriages are indicative of majority-minority group relations, and, thus, of the level of social cohesion in the country. Consequently, scholars who have studied intermarriage generally agree that mixed unions facilitate migrant integration, and therefore reflect attitudes toward out-groups and the level of social change in a society. Whether mixed unions are long-lasting or end in disruption and divorce may serve as the ultimate test of majority-minority relationships. In addition, examining mixed unions is helpful for understanding the socialization of migrant children, as children born to parents in mixed unions have to deal with their parents' different cultural and social backgrounds. Given that it is often assumed that the native partner in a mixed union influences the migrant partner and their joint children, we found it somewhat surprising that only a few studies have actually investigated the characteristics of the native partner in a mixed union. Thus, the aim of our research was to examine the attitudes of natives in mixed unions. This empirical study was based on the underlying assumption that the decision to enter a mixed partnership reflects openness to out-groups and population diversity. Against this background, we posed the research question of whether the social attitudes of natives in exogamous unions differ from those of natives in endogamous unions.
In order to investigate this question using quantitative data, we had to deal with the well-known scarcity of data on minority populations in Europe. In the end, we decided to use pooled data from seven European countries (GGS), as this was the only way to obtain a sample that included a sufficiently large number of mixed unions, as well as information on the attitudes of the partners. However, our study has several limitations. Although the number of mixed unions in Europe has been growing, the prevalence of intermarried natives is still rather small. Hence, our sample size of natives in exogamous unions was rather small, and the size of the effects found was also quite small. It is important to keep these limitations in mind in the following discussion, in which we offer our interpretations of the results of our analysis.
Our main working hypothesis was that there is an association between being in a mixed marriage and attitudes. We used four different attitudinal dimensions to test this hypothesis, all of which are included in modernization theories, and are correlated with attitudes toward diversity. Three of these dimensions -sexual liberalization, gender equality, and religiosity/secularization -feature prominently in public and scientific debates on migrant integration and cultural diversity in Europe today. The fourth dimension captures attitudes toward family solidarity. This dimension is also related to modernization, but has received less attention in current integration debates. We worked with competing hypotheses. First, we posited that compared to natives in an endogamous union, natives in a mixed union may be more resistant to modernization; i.e., they may be more religious, and they may have more conservative attitudes regarding gender and family roles. If these natives are more "traditional" than average, they may perceive immigrants or marriage migrants as secondary options in the partner market. Thus, exogamy may be a means of coping with resistance to modernization. Crossing ethnic boundaries could imply that a native is less liberal in terms of his/her other attitudes. Alternatively, it could mean that the native is more liberal in terms of his/her attitudes regarding culture and religion; i.e., that s/he is more open to modernization in general, and has positive views on diversity. In the latter case, exogamy would also be associated with more individualistic attitudes on other dimensions. Interestingly, our study provided evidence for both hypotheses. Being in a mixed marriage was found to be associated with higher levels of secularization and more openness towards sexual liberalization and gender equality, i.e., pointing toward more individualistic values, but with more traditional attitudes toward family solidarity; i.e., pointing to more collectivist values.
These findings are noteworthy for three reasons. First, family solidarity is one dimension of modernization theory that is also subject to change. We expected to find that as individualism increased, family solidarity would also become less important, especially as we used variables that referred to intergenerational relationships and to support of the elderly by the younger generations. In modern Western European countries (despite crosscountry variation), care responsibilities are often delegated to the welfare state. We may conclude from our finding that exogamy was associated with a higher degree of familism, and that crossing the "ethnic line" in union formation is a modern behavior that reflects traditional attitudes. Perhaps these natives perceived the choice of a migrant partner as a secondary option that enabled them to form a family; i.e., they cast a wider net in order to avoid being single. Crossing ethnic boundaries when choosing a partner was, however, found to be associated with having more liberal attitudes toward sexual liberalization and gender equality, and with lower religiosity. Our observation that these three dimensions are linked is in line with the literature. But our findings challenge Inglehart's individualism/collectivism dichotomy. In "real life," individuals may "mix" traditional and modern elements, and show a value synthesis (Klages 2002). Exogamous unions may be such a case, as it appears that the partners often have divergent attitudes.
Second, we assessed the role of marriage migration. Our assumption was that natives who are in a union with a marriage migrant had more conservative attitudes than natives in other exogamous unions. However, our results did not support this hypothesis. As this subsample was very small, we were unable to examine this potential association in detail. We could not, for example, distinguish the migrant partners by their migrant generation, country of origin, or religious affiliation. Placing all migrant partners in one category may be too simple, as it does not account for the diversity in European immigrant populations. This limitation may explain this finding. Whether a European native married a person from a religiously distinct country (Foner/Alba 2008;Koopmans 2015), a first-generation migrant from another EU country, or a member of the second migrant generation might also make a difference.
The third point we want to raise is that the pattern of the association between exogamy and attitudes is similar for women and men (regardless of whether they are marriage migrants). Hence, our results do not indicate that there are gender differences in the likelihood of using exogamy as a strategy for coping with modernization processes, or in openness to diversity. If exogamy is a strategy for coping with modernization or constraints on the partner market, then its association with social attitudes would be similar for both sexes. However, we found differences by gender in the socio-demographic background variables. The education of the respondent and the comparative education of the partner were shown to have more explanatory power for the exogamy effect among wom-en than among men. This result suggests that future research should continue looking into mixed partner choice from the natives' perspective, with a focus on gendered patterns and social attitudes.
Unfortunately, our analyses were based on cross-sectional data. Hence, the study of attitudes allows for two potential interpretations. Attitudes may influence either the partner selection as such or the reason for seeking the union. Future research should, therefore, investigate how attitudes affect partner choice among natives and immigrants. We also need to acknowledge that our sample contained only existing unions. Previous literature has shown that exogamous unions have a higher risk of dissolution than endogamous unions (e.g., Milewski/Kulu 2014). At the same time, divorce tends to be associated with more modern attitudes. Therefore, it is likely that natives who ever had an exogamous union were underrepresented in our sample. We would, however, expect that this underrepresentation led to an underestimation of the impact of exogamy on attitudes.
As cultural differences between the partners in mixed unions are often cited as potential explanations for difficulties in the partnership or union disruption, future research should include information on values, attitudes, and opinions in analyses, and, of course, in data collections. Ideally, future studies would also compare the attitudes of the partners in each couple, which we were not able to do with our data source. We may speculate that what matters in determining whether a union endures is not just how conservative or liberal the partners are, but whether they are well-matched and get along.
Finally, we want to take a step back and focus our attention on the prevalence of intermarriage among natives in Europe. Although some of the previous literature reported that the number of mixed unions has been increasing, the prevalence of exogamous unions among natives is still rather small: i.e. of the natives in our data, only about 4% were in a mixed union. We suggest that future research on modernization processes and social cohesion focus not only on gender issues and religiosity, but on attitudes regarding diversity in a perhaps increasingly hostile Europe (Turper et al. 2015); and investigate how these dimensions are interrelated, and how the societal climate affects partner choice and the family life in mixed unions.
To conclude, our findings imply that the social adaptation process of migrant partners may be facilitated by exogamy in partnerships, because migrants may integrate into a more liberal milieu if they have a native partner. At the same time, the natives in these mixed marriages tend to have more traditional attitudes toward the family in general. Both the percentages of mixed unions and the patterns of our results regarding the attitudes of natives in such unions suggest that the widespread assumption that exogamy is the ultimate boundary breaker and an indicator of social cohesion in increasingly diverse European immigration countries should be challenged more frequently in empirical studies.