Attitudes Towards Intergroup Dating Among Brussels Adolescents: The interaction of gender and ethnicity and its determinants

Social distance is an important component to consider when analysing the challenges of multicultural societies. According to Shibutani and Kwan (1965), social distance is the subjective state of nearness felt for certain individuals and refers to the perception that the other belongs to a different category. This impedes collective identities and shared experiences (Alba and Nee 1997). Accordingly, a decrease in social distance or the blurring of group boundaries fosters immigrants’ structural integration (Shibutani and Kwan 1965). Thus, investigating the mechanisms underlying social distance of both ethnic minorities and the majority group is crucial for understanding the challenges faced by the meanwhile multicultural societies of Western Europe. Research on social distance includes among others intergroup friendships (e.g., Van Houtte and Stevens 2009), partnerships (e.g., Hartung et al. 2011) and intergroup marriages (Lucassen and Laarman 2009). Among those indicators, unions are considered as the strongest measure of social distance because of the high level of intimacy. Social distance is defined as the extent to which people accept contact with members of ethnic out-groups (Bogardus 1933). To date, most of the previous European studies focused on effective intergroup marriages (e.g., Muttarak and Heath 2010; Schroedter 2012) leaving aside intergroup dating preferences. As non-marital unions are gaining in importance, this study analyses attitudes towards intergroup dating as an indicator of social distance.

However, intermarriage data do not allow disentangling the role of individual preferences from the role of opportunities in shaping intergroup love relationships: intergroup marriages are the outcome of both preferences and opportunities. Indeed, the sole investigation of behaviour does not enable to conclude that the low intergroup marriage rates are the result of individual preferences: low intergroup marriage rates might be due to the lack of structural opportunities to meet out-group members. Therefore, determinants of intergroup dating preferences among both ethnic minorities and the majority are crucial to investigate: they can shed light on the mechanisms of the main individual factor driving intergroup marriages. Intermarriage preferences can be helpful for the prediction of actual intermarriage behaviour; both are significantly linked to each other (Carol 2013b). Yet, marital decisions are not only a joint decision of two people from two different ethnic backgrounds (Feliciano et al. 2009); they often involve third parties such as parents and communities that have an interest in intervening, especially in intergroup relationships, which threaten the transmission of traits (Kalmijn 1998). If two people with competing traits marry each other, some traits will prevail while others dwindle. In that sense, intergroup relationships pose a threat to the existence of communities, which have an interest in the preservation of their cultural traits. Particularly intergroup relationships of women and girls are regarded critically if the traits of the male partner succeed and are transmitted to following generations. In this case, intergroup relationships of women are sometimes perceived as a loss by communities (Lucassen and Laarman 2009). In this context parents and religious or ethnic communities play a prominent role. This study goes beyond existing research by emphasizing the role of these third parties driving individual attitudes. As we are able to show attitudes towards intergroup dating are indeed related to parental monitoring, religious attachment, friendship networks and attitudes about premarital sexuality. In research on behavioural indicators of social integration, it has been shown that individuals who have their first sexual relationships with an out-group member are also more likely to marry with an out-group member (King and Harris 2007). Thus, early relationship patterns are likely to persist across the life course and hence important to study. To investigate this, we need to draw attention to an earlier stage of life where the foundation for these relationships is laid – in adolescence when intergroup dating becomes an important issue.

In addition, analysing attitudes towards intergroup dating and their determinants among adolescents of a traditional West European immigration country enables to assess the extent to which the next generation of citizens will cope with cultural diversity. Indeed, a large part of adolescents with an immigrant background in such countries constitute meanwhile the so-called second generation and have thus been entirely socialized in the receiving society. Moreover, ethnic majority adolescents have grown up in an already culturally diverse society. Yet, most of the existing studies on intergroup dating preferences among adolescents focus on the US context, investigating the puzzling high reluctance of certain groups for black–white dating (e.g., Levin et al. 2007). While these US studies provide insights into potential mechanisms leading to the disapproval of intergroup dating, their findings can hardly be generalized to West European contexts, because of the very different ethnic intergroup relations and immigrant groups that are to be found in Western European and in the US contexts (e.g., Koopmans 2013).

In this study, we propose to fill in this gap in the European literature by investigating individual attitudes towards intergroup dating (i.e., the openness to love relationship with someone from another cultural or religious group) and their determinants among 18-years-old adolescents from various cultural and religious groups in Brussels, Belgium. Brussels is characterized by a large cultural diversity: 28% of the population in Brussels does not have the Belgian citizenship (INS 2007). The three most represented countries of origin among Brussels population from non-EU origin are Morocco, Turkey and Democratic Republic of Congo (DRC, Belgium's largest former colony; Willaert and Deboosere 2005). Besides these immigration waves, Brussels – as EU and NATO headquarters – also comprises highly qualified elite of non-Belgian EU citizens. This cultural diversity is not only composed of first-generation immigrants but also meanwhile of a significant proportion of second-generation adolescents. Our representative sample enables us to differentiate between 11 ethnic groups with diverse religious roots consisting of ethnic majority and minority adolescents with an Islamic, Catholic, Protestant, other or no religious background. This remarkable sample heterogeneity provides insights into the role of religiosity and values tied to religiosity such as familial bonds and attitudes towards premarital sexual relationships. These have been identified as crucial in dividing groups (e.g., Hennink et al. 1999). Importantly, this will allow us to disentangle the effects of religiosity and ethnicity. Previous research that mostly relied on public data was not able to conclude whether religion or ethnicity makes a difference. Besides ethnic differences, we will closely look at the interaction of gender and ethnic origin in attitudes towards intergroup dating. Compared to previous research that took gender and ethnic differences for granted without probing its causes, we aim to explain emerging differences through parental monitoring, religiosity and associated values, which are likely to affect the attitudes of boys and girls with and without an immigrant background differently. In this contribution, the term of ethnic majority will refer to adolescents whose parents were born in Belgium, while the term of ethnic minority will be used to define adolescents born abroad or in Belgium with at least one foreign-born parent.

Previous studies in West European contexts highlighted significant ethnic differences in attitudes towards intergroup relationships: Huijnk (2011) showed that immigrants with a Surinamese and Antillean origin are more likely than Dutch natives to favour intergroup marriage and that respondents of Turkish and Moroccan origin oppose intergroup marriages to a larger extent than the other ethnic minority groups. Similarly, Lucassen and Laarman (2009) showed that the effective intergroup marriage rates differ across immigrant groups in Western Europe, with the groups from countries with large Muslim majorities having the lowest intergroup marriage rate. These ethnic minorities from countries with large Muslim majorities have in common the particular importance of family honour and values tied to it (Smith et al. 2006: 234). Based on these previous studies, we expect significant ethnic differences in attitudes towards intergroup dating among Brussels adolescents. More precisely, we expect adolescents from the more religious countries with large Muslim majorities (such as Morocco and Turkey) to oppose intergroup dating the most.

Gender is also a key determinant for attitudes towards intergroup dating: intergroup marriage rates differ among men and women from both ethnic minorities and the majority (Lucassen and Laarman 2009). With regard to attitudes towards intergroup dating, Huijnk (2011) shows that females from ethnic minority and the majority group in The Netherlands oppose intergroup marriage to a larger extent than males. Feliciano and her colleagues (2011) showed that the gender gap in attitudes towards intergroup dating differ for White and Latino American Internet daters. They conclude that ethnic origin interacts with gender in shaping attitudes towards intergroup dating. Transposing these results to the European context would imply that the mechanisms leading women to oppose intergroup dating to a larger extent than men seem to vary for the ethnic majority and minority populations. We can nevertheless expect girls to oppose intergroup dating to a larger extent than boys.

Central to this study is the aim to explain ethnic and gender differences in intergroup dating by several mechanisms that relate to norms, values, parental relationship and peer network. Indeed, Kalmijn (1998) identified three mechanisms underlying intergroup relationships: third parties, contact opportunities and individual preferences mechanisms. These can be applied to different measures of social integration (Martinovic 2010). We will investigate these mechanisms by analysing the role of parents (so-called third parties), religiosity and intergroup friendships in shaping intergroup dating. We expect these factors to function as moderator for the association of ethnic origin and gender with attitudes towards intergroup dating. While all these factors are likely to affect attitudes towards intergroup dating of ethnic majority and minority adolescents as well as girls and boys, some of them might play a larger role either among the ethnic majority, ethnic minorities, boys or girls and thus might explain why ethnic and gender differences on the approval of intergroup dating are significant.

Through socialization, minority and majority parents transmit their values and norms to their sons and daughters. These also include preferences about their son's/daughter's future spouse. It is through vertical transmission processes of cultural traits that groups can ensure their survival. As briefly discussed in the Introduction, intergroup relationships can potentially undermine these attempts if the relationship exists between members with competing traits. From an evolutionary perspective, some traits will trump others, leading to an eradication of those traits and going against the interest of communities. However, these transmission processes differ in their intensity across groups with minority parents sometimes having a keen interest in the transmission as their cultural traits are potentially at a greater stake than those of majority parents (Mchitarjan and Reisenzein 2014; e.g., Idema and Phalet 2007; Maliepaard and Lubbers 2013).

Socio-psychological studies showed that teenagers’ consideration of intergroup dating is related to parental wishes for endogamy in partnerships (e.g., Edmonds and Killen 2009; Uskul et al. 2011). Moreover, discrepancies between parental values and their son's/daughter's values might be significantly related to intergroup dating: Kwak (2003) has argued that parent–child value discrepancies result in intergenerational conflicts which affect immigrant adolescents’ integration outcomes. However and to our best knowledge, only few quantitative studies (e.g., Hunter et al. 2012; Mok 1999; Munniksma et al. 2012; Reinders 2004) have investigated the way parents shape their son's/daughter's intergroup behaviour. More recent studies related parental preferences for the intergroup behaviour of their sons and daughters to norms and values (e.g., Munniksma et al. 2012). Indeed, ethnic minority parents’ values sometimes stand in stark contrast to values that are widespread in Western Europe's receiving societies. Ethnic minority families emphasize obedience, intergenerational responsibility and solidarity within families to a greater extent than families without immigration background (Huijnk 2011). As a consequence, ethnic minority adolescents who adopt views from receiving society members but have parents who preserve their views from their country of origin find themselves in a dilemma between parental values and those transmitted in the majority group (e.g., Merz et al. 2009; Portes and Zhou 1993). This intergenerational discrepancy is not only higher among ethnic minority families than majority families but also present among majority families (Phinney and Vedder 2006). Discrepancies in these values, in turn, might have consequences for social integration – for both majority and minority adolescents. Although adolescents might be more willing to intermingle with out-group members, their parents are sometimes doubtful about the consequences of intergroup contact, which affects adolescents’ intergroup mixing as Edmonds and Killen (2009) pointed out. Therefore, we will explore whether parent–child discrepancies are associated with attitudes towards intergroup relationships among both ethnic majority and minority adolescents.

Parental discomfort also concerns premarital sexuality, which is a crucial theme among individuals of different religious denominations and ethnic groups (e.g., Hennink et al. 1999), because it is linked to the continuity of groups and the demarcation between in-group and out-group members (Buitelaar 2002). While love relationships outside of marriage are regarded as common in Western Europe, some ethnic groups perceive them as putting marriage prospects and cultural heritage at risk. While in Europe mostly the affinal kinship regimes occurs (where relationships are based on love and the decision to marry is taken by the couple), the descent kinship regime, existing for instance in some Asian countries is characterized by a high degree of intergenerational solidarity, where parents are included in the decision-making process (Nauck and Suckow 2006). In these kinship regimes, the honour of families is strongly tied to the behaviour of girls and their chastity. For this reason, premarital relationships are perceived as endangering marriage arrangements (Bradby 2006). Thus, families are more protective of their daughters than their sons (e.g., Gapp 2007). According to Hennink et al. (1999), this behaviour is more prominent among families who follow Islamic traditions. Moreover, it has been shown that girls and boys (from the majority) largely internalize the attitudes of their own parents towards premarital sex (Thornton and Camburn 1987). We will therefore examine whether adolescents’ attitudes towards premarital sex are negatively related to their attitudes towards intergroup dating and whether this relationship is stronger for girls.

Parents counter their son's/daughter's intergroup relationships by exercising social control (e.g., Updegraff et al. 2010). Indeed, Reinders (2004) showed that parental control decreases the likelihood of intergroup contacts among both German and Turkish–German adolescents. Thus, parental control is expected to affect attitudes towards intergroup dating of both ethnic majority and minority adolescents. Furthermore, Munniksma et al. (2012) showed that parental acceptance of intimate intergroup relationships was generally lower for daughters than for sons. Therefore, we expect parental control to affect girls’ attitudes towards intergroup dating to a larger extent than those of boys.

Homophily means favouring partners who share certain characteristics and values. Lazarsfeld and Merton (1954) distinguish two types of homophily. The first one refers to status homophily and is based on ascribed or acquired characteristics such as ethnicity and religion. The second type is value homophily: similarities in attitudes, beliefs and values enhance attraction and interaction. Accordingly, attitudes towards intergroup dating are likely to be shaped by religiosity: individuals with differing religious beliefs and practices will be less inclined to date someone outside their (religious) community. However, females and males are attributed different roles according to the traditions and texts of the main religions (Glick et al. 2002): females are regarded as transmitters of cultural norms and values and thus embodying the continuity of cultures. This is particularly pronounced in religious communities. This means that intergroup boundaries for dating should be more salient for religious girls than for religious boys.

A perceived gap in religiosity, norms and values can be reflected in ethnic hierarchies, where for instance, majority group members favour some groups over others (Sniderman and Hagendoorn 2007). At the extreme end, these ethnic preferences are linked to perceived symbolic threat arising out of these differences and resulting in increased social distance, negative out-group attitudes or stereotypes.

The last mechanism that is likely to shape attitudes towards intergroup dating concerns contact opportunities. Accordingly, adolescents with a culturally diversified network of friends and acquaintances are likely to be more willing to date an out-group member than adolescents with a homogeneous peer network: in addition to reducing ethnic prejudice (Allport 1979), heterogeneous peer networks provide opportunities for dating and relationships (e.g., King and Harris 2007; Levin et al. 2007). We will therefore expect adolescents with a large amount of intergroup friends to be more willing to engage in intergroup dating.

The data were gathered in secondary schools in 7 out of the 19 municipalities of the Brussels-Capital Region, which are representative of the geographic and demographic diversity of Brussels. The response rate of schools was 88%. The response rate among students amounts to 70% (Teney 2009). Altogether, 3121 self-administered questionnaires were filled in by pupils attending the last year of their secondary education. The number of respondents represents about one-third of Brussels school population, meaning all pupils who attend their last year of compulsory education in 2007. Pupils were on average 18 years old.

The dependent variable is the binary response of the following item: ‘Could you imagine yourself having a loving relationship with a person from another cultural or religious group than yours?’ The main explanatory variables are parental control, parent–child discrepancies, attitudes towards premarital sex (sexual conservatism) and religiosity. These constructs were measured by using several items.

Parental control is assessed by six items, each measuring whether at least one parent knows what the child does after school, where he/she goes out, how the pocket money is spent, what he/she does in leisure time, which friends he/she has, and whether parents know about their child's school notes. Each of these was indicated on a five-point scale ranging from one (never) to five (always). Exploratory factor analyses based on a polychoric correlation matrix for categorical variables resulted in a one-factor solution. The items have a Cronbach's alpha of 0.81. For the analyses, the row mean was used.

Parent–child discrepancies were measured by two sets of items which addressed both the mother and the father: ‘I often quarrel with my father/mother’ and ‘I have the feeling that my father/mother does not understand me at all’ on a five-point scale ranging from one (totally disagree) to five (totally agree). These four items load on one factor with a Cronbach's alpha of 0.75. Again, the row mean is used for analyses. That means, if only the evaluation of one parent was available, the average of the items referring to that parent was used.

Sexual conservatism is composed of four items: ‘The use of condom should not be promoted’, ‘One should have sexual intercourse only to have children’, ‘It is preferable not to have any sexual intercourse before getting married’ and ‘Boys but not girls can have sexual intercourse before getting married’. Each of these items was measured on a scale ranging from one (totally disagree) to five (totally agree). A one-factor solution was retained. The Cronbach's alpha is 0.72. The items were averaged to form the scale.

The religiosity measurement combines religious identity and practice. It consists of three questions. For the first question ‘Do you have things in common with people who belong to the same religion as you?’ respondents indicated the strength of their identification on a scale from one (nothing in common) to five (a lot in common). A second question measures whether respondents consider themselves as religious. The scale varies between one (not believing) and five (very religious). The last question measures how often adolescents attend religious services (1: never–7: everyday). Again, these items load on one factor. The factor scores were saved and used in the analyses. The Cronbach's alpha is 0.76.

Intergroup friendships were measured with the item: ‘How many persons among your friends and acquaintances are of the same ethnic origin as you?’ The categories range from one (all) to five (none).

Moreover, we included the following socio-demographic characteristics in the analysis: gender, educational track (vocational, technical and general), socio-economic status of the family (low, middle, high) and ethnic origin. Ethnic origin was measured by the parents’ country of birth. Pupils with at least one foreign-born parent are categorized as ethnic minority adolescents. If both parents are from different foreign countries, the birth country of the mother was decisive for the grouping. Besides several nationalities that are well represented in the data (Congolese, Turks, Moroccans and Belgians), the 92 other origin countries are regrouped into regions of origin (Belgian, n = 1300; sub-Saharan African, n = 111; North African, n = 88; Asian, n = 70; Middle Eastern, n = 40; East European, n = 64; South European/American, n = 280; Moroccan, n = 593; Congolese, n = 221; Turkish, n = 151; and Western Countries, n = 171). The immigrant generation is measured with pupils’ country of birth: 83.2% of the pupils were born in Belgium.

We use linear probability models with robust standard errors for the analysis. The parameters were estimated with the full information maximum likelihood estimator which includes incomplete data records in the estimation procedure (Enders 2010) and increases the number of cases included in the model. It yields comparable results to ordinary logistic regression.¹ The variance that is due to the nested structure of the data (i.e., pupils nested within schools) of the models comprising the socio-demographic characteristics of the pupils is not significant (school variance: 0.002, school standard error: 0.002). Moreover and for the sake of model convergence, we avoid interaction terms between the 11 ethnic origin categories and the other independent variables. Nevertheless, we ran separate models for ethnic minority and the majority adolescents (results available upon request). Since the effects of the independent variables have a similar magnitude and go in the same direction for ethnic minority and majority adolescents, we present the results of a common regression model comprising both ethnic minority and majority pupils.

Table A1 (in the  Appendix) presents the descriptive statistics. Taking a look at the distribution of the dependent variable – approval of intergroup dating – reveals that most pupils approve of intergroup dating, indicating a relatively low social distance (for ethnic differences, see Table B1 in the  Appendix). The level of approval resembles the level of approval of religious intermarriage among adults (Carol 2013a). However, taking a closer look, only half of the minority females would consider intergroup dating (p < .001). Further ethnic and gender differences emerge: minority adolescents are overall more religious than majority adolescents (p < .001) and differ from majority adolescents in terms of sexual conservatism (p < .001). Moreover, majority parents seem to control their son's/daughter's activities to a larger extent than ethnic minority parents (p < .001). Nevertheless, ethnic minority and majority adolescents show similarities with regard to parent–child discrepancies and interethnic friendships. However, differences arise between majority girls and boys: majority girls are on average significantly more religious than majority boys (p < .05). At the same time, ethnic minority girls not only experience significantly more parental control than ethnic minority boys, but majority girls also experience significantly more control than majority boys (p < .001). These findings are in line with the expectations: daughters seem to be raised in a less liberal way than sons. All in all, the descriptive statistics suggest that both ethnic and gender differences should be considered in the investigation of intergroup dating.

The merits of the hypotheses are assessed with stepwise multiple regressions (Table 1). Starting out with a model that controls for socio-demographic variables, the second model contains parent–child discrepancies. Parental control serves as additional third-party influence measure and is introduced in a third step. The fourth model includes attitudes towards sexuality. Religiosity is entered in the fifth model, because previous research has shown that religiosity matters for sexuality (King and Harris 2007) and family relationships (Komter and Knijn 2006: 120). Lastly, the sixth model comprises intergroup friendship to test the contact theory. In the interpretation of the results, we will assess the extent to which these variables can explain ethnic disparities in social distance measured by attitudes towards intergroup dating. The second part of the results focuses on the interactions of gender with religiosity, sexual conservatism and parental control in explaining further gender differences in attitudes towards intergroup dating.

Controlling for gender, education and socio-economic status, the first model shows that most of ethnic minority adolescents show similar levels of social distance to the majority group. Adolescents with a higher educational and socio-economic background are more likely to approve of intergroup dating than those with a lower status (Table 1). With regard to group differences, only 3 out of the 10 ethnic minority groups reject intergroup dating to a significantly larger extent than majority adolescents. More precisely, only adolescents from countries with large Muslim majorities (Turkey, Morocco and the Middle East, with the exception of North Africa) have lower likelihoods of approving intergroup dating than adolescents from the majority group. Indeed, once the religious affiliations (either Muslim, Christian, other or none) are introduced in the model (results not shown), these ethnic differences become insignificant. This finding corresponds to actual intermarriage behaviour (e.g., Schroedter 2012). As already mentioned, family honour and values tied to it are likely to be of particular importance for ethnic minorities from countries with large Muslim majorities. As we will see in the next models, these ethnic differences decrease significantly once parent–child discrepancy, parental control and sexual conservatism are taken into account. Gender differences in attitudes towards intergroup dating are also significant: as expected, females are significantly less likely than males to approve of intergroup dating (see also Osanami Törngren 2011: 149–150).

The second model includes parent–child discrepancy, which – contrary to our expectation – is not significantly associated with attitudes towards intergroup dating and can thus explain neither ethnic nor gender differences in attitudes towards intergroup dating. The sign of the coefficient is in the hypothesized direction, but becomes negative, once parental control is added to the model. Finally, it becomes marginally significant in the last model. Interestingly, additional analyses (not shown) reveal that intergroup friendships are, as hypothesized, associated with higher degrees of parent–child discrepancies even when parental control is included. The change of the parent–child discrepancy coefficient for intergroup dating suggests that the effect of parent–child discrepancies depends on other included variables. To test this, we conducted a mediation test. It turned out that parent–child discrepancy has a significant indirect effect. Parental control mediates 75% of the effect of parent–child discrepancy on the approval of intergroup dating. Parent–child discrepancies and the freedom of choosing a dating partner are only possible if parental control is low. However, the results call for a longitudinal perspective to fill the gap on what comes first: intergroup contact or parent–child discrepancies? The causality could work in both directions. On the one hand, parent–child discrepancies can be a result of unwanted intergroup openness by adolescents; on the other hand, adolescents might want to avoid further conflicts by suppressing intergroup contact and conform at least on the level of attitudes. Talbani and Hasanali (2000) found that girls of South Asian origin accepted conditions in order to avoid the high social costs attached to protest. Or there is a third possibility – these discrepancies are not only about intergroup contact, but other issues that also funnel adolescents in their attitudes towards dating.

By contrast, parental control is significantly and negatively associated with the approval of intergroup dating (Model 3), which confirms our expectation. However, the coefficients of Middle East, Morocco, Turkey remain stable in this model (compared to the first model). Thus even if parental control is significantly associated with attitudes towards intergroup dating, it cannot explain why some ethnic minority adolescents reject intergroup dating to a larger extent. Model 4 shows that sexual conservatism is negatively and significantly associated with the approval of intergroup dating what confirms our hypothesis. Moreover, the introduction of sexual conservatism to the model reduces to some extent ethnic differences in intergroup dating. In other words, the fact that adolescents from Middle Eastern, Moroccan and Turkish origins oppose to a larger extent intergroup dating is partly due to their higher level of sexual conservatism compared to the other ethnic minority adolescents. In the fifth model, the coefficient of religiosity is negative and significant as expected: the higher the degree of religiosity, the lower the likelihood to approve of intergroup dating. Moreover, the coefficient of sexual conservatism decreases but remains significant with the introduction of religiosity in the model. This means that the associations of religiosity and sexual conservatism with intergroup dating somewhat overlap: highly religious adolescents oppose to a larger extent intergroup dating partly because they are highly conservative in terms of sexuality. In addition, with the introduction of religiosity to the model, the coefficients of Middle East, Morocco and Turkey become insignificant. Adolescents of Middle Eastern, Moroccan and Turkish origin oppose thus intergroup dating to a larger extent than the majority group partly because they are more religious than the other ethnic minority adolescents. Lastly, interethnic friendships are included in the sixth model. As expected, adolescents with a large number of interethnic friends are significantly more likely to favour intergroup dating. Moreover, with the introduction of intergroup friendship in the model, the coefficients of Middle East, Morocco and Turkey become slightly significantly negative.

All in all, the few significant ethnic differences in social distance measured by attitudes towards intergroup dating (i.e., among adolescents from Middle Eastern, Moroccan and Turkish origin) are largely due to a higher level of sexual conservatism and religiosity. These statistical results also highlight a striking finding: gender differences remain stable across the six models, even if boys and girls differ in terms of degree of parental control, sexual conservatism and religiosity. In order to gain a deeper insight into this striking finding, the next section investigates the interaction effects of gender with parental control, sexual conservatism and religiosity on attitudes towards intergroup dating for both ethnic minority and majority adolescents.

Based on the expectation that the association of religiosity with social distance is gendered, we calculated interaction effects (not shown) and plotted these gender differences for ethnic minority adolescents and majority adolescents (Figure 1). For ethnic minority adolescents the interaction is significant at the 1% significance level and for the majority adolescents at the 10% significance level. These interactions indicate among both ethnic minority and majority adolescents a significantly stronger negative relationship between religiosity and favouring intergroup dating for girls than for boys. In other words, the likelihood of approval of intergroup dating is lower for religious girls than religious boys.² Moreover, for both ethnic minority and majority boys, the coefficient of religiosity on intergroup dating is insignificant. Thus, religious girls both with and without an immigrant background reject intergroup dating to a significantly larger extent than non-religious girls. By contrast, religious boys with or without an immigrant origin, favour intergroup dating to the same extent as non-religious boys.

In line with the religious demarcation of group boundaries, we observe a similar pattern among ethnic minority adolescents when it comes to premarital intercourse (Figure 2). Religiosity and sexuality are crucial in dividing societies along ethnic and religious lines. Sexual conservatism has again for ethnic minority girls a significantly greater impact on their attitudes towards intergroup dating than for ethnic minority boys (p < .001). There is no comparable pattern observed for adolescents from the majority (and thus interaction is not shown). Moreover and similarly to the association of religiosity with attitudes towards intergroup dating, sexual conservatism is not significantly associated with intergroup dating among ethnic minority boys. Thus, ethnic minority girls are the only group for which sexual conservatism implies a significantly larger disapproval of intergroup dating: the more premarital sex and contraception are regarded as inappropriate, the lower the likelihood of accepting intergroup dating partners.³

Adolescents’ sexuality is a crucial issue for immigrant parents. In order to limit adolescents’ sexual activity parents exercise social control (see, e.g., Hennink et al. 1999). This brings us to one of the main explanatory variables of third-party influence that significantly affect attitudes towards intergroup dating: parental control. Here again, we would expect a gendered association of parental control with attitudes towards intergroup dating, since parents have been shown to be more protective of their daughters than their sons. The interaction between parental control and gender on attitudes towards intergroup dating for majority adolescents is not significant. By contrast, a significant interaction emerged for ethnic minority adolescents (p < .10): ethnic minority girls who report higher levels of parental control approve of intergroup dating to a significantly lesser extent than ethnic minority girls with lower parental control (Figure 3).⁴ By contrast, the association of parental control with attitudes towards intergroup dating is insignificant for ethnic minority boys.

Next to structural opportunities, individual attitudes towards intergroup dating are the main factor driving intergroup unions. Attitudes towards intergroup dating of both ethnic minority and majority adolescents are therefore crucial to investigate in order to better comprehend the challenges facing the meanwhile multicultural West European societies. Intimate intergroup relationships are often discussed as reducing conflicts. To date, this study is the first one to analyse attitudes towards intergroup dating as an indicator of social distance and their determinants among European adolescents from various religious and cultural backgrounds: we could investigate differences in attitudes towards intergroup dating between ethnic majority adolescents and 10 ethnic minority groups from different religious backgrounds. In addition, our gendered approach highlights the limitations of previous studies due to their implicit assumption that ethnic differences in intergroup dating preferences and their determinants are similar for men and women. Following up on previous research that revealed gendered effects (Lucassen and Laarman 2009), we showed that the factors linked to attitudes towards intergroup dating such as religiosity, sexual conservatism or parental control are also highly gendered. As our results showed, a gendered approach is vividly recommended for future studies on social distance indicators in order to avoid misleading conclusions regarding the role of gender in shaping intergroup contacts. Moreover, we shed light on important factors that are significantly associated with attitudes towards intergroup dating, namely religiosity, sexual conservatism, parental control and intergroup friendships.

Our results have important implications for theories on social distance and future research. So far, the willingness to intermarry or date someone of another group has been treated as the strongest indicator of social distance. However, we argue that religious intergroup relationships are an even stronger indicator: we show that group boundaries persist along religious instead of racial lines; adolescents from countries with a Muslim majority (i.e., adolescents of Middle Eastern, Moroccan and Turkish origin) are the only minority groups opposing intergroup dating to a significantly larger extent than the ethnic majority. These significant ethnic differences in attitudes towards intergroup dating are entirely explained by sexual conservatism and religiosity. That is, Muslim ethnic minority adolescents oppose to a larger extent intergroup dating because they are more religious and less liberal in terms of sexuality than the other non-Muslim ethnic minority adolescents. Religiosity and level of sexual liberalization do not only explain why Muslim ethnic minority adolescents oppose to a larger extent intergroup dating but also significantly explain attitudes towards intergroup dating of adolescents from non-Muslim origin (including the ethnic majority). In other words, lower levels of sexual liberalization and higher levels of religiosity reinforce social distance between ethnic minority and majority groups. However, the associations of religiosity and sexual conservatism with social distance are gendered. Indeed, the relationship between religiosity and the approval of intergroup dating is significantly more negative for girls than for boys from both ethnic minority and majority groups. Furthermore, ethnic minority girls who are less liberal in terms of sexuality oppose intergroup dating to a much larger extent than ethnic minority boys with low sexual liberalization. The attributed role of girls as transmitters of cultural norms and values that is to be found in most cultures and religions may explain the gendered association of religiosity and sexual conservatism with attitudes towards intergroup dating. Our findings concerning parental control among ethnic minority adolescents give further support to this explanation. While the association of parental control with attitudes towards intergroup dating is similar for girls and boys from the majority group in multivariate analyses, this association is gendered among ethnic minority adolescents: ethnic minority girls who report high levels of parental control approve of intergroup dating to a significantly lesser extent than ethnic minority girls with lower parental control. By contrast, the association of parental control with attitudes towards intergroup dating is insignificant for ethnic minority boys. The traditional role of girls as transmitter of cultural norms and values could explain why the relationships of attitudes towards intergroup dating with religiosity, sexual conservatism and parental control are much stronger for girls than for boys. But it may also help to prevent girls from teenage pregnancies. Future studies could assess the validity of this explanation and first try to capture the extent to which parents already try to avoid their child's intermingling with members of other ethnic and social groups by choosing different schools and neighbourhoods. Moreover, future research may want to look at the role of perceived symbolic threat (arising out of perceived group differences in values, norms and beliefs) in order to explain why parents may support intragroup relationships and whether they see these relationships as a way to ensure cultural transmission.

While parental influence and religiosity are associated with more opposition towards intergroup dating, our results also show that social distance can be overcome by having interethnic contacts, which support more positive out-group attitudes (Allport 1979). This underlines once more the importance of less segregated environments to impinge on social distance of both ethnic majority and minority groups.

All in all, the three mechanisms underlying intergroup relationships discussed by Kalmijn (1998) also apply to attitudes towards intergroup dating of European adolescents: third-party influence, contact opportunities and belief systems turn out to be highly associated with attitudes towards intergroup dating. However, the limitation of our cross-sectional data prevents us from drawing any conclusion concerning the causal relationship between attitudes towards intergroup dating and religiosity, sexual liberalization, parental control and intergroup friendships. Nevertheless, our study provides solid starting points for future research on contested issues such as parent–child relationships among ethnic majority and minority groups, the importance of religiosity for specific ethnic groups or gender differences in the upbringing of ethnic minority and majority adolescents and their roles in shaping social distance.

We wish to thank the MaD research department from the WZB, in particular Ruud Koopmans and Merlin Schaeffer for their insightful comments on a previous draft.

Sarah Carol is research fellow at the Mannheim Centre for European Social Research (MZES) and guest researcher at the WZB Social Science Center Berlin. She obtained her Dr. Phil. from Humboldt-Universität zu Berlin in 2013. Her work covers the social distance between Muslim migrants and natives in Western Europe, religious rights of Muslims, transnational marriages and the role of immigration policies.

Céline Teney is junior research group leader at the Centre for Social Policy Research at the University of Bremen. Previously she held positions at the WZB Berlin Social Science Center and the Université libre de Bruxelles. Her research interests cover the sociology of immigration, the sociology of the EU, political sociology and quantitative methodology.