Dating relationships of the Turkish second generation in Europe

Research on the union formation of the children of immigrants, the so-called second generation, focusses primarily on marriage and cohabitation. Even though dating relationships are often the first romantic relations and lay the foundation for more committed relationships, very little is known about dating among second-generation young adults. This is unfortunate because dating across different origin groups may point to changing or shifting group boundaries, that may only later be reflected in interethnic marriages. In this paper we aim to fill this gap in the literature by exploring the dating behavior of Turkish second-generation young adults in five European countries (the Netherlands, Austria, Switzerland, Germany, France) using the comparative TIES survey data. Our results show that it is common for the Turkish second generation to date a partner with a non-Turkish background. And, moreover, that it is important to distinguish be-tween different types of intra-ethnic partnerships. Although only few parents are reported not to approve of the relationship of their child, a substantial part of the second generation seems reluctant to inform their parents about their dating life. With regard to the role of parents, it seems that their influence on the origin of dating partners is limited. In order to enhance our understanding of the meaning of (intergroup-) dating, its determinants and outcomes, future data may focus more on meeting places and the role of peers in a longitudinal way. We argue that dating is an important indicator for intergroup relations that deserves further attention in increasingly diverse societies.


Introduction
Dating is the first step in romantic partnerships that young adults develop.Whereas dating used to be associated with risk, it is now considered as normative behaviour in western societies, serving important developmental purposes (Collins, 2003;van de Bongardt, Yu, Deković, & Meeus, 2015).First evidence suggests that this is not only the case among majority populations but also among immigrants and their descendants (King & Harris, 2007; Van Kerckem, Van der Bracht, Stevens, & Van de Putte, 2013; van Zantvliet, Kalmijn, & Verbakel, 2015).Yet, studies on the romantic life of the so-called second generation by and large focus on marriage (Carol, Ersanilli, & Wagner, 2014;Hannemann et al., 2018;Huschek, de Valk, & Liefbroer, 2012), and to some extent cohabitation (Hannemann et al., 2014;Hartung, Vandezande, Phalet, & Swyngedouw, 2011;Holland, Wiik, & Dommermuth, 2018).Dating experiences in early adulthood, however, lay the foundation for these more committed relationships.In addition, dating relations may also indicate changing or shifting interethnic boundaries at an early stage.Given the growing shares of the second generation in developed countries it is therefore important to increase our understanding of the patterns and mechanisms behind inter-and intra-ethnic dating relationships.
In absence of suitable data, very few studies have examined partner choice in dating relationships of the second generation in Europe.So far, the limited existing literature on immigrants' and minorities' dating relationships and interethnic dating partners originates from the United States (e.g.Fiebert, Karamol, & Kasdan, 2000;Fujino, 1997;King & Harris, 2007;Levin, Taylor, & Caudle, 2007;Luo, 2008;Manohar, 2008;Nesteruk & Gramescu, 2012;Shenhav, Campos, & Goldberg, 2017).These studies show that the children of immigrants are more likely to date outside their own group than the first generation (Fujino, 1997;Shenhav et al., 2017) and that dating relationships are more often interracial than cohabitations and marriages (Blackwell & Lichter, 2004).Due to the very different migrant origin groups and intergroup relations in the US versus Europe, these findings can however not be directly generalized to the European context (Koopmans, 2013).Studies that do consider interethnic dating in Europe focus on adolescents (Bucx & Seiffge-Krenke, 2010;Carol & Teney, 2015), and show that the probability of dating a majority group partner increases across immigrant generations (van Zantvliet et al., 2015).A German study nevertheless suggests that Turkish second-generation adolescents still mostly date partners of the same (Turkish) origin (Weißmann & Maddox, 2016).Since so little is known about the European context, the aim of this paper is to explore the patterns, origin of the partner and parental influence in dating relationships of the numerically largest second-generation group living across Europe, those of Turkish origin.
Our contribution to the literature on romantic relationships of the second generation is twofold.First, we go beyond the focus on adolescents which so far dominated most (European) studies on dating and study patterns of dating relationships (having a boy-or girlfriend) of second-generation young adults.Although dating starts in adolescence, it is not restricted to this age group.With the postponement of committed relations and increase in age at first marriage, that is prevalent in most developed countries, romantic lives of young adults become more diverse (Rauer, Pettit, Lansford, Bates, & Dodge, 2013).Several studies show that it is not uncommon for people in their early twenties to be in a dating relationship (Kiernan, 2002;Wood, Avellar, & Goesling, 2008).Based on the theory of emerging adulthood (Arnett, 2000) we focus on the Turkish second generation aged 18-25, as this is argued to be an important time for self-exploration that is characterized by great demographic diversity.Second, we are among the first to study changing group boundaries and differentiate between inter-and different types of intra-ethnic dating partners.Since the meaning and definition of intra-ethnic relationships is diffuse (Song, 2009) and can entail having either a Turkish origin partner of the first or second generation, it is essential to differentiate between these types of intra-ethnic partners in addition to interethnic ones.
We use survey data from the cross-national TIES (The integration of the European Second Generation) project from five countries; the Netherlands, Austria, Switzerland, Germany and France.We start by exploring patterns of dating and the characteristics in terms of origin of the dating partners.This will give insight into who is dating across group boundaries and what characterizes these dating relations.Second, we study the extent to which different dating relationships are approved of by parents, as dating relationships can be a source of conflict between immigrant parents and their children.In the final step, we examine how parents and the family context influence partner choice in dating relationships.We do so taking gendered patterns into account.

The meaning of dating
The meaning of dating relationships and it's acceptability largely depend on cultural norms and values regarding romantic relationships (Dion & Dion, 1996).In western European societies, dating in young adulthood is nowadays seen as normative, salient behavior that is based on individual choices and preferences (Collins, Welsh, & Furman, 2009;van de Bongardt et al., 2015).Dating is often perceived as the first step in exploring romantic relationships before committing to a serious relationship (Meier & Allen, 2009;Rauer et al., 2013).In the Turkish culture, however, dating and marriage are still relatively strongly associated and regulated or monitored by the family (Hortaçsu, 2007;Kavas & Thornton, 2013;Medora, Larson, Hortacsu, & Dave, 2002).Dating is increasingly accepted also in Turkey, yet it is mainly seen as a trial period prior to marrying a partner.Although 'western style' dating is becoming more prevalent as well, it is still mostly common among the higher educated living in urban areas (Hortaçsu, 2003).The parents of the Turkish second generation in Europe are generally lower educated and originate from rural areas in Turkey, and as such are overall more traditional when it comes to union-and family formation (Caarls & de Valk, 2018).Research suggests that the influence of parents on the romantic life of their children remains strong within Turkish migrant families in western Europe (Buunk, 2015).However, a double standard prevails when it comes to dating of boys versus girls (Boratav & Çavdar, 2012).For girls, dating is often forbidden because a girl's chastity, which is seen as at risk in dating relationships, is important in order to protect the honor of the family (Es ¸sizoglu, Yasan, Yildirim, Gurgen, & Ozkan, 2011).For boys, on the other hand, pre-marital dating and sexual activity is less criticized and sometimes even viewed as part of true masculinity (Delevi, 2013).

Inter-or intra-ethnic dating; endogamy versus exogamy
Different scholars have addressed the importance of ethnic, cultural and religious boundaries in society and have linked them to how the individual may deal with this as well as to how this may be relevant for understanding intergroup relations (e.g.Alba, 2005;Berry, 1997).Research has also repeatedly shown that partners are often (culturally) similar to each other (Kalmijn, 1998;McPherson, Smith-Lovin, & Cook, 2001).Intra-ethnic (endogamous) relations in which partners have the same ethnic origin are perceived to ensure a level of shared cultural norms and values which facilitates communication and mutual understanding and as such would explain high rates of intra-ethnic marriages (e.g.Dribe & Lundh, 2011;Kalmijn & Van Tubergen, 2006;Van Kerckem et al., 2013).Yet, whereas marriage is a long-term commitment, dating may be of a more casual nature and associated with excitement and self-exploration.As a result, cultural similarity may be of less importance in dating relationships, which makes it easier to cross ethnic boundaries (Blackwell & Lichter, 2004).
Also for the Turkish second generation, the few existing studies suggest that endogamy is more prevalent in marriage than in dating.Huschek et al. (2012) find that 24 and 13 percent of Turkish second generation men versus women are married to a partner with a non-Turkish background.When it comes to dating partners of Turkish second-generation adolescents, a German study suggests much lower endogamy rates, with around 40 percent dating outside the Turkish origin group (Weißmann & Maddox, 2016).Endogamy rates in dating relationships in adolescence may however differ from young adults (our study population), since partner choice has a different meaning with increasing age (Joyner & Kao, 2005).Moreover, previous studies on dating relationships of the second generation have not distinguished between different types of endogamous partners: those born in the parental country of origin and those of second-generation same origin born in the country of residence.As suggested by research on married unions, having a partner born in Turkey may indicate a more traditional choice and a stronger focus on the country of origin (Carol et al., 2014;Huschek et al., 2012).Having a second-generation partner of the same origin is, although still within the own group, perceived to suggest a change in group relations.We therefore argue that it is important to go beyond a simple dichotomy (endogamy versus exogamy) and in addition to exogamous unions also distinguish a gradient of endogamous partners: those born in the European country of residence (second generation), those born in Turkey but living in the European country of residence (first generation) and partners born and living in Turkey (transnational partner).If we expect that endogamy in general is preferred we can expect higher levels of endogamous than exogamous dating relations (H1a).In line with the argumentation on the different meaning of dating for men and women we additionally expect that this holds even stronger for women.In other words, we expect women to more often date an endogamous partner born in Turkey (first generation and transnational) than men (H1b) while we expect men to more often date an exogamous partner from the majority population than women (H1c).

Parental approval
Dating without parental approval is potentially a major source of conflict between immigrant parents and their offspring (Giguère, Lalonde, & Lou, 2010).In order to avoid these negative consequences of dating, the second generation might either refrain from dating or keep their parents uninformed about their relationship (Manohar, 2008;Nesteruk & Gramescu, 2012).Since dating is still not normative acceptable in our study population and the fact that parents traditionally play a key role in partner choice and union formation in the Turkish context, parental approval can be expected to be relevant also after adolescence.In this paper we explore to what extent the parents of the Turkish second generation approve of the dating relationships of their sons and daughters.In a similar vein we also examine whether the second generation hides their dating relationships from their parents.As rules regarding union formation in general, and dating in particular, are stricter for women in this origin group (see before), we expect women to be more secretive of their dating life than men (H2a).In case parents are aware of their child's dating relationship we would, for the same reason, expect higher disapproval rates for their daughters than their sons (H2b).Furthermore, we also study whether parental approval relates to the origin of the partner, as has been suggested by previous research (Shenhav et al., 2017;Shibazaki & Brennan, 1998).Parents in particular may prefer a same origin (ethnic endogamous) partner for their children (Huijnk & Liefbroer, 2012;Van Pottelberge & Lievens, 2018).We therefore expect to find a gradient of parental approval in our descriptive part of the study which runs from a transnational, first generation, second generation to a majority group partner for the Turkish young adult respondents in our study (H3).Given the cross-sectional nature of our data it's impossible to disentangle causality and as such this expectation is not tested in the multivariate part of our study.

Parental socialization
The development of romantic relationships does not occur in a social vacuum, but is influenced by an individual's social surroundings (Cavanagh, 2007;van de Bongardt et al., 2015).Although it can be expected that crossing ethnic boundaries in the choice for a partner is more common in dating relationships, the degree to which this occurs depends, amongst others, on the characteristics of and socialization in the family in which the individual is born and raised.
Parents influence their children via the transmission of cultural norms and values (Kwak, 2003).As such, parental preferences are found to have a substantive influence on union formation preferences of their children both among majority populations (Barber, 2000) as well as in migrant families (Carol, 2014;De Valk & Liefbroer, 2007).A study on interethnic dating in Germany by Weißmann and Maddox (2016) shows this to also be true with regard to the dating behavior of second-generation adolescents.Around 60 percent of these adolescents perceive their parents to have similar homogamy preferences as they themselves, including adolescents from the Turkish second generation.
In more traditionally oriented families (indicated for example by educational level, family size and religious upbringing of children) the preference for a Turkish partner may be stronger than in more liberal, modern families (Munniksma, Flache, Verkuyten, & Veenstra, 2012).Traditional Turkish families are characterized by an emphasis on collectivism and interdependency, and as such ties with the family are deemed important (Hortaçsu, 2003).Moreover, children are raised to comply with their parents and to be family, rather than individualistically orientated (Kagitcibasi & Ataca, 2005).This is reflected in both strong intergenerational support networks and the continued importance of the family in partner choice (Baykara-Krumme, 2016;Kalter & Schroedter, 2010;Milewski & Huschek, 2015).As (more) traditional families attach more importance to the maintenance of communal and familial ties, they are also likely to have a stronger desire to protect ethnic group boundaries which are, either implicitly or explicitly, communicated in the family.Therefore, we expect the Turkish second generation from more traditional families to be less likely to date a non-Turkish partner (H4a).Moreover, when dating endogamous, we expect that those from more traditional families are more likely to date a partner born in Turkey instead of a second-generation partner, as this enables a continuation of the Turkish culture and enhances ties to the parents' country of origin (H4b).
Despite our focus on parental influence, we acknowledge that a range of other individual and contextual factors may influence partner choice in dating relations.We therefore take the peer context during adolescence and educational level of the individual into account. 1Friendships with majority group members and going to a school with large shares of majority group members may promote positive attitudes towards members of the majority population and enhance identification with the mainstream culture as well as create opportunities to meet a potential partner (Leszczensky, 2013).This does not necessarily imply that the second generation no longer identifies with their ethnic in-group, but their endorsement of mainstream culture might make them more likely to support interethnic dating (Uskul, Lalonde, & Konanur, 2011).Moreover, interactions between members of different groups can promote positive intergroup relationships by reducing prejudice and improving attitudes (Pettigrew & Tropp, 2006), which might enhance the likelihood of dating outside ethnic group boundaries.The young adults' level of education is relevant as it has been shown in previous studies that higher educated second-generation young adults in Europe are more likely to have a partner outside the ethnic group (Hartung et al., 2011;Huschek et al., 2012).

Data
We use cross-sectional data from "The Integration of the Second Generation" (TIES) survey, collected in eight countries between 2006 and 2008 through face-to-face interviews.This cross-national comparative survey was designed to study the lives of secondgeneration young adults (age 18-35) living in European cities.In this paper we focus on the Turkish second generation living in the two largest cities of Austria, France, Germany, Switzerland and the Netherlands.These countries were selected based on the inclusion of the Turkish second generation in their sample and the inclusion of questions on dating relationships as these were not part of the study in all TIES survey countries.Respondents were defined as second-generation Turkish if they themselves were born in the survey country but at least one of their parents was born in Turkey.The sampling procedure in the Netherlands was based on population registers.In Austria, Switzerland, Germany and France the limited access to population registers proved insufficient to obtain enough details to identify the second generation.In these four countries onomastic techniques were therefore used (Groenewold & Lessard-Phillips, 2012).An urban sample frame was chosen because most migrants and their descendants in Europe live in urban areas.Although technically our sample is urban by design, we refer to countries rather than cities throughout the rest of the paper.
We select only those Turkish second-generation respondents who were between 18 and 25 and who indicated to be in a dating relationship at the time of the survey or who indicated to have dated before (N = 627).This subsample corresponds to 43 percent of the total Turkish second-generation sample in this age group (the remaining 57 percent either lived already with a (married) partner, had no dating experience or didn't answer; see for more details below).Inclusion in our study was based on the question, "Do you currently have a girlfriend/boyfriend/partner?" which was only asked to those who, up to the survey, did not live with, or marry a partner. 2 Those who reported to not have a girlfriend/boyfriend/partner currently were subsequently asked: "Did you ever have a girlfriend/ boyfriend/partner?".We exclude thirteen individuals for whom we lack information on the origin of the partner.The final analytical sample (N = 614) therefore consists of 37 Turkish second-generation respondents in Austria, 241 in France, 77 in Germany, 140 in Switzerland and 119 in the Netherlands. 3Although all countries are part of western Europe they do differ with regard to the size and position of the Turkish second generation, which may result in cross-country differences in partner choice.Countries with larger number of Turkish immigrants, such as Germany, have a different pool of dating partners to choose from than countries in which the number of immigrants and their descendants are lower (Crul & Vermeulen, 2006).Moreover, living in countries with stronger anti-immigrant attitudes, such as in Austria and France (Czaika & Di Lillo, 2018;Strabac & Listhaug, 2008), may influence dating behavior and partner choice as well.Our sample size does, however, not allow to conduct comprehensive cross-country comparisons.Instead we pool the data in the multivariate analyses while controlling for survey country and limit country specific findings to the descriptive statistics.
1 Age at entering the dating relationship is not included because we do not know the exact age at which this occurred.Including age at the moment of filling out the survey is not informative since this does not necessarily coincide with the age at which the respondents were in the dating relationship.
2 It should be noted that respondents who are currently or were previously in a cohabiting/married union might have dated with a partner before.
Our analyses may thus underestimate the share of second-generation Turks who have dating experience.The data do not allow to include these respondents, but additional robustness checks indicate that they are comparable in the indicators used in our analyses to those in our sample.Only with regard to religion and educational level we found that our sample is potentially slightly more modern.We reflect on this further in the discussion section. 3The relatively small sample size in Austria and Germany can be explained by the fact that the majority of the respondents in both countries (both 63%) answered to have never had a boyfriend or girlfriend.Due to this small sample size the country specific descriptive findings should be interpreted with caution.

Dependent variable
The origin of the dating partner is determined by combining information on his or her country of birth, place of residence and nationality.We distinguish five categories of partners: 1) majority group member: born in the survey country and only has the nationality of the survey country; 2) second-generation Turkish: born in the survey country, but has the Turkish nationality; 3) firstgeneration Turkish: born in Turkey but lives in the survey country and 4) transnational Turkish: born in Turkey and still lived there at the time of survey.All other partners are categorized as having an 'other' origin (5).

Independent variables
Parental approval is measured with a question on whether the respondent perceives their parents to approve of their dating relationship.The answer categories are: "Yes", "No" and "They do not know about the relationship".Respondents who did not answer this question are categorized as missing but included in the analyses and presented as such.This variable is only shown in our univariate and bivariate models, as it is an assessment of an existing relationship.Given the cross-sectional nature of our data we do not include this variable as a predictor in the multivariate models.
Our key explanatory variables refer to the characteristics of the families in which the second-generation adults were raised.Traditional families are generally characterized by a larger family size, rural background, being religious and parental human capital (Kagitcibasi & Ataca, 2005).Whether the young adult was raised in a more traditional family is therefore indicated by several often used proxies.In order to avoid causality issues, we include only time-invariant family characteristics and characteristics that were retrospectively asked about the respondents' childhood.Being raised as a Muslim is measured in two steps.First, respondents indicated whether they were raised according to a religion.Those who answered positively were subsequently asked according to which religion.A dichotomous variable was constructed distinguishing between Muslim (1) and non-Muslim (0) respondents (which included in addition to those who were raised without religion also 5% who were raised in another religion than Islam).Whether or not parents come from a rural area in Turkey is measured with a question on where the mother and father grew up until they were fifteen: 'a village', 'a medium size town' or 'a large city'.The latter two categories were combined, resulting in a dichotomous variable that distinguishes between a rural area (1), if at least one parent grew up in a rural area, or a non-rural area (0).The majority of the parents come from a similar urban/rural region (83 %).A large family size is defined here as those respondents who have more than three siblings (1) compared to three or less (0).As indicators of parental human capital we include a variable that measures whether parents were employed when the respondent was 15 years old (1 = yes; 0 = no).We also include the highest level of education of the parents (0 = Primary school or less, 1 = Lower secondary education, 2 = Higher secondary education, 3 = Post-secondary/Tertiary education).Separate variables are constructed for mothers and fathers as their respective intergenerational transmission to children may differ (Putney & Bengtson, 2002).Furthermore, we control for contacts with majority group peers and individual characteristics.As a measure for peer context we use retrospective questions about friendships and the school composition during adolescence.The number of majority group friends is measured by inquiring the ethnicity of the three best friends during secondary school.We calculated a sum-score of the number of majority group best friends if the ethnicity of at least one of the best friends was known (range 0− 3).The proportion of majority group students in secondary school is included as a categorical variable.Respondents were asked whether their school had "almost no majority group students", "up to 25 %", "around 50 %", "up to 75 %", or "almost only majority group students".We include this variable in three categories (the first two and the last two categories were merged) resulting in a variable indicating: the minority of students are majority group members (0), around half of the students are majority group members (1), the majority of students were majority group members (2).
Educational level of the young adult is measured with a categorical variable indicating the highest completed level of education of the respondent: lower secondary education or less (0), upper secondary education and apprenticeship (1) and tertiary education (2).Gender is coded as 1 (male) and 0 (female).To capture country effects, the survey country is included as a categorical variable in which France is the reference category.Finally, we also control for whether the respondent was dating at the time of the survey (1) or before (0).

Analytical strategy
We first show descriptive results on the origin of the partner and parental approval.Chi-square tests with Bonferroni correction are used to examine whether these aspects of dating differ by country and gender.In the same way we also test whether parental approval varies across different partner origins.Next, multinomial logistic regression models are estimated to examine the influence of the family on the origin of the dating partner.Whereas we make a distinction between transnational partners and first-generation partners in the descriptive analyses, these groups are combined (and referred to as first generation) in our multinomial models; due to small numbers separate analyses were impossible and patterns were overall found to be similar for both groups.Partners who neither belong to the majority population, nor have a Turkish origin are shown in the descriptive results only.Since this group is relatively small and composed of a large variety of backgrounds, meaningful interpretation of the results is difficult with the data at hand.The multinomial regression analyses are run twice in which the first-generation versus the second-generation partner is defined as the reference category.Missing values on the independent variables (which are limited to a maximum 9% of responses) are imputed ten times using multiple imputation.Table 1 shows the descriptive statistics for the full sample and by survey country.

Dating partners: exogamous or endogamous unions?
Table 2 shows the distribution of partner origin, for the full sample as well as split by survey country.Starting from the full sample we find that the largest share of the dating Turkish second generation has a partner from the majority population (46 %), followed by having a second-generation partner (20 %) or a partner that is neither from the majority population, nor has the same ethnic background (15 %).The share of first-generation partners and transnational partners is respectively eight and eleven percent.Contrary to our hypothesis (H1a) exogamy thus overall seems to prevail (61 % in total) according to these descriptive findings.At the same time those who are dating another Turkish partner seem to be equally shared between those with a Turkish second-generation partner and a Turkish partner born in the parents' country of origin (both around 20 %).
There are, however, substantial differences between the survey countries (X 2 (16) = 97.019,p < .001).In Germany, Switzerland and France the majority of the dating partners belong to the majority population, while in the Netherlands and Austria dating a secondgeneration partner is most frequent (see Table 2 for all country differences).
Partner choice not only differs across countries, but also by gender (X 2 (4) = 55.763,p < .001).In line with our expectations (H1b), women more often have a partner born in Turkey (both first-generation and transnational partners; respectively 14 % and 17 %) than men (4 % and 5 %).There is no significant difference between men and women in the share of dating second-generation partners or Note.Each subscript letter denotes a subset of country categories whose row proportions do not differ significantly at the .05level.
G.G. Wachter and H.A.G. de Valk partners with a different migration background.At the same time we find that men (56 %) more often date someone from the majority population than is the case for women (34 %) (confirming H1c).

Parental approval
Table 3 shows whether the second generation perceived their parents to approve of their dating relationship or not (please remind the cross-sectional nature of our data so causal relations can't be drawn based on this).In each survey country, the majority of the parents seem to approve of the relationship their child is in, ranging from an approval rate of 60 percent in Switzerland to 41 in France.No more than 11 percent of the parents do not approve.This seemingly high level of approval is, however, combined with a rather high number of respondents indicating that their parents do not know about the relationship; Around one third of all respondents indicate that their parents are unaware of their dating relationship.We do however find that approval rates between survey countries seem to differ (X 2 (12) = 34.568,p < .001).At the same time there is no difference between men and women in the extent to which parents know about the relationship (contradicting H2a), disapproval rates do differ and are higher among women (12 %) than men (7%) (X 2 (3) = 8.666, p < .05),which confirms H2b.In the next step, we study the perceived parental approval by origin of the partner (Table 4).Contradicting our expectations (H3), we do not find substantial differences in approval rates depending on whether there is an exogamous or endogamous dating partner (X 2 (12) = 10.049,p = .612).The fact that we do not find significant differences could be driven by our sample size.The descriptive findings seem to suggest the highest parental approval in case of transnational dating partners.This could be related to the fact that parents themselves might have been involved in bridging between their child and the dating partner.This has been commonly reported to take place in case of marriage in which there is a change from fully arranged marriage to parents having a facilitating role to connect potential partners.

Multivariate results
Table 5 shows the results of the multinomial regression models.The first two columns refer to the analyses in which having a second-generation dating partner is the reference category, whereas in the third column having a first-generation partner is treated as the reference.
First, we analyze the effect of coming from a more traditional family on partner origin in dating relationships.We study the effects of several indicators separately in our models (all perceived as proxies for the level of traditionalism of the parents and the family of origin).Regarding religion, the findings are partly in line with our expectations as they indicate that second-generation young adults who are raised according to Islam are less likely to date a majority group member compared to a second-generation partner than non-Muslims.No difference between Muslims and non-Muslims is however found in the likelihood of dating a first-generation partner, neither in comparison to a majority group member nor to a second-generation partner.With regard to parental human capital we find mothers' human capital to be of somewhat more importance than fathers' human capital.When the mother is higher educated, the second-generation young adult is more likely to date a majority group member.Yet this finding holds mainly in comparison to dating a first-generation partner.No difference is however found when comparing young adults with the lowest and highest educated mothers, which is likely due to very low number of mothers having obtained post-secondary education.We find the same when it comes to employment of the mother; If the mother was employed when the respondent was 15 years old the person is more likely to have a majority group partner rather than a first-generation partner.Although the findings do not reach significance, the influence of employment and education of the father on partner choice seems to be in the opposite direction to that of the mother.Finally, coming from a large family and having parents with a rural background were found to be unrelated to endogamy or exogamy.Overall these findings do not support our general hypothesis that the second generation who was raised in a more traditional family is more likely to date outside group boundaries (H4a).However the role of the mother does seem to be important and have the expected effect when it comes to her educational level and work history.Suggesting that it is more human capital rather than traditionalism of the family that matters for exogamous partner choice in dating.We do not find support for the hypothesis (H4b) that there is a difference between different types of endogamous unions (a second-compared to a first-generation partner).
With regard to our control variables, we find clear gendered patterns.As expected, men are less likely than women to date a firstgeneration partner and more likely to date a majority group member.In other words, women are more likely to date someone born in Turkey than is the case for men.Thus even when controlling for other characteristics, our hypotheses (1b and 1c) are confirmed.With regard to interethnic friendships we find that friendships with majority group members during secondary school increase the Note.Each subscript letter denotes a subset of country categories whose row proportions do not differ significantly at the .05level.
G.G. Wachter and H.A.G. de Valk probability of dating a majority group member in young adulthood, both in comparison to dating a first-and a second-generation partner.Friendships with majority group members do, however, not influence the probability of dating a second-compared to a first-generation partner.Furthermore, the ethnic composition of secondary schools seems to partially matter in explaining partner choice in young adulthood.Second-generation young adults who attended schools in which more than 75 percent of the students were majority group members seem to be more likely to date a majority group partner compared to a second-generation partner.
In line with our descriptive findings, we find clear differences in the origin of the dating partner between the survey country in which the second generation is born.The second generation from the Netherlands and Austria is less likely to date a majority group member than in France.Whereas the second generation in Germany seems most likely to date a majority group member and least likely to date someone from the second generation.Finally, we find that the probability of dating a second-generation partner or a majority group partner compared to a first-generation partner is smaller when respondents reported about their current rather than a previous dating relationship.Note.Each subscript letter denotes a subset of partner origin categories whose row proportions do not differ significantly at the .05level.

Discussion
The aim of this paper was to explore dating relationships of the young adult Turkish second generation in Europe with regard to partner choice patterns, parental approval and the influence of the family on partner choice.First of all, our findings show that dating outside the Turkish community is common practice among the second generation.A large share of the respondents date a partner from the majority population, and this seems to be especially true for young adult men.In comparison to findings from research on interethnic marriage, having a majority group dating partner thus seems to be more common among the Turkish second generation than having a majority group marriage partner (Huschek et al., 2012).This is in line with previous studies showing lower homogamy rates in relationship types involving less commitment (Blackwell & Lichter, 2004;Kalmijn & Van Tubergen, 2006).Moreover, the pressure of parents on finding a Turkish partner may also be lower in dating relationships.This is supported by the fact that, besides human capital of the mother, we find little effect of the family on partner choice in our multivariate analyses.Parents might allow a period of "fun and self-exploration", although mainly for men, as long as this does not result in an exogamous marriage (Nesteruk & Gramescu, 2012).Unfortunately, we were not able to test whether the influence of the family on partner choice in dating relationships differs across gender, due to the relatively small sample size in this study.Research using larger samples of second-generation men and women is needed in order to better understand whether parental socialization regarding dating is gendered.
Furthermore, our results show that it is essential to include different partner types and go beyond the endo-exogamy dichotomy also when studying dating behavior.A significant share of the second generation who dates within the Turkish community does so with another member of the second generation.Interestingly, the second generation is also frequently involved in transnational dating.On the one hand this may indicate a modern partner choice for those who are highly educated and meet their partner during their studies or work abroad (Sterckx, 2015).On the other hand, partners may be introduced by family in order to maintain ties to the country and region of origin.Our findings point in this direction as transnational dating relationships seem to most often be known to parents and have the highest perceived approval rates.
With regard to parental approval, our findings suggest that the gendered norms about dating are reinforced since women more often perceive their parents to disapprove of their relationships.Interestingly, this does not seem to make them more inclined than men to hide their dating relationships for their parents.As expected, part of the second generation does seem to be hesitant to inform their parents about dating relationships.In order to prevent conflict with parents on appropriate dating behavior and partner choice, the second generation might choose to keep their parents uninformed.It would be interesting to pursue further analyses on how and to what extent parents are aware of the dating relationships of their children among both other migrant origin groups as well as among majority populations.In our sample we can only get a first impression on this issue: according to majority group respondents (in similar ages and who answered the question), 12 % of their parents do not know about the dating relationship.This number is clearly lower than among the Turkish second generation in our sample.Whether this is the result of actual less parental knowledge or whether it reflects the respondents' perception of parental knowledge would be interesting to study in more detail but is impossible to assess with our data.Especially in a digital age where parents more than ever follow their children online, it has been questioned how and in what setting children may develop relationships without "constant" parental monitoring.It was also expected that there would be a gradient in parental approval according to the origin of the dating partner.Yet, our data do not point to large differences in perceived parental approval by partner origin.It could be that parents who approve of dating are overall less traditional and therefore not that concerned about the origin of the child's dating partner.As mentioned, this needs further study and should include dyad information also from the side of the parents.
In addition to the role of parents and the parental home, our findings seem to support the importance of peers for dating among Turkish second-generation young adults.Our rather crude measures on peer relations indicate that having more majority group friends during adolescence and more majority group members in school is related to a higher probability of dating a majority group partner in young adulthood which suggests a socialization effect of contact with majority group members (van Zantvliet et al., 2015).Future studies should include better measures of peer relations over time in order to shed more light on this.Too often surveys do not capture peer relations which are known to become more important in the course of the child's life and particular in the transition to adulthood.
Finally, we found important country differences in dating relations by partner type in which the Turkish second generation in Germany is most likely to date a majority group member.A similar finding was also reported for partner choice in married unions, albeit that the prevalence of having a majority group partner is lower in marriages it was found most common in Germany (Huschek et al., 2012).This could potentially be related to the fact that levels of segregation between the Turkish second generation and the majority group in e.g.neighborhoods is lower than it is in other study countries like France (Lichter, Parisi, & De Valk, 2016).Meeting opportunities may thus be larger and clear divides along origin lines may be less prominent.In particular in a city like Berlin the majority group may be more open to engage in dating a person of the Turkish second generation.Partner origin may however also have different meanings for different types of unions and stages in life.Compared to a study on adolescents of the Turkish second generation in Germany (Weißmann & Maddox, 2016), our findings indicate a much higher occurrence of dating a majority group member.Perhaps young adults of the second generation are more inclined to date majority group members than adolescents as age influences partner choice (Joyner & Kao, 2005), but this difference may possibly also be explained by the different ways of measuring partners' ethnicity.Whereas Weißmann and Maddox (2016) asked respondents what the ethnic background of their partner was, in our study it is a combination of reported birth country, country of residence and nationality determining the ethnicity of the dating partner.It could be that our respondents were not always aware of the exact nationalities or origins of their dating partners.Moreover, the observed differences between countries are potentially the result of different nationality laws that apply to the second generation (Dronkers & Vink, 2012).This may have influenced our measurement of partner origin which is partly based on nationality.Finally, the observed country differences can also relate to different levels of openness of the majority group towards dating a Turkish second-generation young adult across the countries under study.It is, however, relevant to bear in mind that our sample is based on analyses of two cities per country, which implies that one should be cautious with interpreting our findings as county level differences as this could lead to issues of exception fallacy (Lessard-Phillips, Galandini, de Valk, & Fibbi, 2017).As addressed by Lessard-Phillips et al. (2017) it is challenging to decide the best analytical level when studying the second generation.This is also true when examining intergroup relations such as interethnic dating and partner choice, as union formation is a two-sided process that requires positive intergroup attitudes from both the second generation and the majority population.The level of openness towards outgroup members not only differs by country, but also on a lower local level (Czaika & Di Lillo, 2018).Whereas conclusions from studies making comparisons on the national level may be too general, comparing regions or cities across different countries runs the risk of ignoring the influence of the national and institutional context in which these localities are embedded.Future analyses should more fully address the different levels of analyses and the role of the majority population in these different settings in order to understand the dynamics between groups.In this way we would be able to better grasp the influence of country level differences versus regional and local differences on the choice for a partner within or outside own group boundaries.
Two data limitations need to be mentioned here.First, in our data questions about dating are only asked to those who were never in a cohabiting or married union.It could be that those who have not made the transition to a committed relationship yet are less traditional and as such this might result in having a selective, more progressive sample of the Turkish second generation in our analyses.Moreover, it could also be that the respondents who were in a dating relationship at the time of the survey will eventually turn their relationship into a cohabitation or marriage.Therefore these dating relationships may differ from the dating relationships that ended before the survey, as also indicated by the control variable capturing this distinction.Second, we do not know the exact age at which the dating relationships started and occurred in the life of the respondents as some respondents reported about a previous dating partner.Dating at younger ages may however be different from dating in young adulthood and as such involve different (partner) choices (Joyner & Kao, 2005).In order to find out whether patterns observed in this study are generalizable, future research would benefit from including dating relationships at different stages in the life course.Gathering (so far absent) panel rather than cross-sectional data, would enable future studies to examine whether dating functions as a trial period before more serious commitment or as a time to explore different partners.This is important in order to increase our understanding of partner choice throughout the second generations' life course and to better understand the duration and strength of intergroup dating relationships.
In sum, we explored the dating behavior among Turkish second-generation young adults living across Europe.Our work shows that it is relevant to not only focus on marriage and cohabitation but to also include dating relationships if we want to understand how the life courses of the second generation evolve.More research is needed to understand the meaning and implications of different types of dating partners across the life course as it sheds important light on intergroup relations in society.This holds not only for those of Turkish origin but is equally important for those of other immigrant backgrounds living across Europe and elsewhere.

Table 1
Descriptive Statistics of the Full Sample and by Country.

Table 2
Origin of the Dating Partner by Survey Country (%).

Table 3
Perceived Parental Approval of the Dating Relationship by Survey Country (%).

Table 4
Perceived Parental Approval by Origin of the Partner (%).