Distinct Boundaries? Preferences of Immigrants' Descendants Regarding Partnerships with Recent Refugees from Syria and Afghanistan in Germany.

This study examines the preferences of young people from the majority group and minority ethnic groups in Germany regarding partnerships with recent refugees from Syria and Afghanistan. On the basis of a factorial survey experiment implemented in the year 2022 in the 9^th wave of the CILS4EU-DE data, we demonstrate that young people in Germany – both with and without migration background – are much less favourable to partnerships with refugees compared to partnerships with native Germans. Of all respondents, young people who either themselves or whose parents come from Turkey are most opposed to partnerships with refugees. Particularly those who tend to extensively consume Turkish media and those who identify more strongly with their origin country are more negative about partnerships with refugees.

Zusammenfassung: Die Studie untersucht Präferenzen junger Menschen mit und ohne Migrationshintergrund in Deutschland in Bezug auf Partnerschaften mit Geflüchteten aus Syrien und Afghanistan. Die Analysen basieren auf einem faktoriellen Survey-Experiment, das im Jahr 2022 im Rahmen der 9. Welle der deutschen CILS4EU-Panelbefragung durchgeführt wurde. Die Ergebnisse zeigen, dass junge Menschen in Deutschland einer Partnerschaft mit Geflüchteten weniger positiv gegenüberstehen als einer Partnerschaft mit einer Person ohne Migrationshintergrund, unabhängig davon, ob sie selbst einen Migrationshintergrund haben oder nicht. Von allen Befragten lehnen jedoch junge Menschen mit einem türkischen Migrationshintergrund Partnerschaften mit Geflüchteten am stärksten ab. Dieser Effekt ist ausgeprägter für Personen, die türkische Medien konsumieren und für diejenigen, die sich stark mit ihrem Herkunftsland identifizieren.

Keywords: Attitudes towards Immigrants; Partnership Preference; Refugees; Young Adults; Factorial Survey Experiment; Germany; Einstellungen gegenüber Zuwanderern; Partnerschaftspräferenz; Flüchtlinge; junge Erwachsene; faktorielles Befragungsexperiment; Deutschland

With the growing diversity of European countries, scholarly interest in attitudes of the host country population towards newcomers has flourished ([11] & Escandell 2010; [23] & [42] 2021; [40] et al. 2019). An increasing number of studies have focused on the sentiments of citizens with a migration background towards newcomers (C. C. [3] 2019; [7] et al. 2017; Mayer et al. 2023; [43] et al. 2019; [45] & Richards 2019; [48] & Schulte 2022; [49] & Sinnott 2006; [63] et al. 2017). Scholars remain uncertain whether and under what conditions host country minorities share similar attitudes towards newcomers as the majority ([31] & Anderson 2015; [53] et al. 2015). Furthermore, research addressing the attitudes of established immigrants and immigrant offspring towards refugees who arrived during the 2015–16 mass refugee migration is still in its infancy. Our study aims to fill this void while focusing on a particularly telling indicator of (positive) attitudes towards newcomers―individuals' openness towards a refugee as a romantic partner.

Partnerships and marriages between members of the majority and members of different ethnic groups are considered one of the most significant indicators of the permeability of social boundaries within an immigrant-receiving society ([55] 2013) and of growing societal integration across various population groups ([22] 1964). Even though the ethnic composition of Western societies' populations has shifted strongly during recent decades towards higher shares of immigrants and their descendants ([13] 2009), research indicates that intermarriage rates are still rather low ([8] 2018; [12] 2004; [28] et al. 2012; [46] 2010; Muttarak & [24] 2010).

Focusing on societal openness to partnerships with refugees, this study contributes to the existing literature on intergroup attitudes and partnership formation in three important ways. First, we examine individual preferences regarding a romantic partnership with a refugee, focusing explicitly on the variation in partnership preferences among both the majority group and minority ethnic groups of the host country population. Both population groups might have different attitudes towards recent newcomers, particularly if those are forced migrants. Moreover, the population with a migration background is not homogeneous either, with some groups of descendants of immigrants showing more similarities with newcomers than others.

Second, we propose and test a number of theory-driven hypotheses on a range of potential mechanisms behind partnership preferences. These are derived from sociopsychological and sociological theories and refer to the explanations of cultural similarity and common socialization as well as to the reactive distinctiveness hypothesis. Our findings are largely in line with two of the three hypotheses regarding the descendants of Turkish immigrants.

Third, we use a factorial survey experiment implemented in the CILS4EU-DE data to identify the effects of focal variables net of potential confounders. Although previous research has examined to what extent the German society is open to marriages with asylum seekers, these results might be confounded. Asylum seekers come from various countries, may be religious or secular, may have different religious affiliations, and some may have a tertiary education while others have not even completed primary school. All these factors are relevant in attitudes towards one's own future partner or that of relatives. Individuals may have different perceptions of asylum seekers and therefore attribute certain characteristics in their assessments, which would confound the true effect of refugee status or origin, a limitation our study seeks to overcome. Furthermore, much of the earlier research focused on the general population. With the CILS4EU-DE data, which cover a cohort of young adults in their mid to late 20ies, we address a population group for whom partnership formation is a relevant step in the life course and who are likely to think about their future partner.

The paper proceeds with a review of the literature on individual preferences and attitudes towards partnerships with recent immigrants and refugees. We pay special attention to the research findings for Germany in order to place our research in the relevant context. Then, in an unorthodox fashion, we present our descriptive results and compare them to previous findings. Unlike earlier research, our descriptive findings are based on a factorial survey design, which is well suited to analyze partnership preferences and detect the causal effect of the potential partner's origin within an experimental setting. Our key result is that, among all respondents―both those with a migration background and those from the native majority―, the descendants of Turkish immigrants are most unfavorable regarding partnerships with refugees from Syria and Afghanistan. In an attempt to explain this unexpected finding, we dig deeper into the potential mechanisms of intergroup preferences by drawing on a number of sociopsychological and sociological theories and deriving respective hypotheses. After describing the research methodology, we confront our hypotheses with the data and discuss our major findings. We conclude with a summary of our research results and their importance for future research on interethnic relations and ethnic boundary making.

The paper focuses on individual preferences due to their key role in the process of partnership formation ([32] 1998). This process is characterized by (positive) assortative mating, i. e., individuals prefer partners who share similar characteristics, such as the educational level, race, religion, or ethnic background, a phenomenon known as homophily (G. S. [4] 1974; [9] 1985). Cultural similarity is particularly important for a relationship, in terms of both the initial attraction and the stability of a partnership: individuals prefer partners who share similar norms, values, attitudes, beliefs, and worldviews (Kalmijn 1998). Unsurprisingly, most studies on interethnic marriages show that intragroup (or endogamous) partnerships are more likely than intergroup (or exogamous) partnerships between members of two different ethnic groups ([47], 2010; Muttarak & Heath, 2010).

Intergroup partnerships as well as preferences for interethnic partnerships are also not uncommon, as US ([18] et al. 2009; [41] & Lundquist 2013) and European ([5] et al. 2007; [10] & Teney 2015; [29] & Lindholm 2014; [38] & Boos-Nünning 2004; [50] & Mills 2015) research suggests. It is noteworthy that such preferences are marked by stable patterns of ethno-racial hierarchies. In the European context, Potarca and Mills' (2015) analyses of data provided by the dating website eDarling demonstrate that daters—apart from Arab minorities—consistently favor partners of European origin, while individuals of Arab and African origin are chosen least often by daters. Jakobsson and Lindholm (2014) report a similar finding in a Swedish field experiment on an internet dating site: the fictitious profile of an Arab man received significantly fewer clicks than the corresponding Greek or Swedish profile with otherwise identical characteristics.

Several studies that focused on the comparison of attitudes towards interethnic partnerships between majority and minority groups in the Netherlands showed that the Dutch native majority rejected partnerships with migrants or ethnic minorities more strongly than migrants of Turkish, Antillean, and Surinamese origin rejected partnerships with Dutch natives ([26] et al. 2010). The level of rejection of partnerships with migrants from Muslim-majority countries (Turkey, Morocco) was considerably higher compared to migrants with other backgrounds ([27] et al. 2013). Other research, in contrast, reported the level of rejection of interethnic partnerships to be higher among those with Turkish origin than among the native Dutch ([44] et al. 2012).

With regard to preferences regarding partnerships with refugees and asylum seekers, research has so far remained relatively scarce. In her dissertation, Steinbach (2004) analyzed the social distances among Germans and various immigrant groups in Germany, capturing it with the help of the famous [6] (1925) social distance scale. Based on the MARPLAN-BUS 1999 data, she showed the highest level of rejection for immigrants of either Italian, Greek, Turkish, Vietnamese, or African origin as well as Aussiedler "as a member of one's own family" compared to them being a friend, a colleague, or a neighbor. Similarly, Steinbach's (2004) analyses of the ALLBUS 1996 data showed that the level of rejection of immigrants by the native majority is consistently higher when it comes to the immigrants marrying into the family than being a neighbor. The data further revealed that the social distance between Germans and the asylum seekers in this regard is the highest, relatively closely followed by Turkish migrants, whereas it is considerably lower for Italian migrants or Aussiedler. As a study by [51] and Seipel (2023) demonstrates, the social distance to asylum seekers with regard to "marrying into the family" seems to reduce from 1996 to 2016—in fact, it is reduced more strongly for this group than for any other group. Yet, about 20% of Germans still prefer that asylum seekers do not marry into their family, whereas the corresponding figure for Turkish immigrants is 16% and for Italian immigrants about 2.5%.

The hitherto research, which is largely based on the ALLBUS data, has pointed to consistently negative attitudes of the German population towards asylum seekers marrying into the family. Yet we know little about how these attitudes compare to those of the German residents with a migration background. At this point, we break with the academic convention of presenting our own research findings in the empirical section and draw the readers' attention to the first key contribution of our study—a description of the patterns of preferences regarding partnerships with refugees among Germany's majority population and among Germany's various population groups with migration background (see Figure 1). The results are based on a factorial survey experiment implemented in 2022 in the CILS4EU-DE survey. One of the experiment's goals was to establish a causal effect of refugee status and Syrian/Afghan ethnicity on the partnership preferences of young people with and without a migration background in Germany. Besides several other dimensions (such as religious denomination, religiosity, and the level of education), the experiment's vignettes varied in the description of the origin of the potential partner, i. e., born in Germany without a migration background, born in Germany with a migration background, and having arrived a few years before as a refugee from either Syria or Afghanistan (for a detailed description of the data, the vignette experiment, and the method, see below). Due to the similarity of attitudes towards Syrian and Afghan refugees and for ease of interpretation, we created just one group, which we labelled "refugee". Figure 1 maps the patterns of partnership preferences of the various groups, which are listed in the legend to the figure, regarding refugees and Germans with migration background compared to Germans without migration background.

Graph: Figure 1: Partnership preferences of native-born Germans and migrant population Notes: N(Native Germans) = 7,760, N(Turkish) = 1,400, N(MENA) = 552, N(European) = 3,800, N(other) = 764; conditional effect plots with 95 percent confidence intervals. The estimates are reported in Table A5. Source: CILS4EU-DE waves 1, 2, 3, 6, and pre-publication version wave 9, own calculations.

The results demonstrate that all respondents—both those with and those without a migration background—tend to reject refugees as partners compared to Germans without migration background. Yet, one group of respondents particularly stands out, and these are young people of Turkish origin. They reject refugees as partners considerably more strongly than any other group.

The rejection of Syrian and Afghan refugees as potential (marriage) partners by all analyzed population groups in this study stands in sharp contrast to no or rather small differences in their preferences for German partners with a migration background compared to German partners without a migration background. While we found no significant differences in the acceptance of those born in Germany (with and without a migration background) as partners among native Germans and respondents stemming from Middle Eastern and Northern African (MENA) countries as well as Turkey, we detected slightly higher preferences for a German partner with a migration background among respondents of European and other origins.

Despite somewhat different definitions of respondent groups and a methodologically different approach, these findings are in line with previous research on the patterns of rejection of asylum seekers in Germany. This rejection contrasts with largely similar or even more favorable patterns of partnership preferences regarding individuals with a migration background who are born in Germany. Of all respondents, young people who either themselves or whose parents come from Turkey are most opposed to partnerships with refugees. In the following sections, we will try to advance our understanding of why this population group so firmly opposes the idea of having Syrian and Afghan refugees as potential romantic partners.

How can we explain the unexpected finding that descendants of Turkish immigrants, who in their majority are themselves of Muslim faith and stem from families who share more traditional perceptions of marriage and family, are unfavorable towards newly arrived refugees who on average are of the same faith and share largely similar attitudes ([20] & Heyne 2017)? We base our theoretical ideas and hypotheses on social identity theory (SIT) ([58] & Turner 1979, 1986), which emphasizes the importance of intergroup differentiation for the establishment of a distinct group identity ([25] et al. 2014). What emerges is a sociopsychological phenomenon of ingroup favoritism and outgroup bias ([14] 2004), which can provide an explanation for the patterns of inter- vs. intragroup partnership preferences. Threats to intergroup distinctiveness activate an ingroup bias, whereby the categorization into in- or outgroup depends on the identity components that are particularly meaningful for the respective individuals, such as race, ethnicity, religion, or national identification. Whether descendants of Turkish immigrants consider Syrian and Afghan refugees as in- or outgroup depends on the salience of a particular identity component.

On the one hand, Turkish minorities might define their ingroup based on the common cultural heritage and values shared by their families of origin. The underlying mechanism is that of cultural similarity. Accordingly, descendants of Turkish immigrants who share more conservative values and are generally more oriented towards traditions of their heritage country might feel culturally closer to recent refugees from Middle Eastern countries than to the German mainstream. Such traditionalism is manifested in attitudes towards core life areas, including premarital intimate relationships or homosexuality, towards which immigrants from Muslim countries have consistently been shown to hold more negative attitudes ([19] et al. 2014; [39] & Weißmann 2020; [52] & Spierings 2022). Assuming that Turkish minorities define their ingroup primarily based on common cultural heritage, we expect that more traditional individuals within the Turkish community will prefer refugees over Germans compared to their less traditional counterparts (H1).

On the other hand, descendants of Turkish immigrants who are born and raised in Germany might define their identity based on their socialization experiences within German society, thus extending their ingroup definition to include the majority native-born Germans. Common socialization within German society implies a considerable extent of assimilation into the mainstream culture, according to which ethnic minorities adopt norms and patterns of behavior that are widespread among the majority population (Gordon 1964; Heath et al. 2013). Assimilated Turkish minorities are thus more likely to share the social norms and values prevalent in the native population (van der Zwan et al. 2017), leading to the alignment of their intergroup perceptions with those of the native-born majority (Meeusen et al. 2019; [54] et al. 2018; [65] & Martinovic 2012). Assuming that descendants of Turkish immigrants define their ingroups based on common socialization in Germany, we expect that those with stronger German identification will exhibit partnership preference patterns regarding refugees that are comparable to those of the majority native-born Germans (H2).

Finally, as SIT posits, intergroup differentiation occurs as a result of a group's desire to establish a distinct identity that differentiates their ingroup members from outgroups. Threats to intergroup distinctiveness, which might occur when outgroup members are easily mistaken with ingroup members due to their phenotypical, linguistic, or cultural similarity, exacerbate the group's attempts to maintain their unique distinctiveness (Hindriks et al. 2014). The reactive distinctiveness hypothesis, which is derived from SIT, addresses this issue by postulating that the more similar an outgroup is perceived to the group in question and the more salient ethnic identity is in the latter group, the stronger the distinctiveness threat and the stronger the tendency for intergroup differentiation ([30] et al. 2004). Thus, the intergroup differentiation is viewed as the reaction to distinctiveness of a threatened group and as Jetten et al. (2004) have shown, the strength of this reaction is moderated by the individuals' identification with the ingroup. In our case, this would imply stronger rejection of a refugee group by the descendants of Turkish immigrants with a stronger ethnic identity as opposed to those whose ethnic identity is less salient (H3).

For our analyses, we rely on the data from the German part of the Children of Immigrants Longitudinal Survey in Four European Countries (CILS4EU-DE, [33] et al. 2016a, 2016b, 2017, 2021, Forthcoming). A wave 1 sample was drawn using a three-stage sampling design, which included general schools enrolling nine-graders as the first stage, classes within these schools as the second stage, and all adolescents in these classes as the third stage. An overrepresentation of adolescents with migration background was achieved by oversampling schools with a high proportion of immigrant pupils. In wave 6, which took place in 2016, a refreshment sample of the same birth cohorts as the initial sample was drawn. Sampling was based on name lists from randomly selected municipalities. Individuals on the lists were classified into a possible migration background using name-based procedures.

Our analyses draw on a factorial survey experiment implemented within the 9th wave of the CILS4EU-DE data, which was collected in 2022 among the ca. 26–28-year-old individuals.

Each respondent received four vignettes that contained a description of a potential partner. The vignette persons' (VP) characteristics varied on several dimensions, namely their origin (levels: 1 "born in Germany to German-born parents, labeled as without migration background", 2 "born in Germany to migrant parents, labeled as with migration background", 3 "Syrian refugee", 4 "Afghan refugee"), religious denomination (levels: 1 "Christian", 2 "Muslim"), religiosity (levels: 1 "not religious", 2 "strongly religious"), and educational attainment (levels: 1 "no tertiary degree", 2 "tertiary degree") (for further details and an example of a vignette, see Figure A1 and Table A1 in the appendix). We used a D-efficient design to select 48 vignettes with different level combinations and allocated them to twelve experimental groups, to which the respondents were randomly assigned. The respondents then indicated whether they would like to engage in a particular type of romantic partnership (marriage/committed relationship/dating) with the VP on a scale from 1 "not at all" to 11 "completely". For the following analyses, we focus on the reaction of respondents towards partnerships depending on the origin of the vignette person. While we focus on the VP's origin, we per design control for all other dimensions included into the vignettes and thus get a causal effect of refugee status and ethnic origin net of VP's religious affiliation, religiosity, or education. This is an important characteristic of the factorial design, which would make our results not entirely comparable to the research findings relying on social distance scales. In the latter case, the resulting point estimates pertaining to ethnic origin or group legal status (e. g., Aussiedler) might be confounded by characteristics not belonging to the group definition, but are correlated with it, such as education or religious affiliation.

Among the respondents, we differentiate between native-born Germans and immigrant descendants using information on their own country of birth as well as that of their parents and grandparents, which has been collected in waves 1-3 and wave 6 respectively (for detailed information on the measurement, see [17] et al. 2014). The descendants of immigrants are classified into distinct groups as follows: those whose ancestors originated in Turkey, MENA countries, Europe, and other countries (for the exact classification of countries into the analyzed categories, see Table A2 in the appendix). A more detailed classification is hardly possible due to the sample size.

The hypothesis testing is conducted for the group of descendants of Turkish immigrants using additional information collected in wave 9. To test the cultural similarity hypothesis (H1), we carry out a within-group comparison of less and more traditional individuals, postulating that more traditional ones should be more favorable towards partnerships with refugees (from traditional societies). We used two constructs to test this hypothesis. First, we measure norms of traditional couple relations by building an additive index of the disagreement with the following items: living together as a couple without being married, divorce, abortion, homosexuality. Second, we use information on whether respondents agree with the statement that immigrants should keep their customs and traditions in host societies.

To test the common socialization hypothesis (H2), we compare immigrants with high and low German identification to one another (and implicitly also to the German majority youth). The aim is to establish whether the partnership preferences towards refugees of immigrants with strong German identification resemble those of the majority German-born. Such an implicit comparison is best done when we operationalize the degree of minorities' German socialization through a question about the extent to which descendants of Turkish immigrants identify as Germans. Secondly, we rely on the question of how frequently respondents consume exclusively German media versus media from the country of origin. Whereas the former variable aims at measuring a subjective aspect of the German identification, the latter variable captures a more behavioral part of the individual identification.

Finally, to test the reactive distinctiveness hypothesis (H3), we use information on whether and how strongly respondents feel they belong to their own ethnic group. This hypothesis pertains to an intragroup comparison of the partnership preferences for immigrants with a strong and weak ethnic group belonging. For a description of the operationalization of the variables used to test Hypotheses 1-3, see Table A3 in the appendix.

To rule out that our results are biased by the survey mode (CAWI or PAPI) or by the order of the presented vignettes, we control for these characteristics in all models. Additionally, we control for age, gender, and partnership status of the respondents. Excluding respondents with missing information on these variables and those who have not answered all four vignettes, our main analyses rely on 14,276 vignette evaluations by 3,569 respondents, of which 1,400 vignettes were evaluated by 350 respondents with Turkish background (see Table A4 in the appendix for a detailed description of the sample composition).

To inquire into the rather unexpected finding that descendants of Turkish immigrants reject partnerships with Syrian and Afghan refugees much stronger than any other group, we leave other groups out of the analyses. In the following, we probe several potential explanations for the above mentioned finding, in particular, the mechanisms of cultural similarity and respondents' common socialization in the country of residence as well as the reactive distinctiveness hypothesis.

To test the cultural similarity explanation, we explore partnership preferences among descendants of Turkish immigrants who (1) share norms of traditional couple relations as opposed to more liberal ones and (2) endorse stronger immigrants' adherence to traditions of the heritage country as opposed to those who do not. Analyses for both respective items demonstrate that overall, more traditional descendants of Turkish immigrants more strongly reject partnerships with refugees over partnerships with the majority native-born Germans compared to less traditional ones (see Figure 2). Even though the differences in the patterns are not statistically significant, they are consistent across both indicators. At any rate, the findings run against the cultural similarity hypothesis, as respondents for whom we expect larger cultural similarity with refugees tend to more strongly reject this group.

Graph: Figure 2: Preferences regarding partnerships with refugees as opposed to the German majority members among descendants of Turkish immigrants, testing cultural similarity hypothesis (H1)Notes: N(Model a) = 1,384, N(Model b) 1,352; conditional effect plots with 95 percent confidence intervals. The estimates are reported in Table A6.Source: CILS4EU-DE waves 1, 2, 3, 6, and pre-publication version wave 9, own calculations.

In the next step, we test the common socialization hypothesis, according to which partnership preferences among descendants of Turkish immigrants with stronger compared to weaker German identification should differ from one another. Furthermore, respondents with stronger German identification should express partnership preferences similar to those of the majority Germans. Analyses presented in Figure 3 are largely in accordance with the hypothesis: the point estimate of the preference pattern for those with strong German identification is smaller than that of the respondents with weak German identification, and it is very close to the pattern observed among the German majority (see Figure 1). The confidence interval is, however, very large due to a small number of Turkish-heritage individuals with strong German identification and therefore the differences between the two groups of respondents are not statistically significant. Yet, despite the lack of statistically significant differences, the group of Turkish descendants with stronger German identification tends to disfavor partnerships with refugees much less.

Another indicator of minorities' socialization pertains to the pattern of media consumption. Respondents who do not consume media of their heritage country are assumed to be highly assimilated into the residence country. Others, in contrast, might be oriented towards their heritage countries, and by closely following the media discourse prevalent in those countries, they might adopt attitudes towards refugees that align with the dominant perspectives in these countries. With the ongoing refugee inflow to Turkey, the attitudes towards refugees from Syria have been oscillating between mixed and negative ([1] et al. 2018; [15] et al. 2021; [21] et al. 2018; [56] et al. 2016; [60] et al. 2017; [61] et al. 2022; [62] & Aydin Çakir 2020). Hence, negative attitudes towards refugees among descendants of Turkish immigrants might be related to the non-positive climate of refugee acceptance in Turkey.

To test this assumption, we juxtapose descendants of Turkish immigrants who consume solely German media with those who exclusively or regularly consume Turkish media. Results indicate that descendants of Turkish immigrants who do not consume Turkish media are indifferent in their partner preferences once a refugee and a German are compared. In contrast, those who exclusively or frequently consume the Turkish media more strongly reject a refugee compared to a German without migration background as a potential partner. Taken together, our findings speak in favor of the common socialization hypothesis. A greater extent of German socialization is associated with a pattern of partnership preferences, which is more comparable to that of the majority native-born Germans. In contrast, Turkish descendants who are more strongly oriented towards their heritage country reject partnerships with refugees on a much stronger level.

In the final step, we test the reactive distinctiveness hypothesis, which maintains that minorities, who are perceived by others as similar to a different minority group―in our empirical case the group of recent refugees―tend to experience distinctiveness threat and therefore a more pronounced tendency towards intergroup differentiation. Empirically, this should result in a stronger rejection of a refugee group by the descendants of Turkish immigrants who maintain a strong ethnic identity. Results presented in Figure 4 show consistent support for this hypothesis. Respondents originating in the first or second generation in Turkey, who express a strong belonging to their heritage country ethnic group, are more likely to reject partnerships with a supposedly similar group of refugees from Syria and Afghanistan. In contrast, those with less salient ethnic identity tend to reject partnerships with refugees to a lesser extent and at a level comparable to other groups in Germany.

Graph: Figure 3: Preferences regarding partnerships with refugees as opposed to the German majority members among descendants of Turkish immigrants, testing common socialization hypothesis (H2)Notes: N(Model a) = 1,372, N(Model b) = 1,396; conditional effect plots with 95 percent confidence intervals. The estimates are reported in Table A7.Source: CILS4EU-DE waves 1, 2, 3, 6, and pre-publication version wave 9, own calculations.

Graph: Figure 4: Preferences regarding partnerships with refugees as opposed to the German majority members among descendants of Turkish immigrants, testing reactive distinctiveness hypothesis (H3) Notes: N = 1,336; conditional effect plots with 95 percent confidence intervals. The estimates are reported in Table A8.Source: CILS4EU-DE waves 1, 2, 3, 6, and pre-publication version wave 9, own calculations.

Contributing to existing research on interethnic partnership choice, our study sheds light on differences in partnership preferences regarding refugees on part of young people from both majority and minority ethnic groups in Germany. Whereas so far only few studies in Germany were able to explicitly focus on the minority groups' attitudes towards forced migrants (Mayer et al. 2023), our study captured partnership preferences of young German residents, who are in the life phase when partnership formation is on top of the agenda.

In pursuit of the question of how young German people with and without migration background react to the presence of refugees from Middle Eastern countries, and particularly, whether they are open to partnerships with newcomers, we carried out a factorial survey experiment in the framework of wave 9 of the representative CILS4EU-DE survey. The resulting data allowed us to tease out causal effects of refugee status and ethnic origin without potential confounding from other factors commonly associated with the two, such as low socioeconomic status or high religiosity. Furthermore, our study proposed and tested a number of potential explanations for the detected pattern of rejection of partnerships with refugees among young people with Turkish roots highlighting the importance of distinctive identity among this population group. In doing so, we built on the sociopsychological and sociological theories and tested the cultural similarity, common socialization, and reactive distinctiveness hypotheses, of which the latter two were largely supported by the data.

Our results suggest that overall, young German adults―regardless of whether they are the majority native-born or with migration background―tend to disfavor partnerships with refugees of Syrian and Afghan origin as compared to partnerships with other Germans. Avoidance of partnerships with refugees is rather consistent and is at a comparable level across respondents with and without migration background, with a single exception: young people who either themselves or whose parents originated in Turkey. These respondents tend to reject partnerships with refugees particularly strongly.

To explain this surprising finding, we referred to different theoretical approaches, including the cultural similarity and common socialization explanations as well as the reactive distinctiveness hypothesis. In contrast to the cultural similarity hypothesis (Hypothesis 1), we established that more traditional descendants of Turkish immigrants rejected partnerships with refugees more strongly than those who were less traditional, albeit the differences between more and less traditional groups were not statistically significant. The common socialization hypothesis (Hypothesis 2) in conjunction with the reactive distinctiveness hypothesis (Hypothesis 3) delivered the most consistent explanations. According to the common socialization hypothesis, descendants of Turkish immigrants who strongly identify themselves with Germany, express a lower rejection of partnerships with refugees and accordingly, their partnership preferences are more similar to those of the majority native-born Germans; yet, the differences between those with and without strong German identification are not statistically significant. Those who tend to extensively consume Turkish media and those who identify more strongly as being Turkish or Kurdish tend to reject partnerships with refugees more fiercely. Presumably, minorities originating in Turkey, who strongly care about their ethnic identity, experience a stronger distinctiveness threat and are under pressure to clearly distinguish themselves from Middle Eastern refugees. At the same time, descendants of Turkish immigrants who consume Turkish media are not confronted with a positive image of a Syrian refugee and therefore more likely reject partnerships with refugees.

Overall, we find that many descendants of Turkish immigrants draw distinct boundaries between themselves and refugees from Syria and Afghanistan, a phenomenon known as boundary-making ([2] 2005). These boundaries to Syrian and Afghan refugees are particularly salient for those who have a strong Turkish or Kurdish identity and are strongly oriented towards the Turkish media and potentially the Turkish media discourse on refugees.

However, the results of our study should be interpreted in the context of the different limitations that require further research. First, focusing on the cultural and identificative aspects of the groups' boundary-making, the current study sidesteps yet another dimension―that of the socioeconomic status―which is inherent to the partnership formation process. As demonstrated by Dasgupta (2004), if a group seeks enhancement of existing social structure and sociocultural hierarchies, a socioeconomically higher-rated outgroup might be preferred over a more natural ingroup. With the current empirical design, we are not able to address this issue, leaving this task for future research. Second, since we focused on refugees from either Syria or Afghanistan, we were not able to conclusively attribute the detected effect to either refugee status or ethnic origin. It remains the task for the future research to figure out whether descendants of Turkish immigrants shy away from refugees from these countries due to their legal status, their ethnic origin, or rather the combination of the two. Third, our data cover a cohort of young adults in their mid-late 20-ies, which makes our results not generalizable to the entire German population. Yet, we are cautiously confident that the findings are generalizable to the cohort of young adults, who are currently in an important life phase of partnership formation. Therefore, studying partnership preferences among a cohort of young adults in their partnership formation phase should not be necessarily considered a study limitation, but potentially an advantage, as we ask respondents not just some hypothetical questions, but the ones, which are particularly relevant to their current life phase. Finally, due to sample sizes no meaningful analyses of attitudes towards refugees and young people with migration background among members of smaller ethnic minority groups were possible. Larger sample sizes would be needed to better understand their partnership preferences.

Overall, our findings tend to suggest that although the German society has undergone a long way towards greater acceptance of immigrants, the patterns of restrained attitudes towards asylum seekers in 1996, 2006, and 2016 as shown in earlier research (Rippl & Seipel 2023; [57] 2004) have persisted into the year 2022 as referring to refugees from Syria and Afghanistan in this study. It seems like the German population―and here we mean both majority native-born and those with migration background―has to make further efforts towards a stronger acceptance of all newcomers, including those who forcefully flee wars, persecution, and other forms of violence occurring in non-European countries.

This work was funded by the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation program (Grant agreement No. 864683). We thank Markus Weißmann and Hannah Soiné for helpful comments and support with the preparation of the CILS4EU data. We thank participants of the Inaugural Lecture in the Research Master in Sociology and Demography at University of Pompeu Fabra, Barcelona, where an earlier version of the paper was presented in October 2022, for their helpful comments.

The code for our analyses is available at https://osf.io/g7aqh/