ABSTRACT
Considering both non-migrant and migrant couples, this paper studies the effect of cohabiting life partners’ attitudes, resources, and social network compositions on their spouse’s interethnic friendships and acquaintances. Thus, partners are conceptualized as important “third parties” for interethnic relationship formation. Analysing representative German household panel data, I find that partner characteristics play an important role for a person’s interethnic relationships. In particular, a strong positive connection between cohabiting life partners’ interethnic relationships indicates that couples’ social networks merge over time. Partner’s preferences for interethnic contacts mainly exert an indirect influence that is mediated by the partner’s own interethnic relationships and – to some extent – by a person’s preferences.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 Migration is the main source of ethnic diversity in many European countries, including Germany.
2 Kalmijn’s framework was originally developed for intermarriage, but has since been applied to interethnic relationships in general in a large number of studies.
3 One example for this ideal case is the study by Kalmijn (Citation2003), which analyses data from cohabiting partners who reported separately about their social ties.
4 The study of Lubbers et al. (Citation2010) is exemplary for this approach.
5 I use the terms “(life) partners” and “spouses” interchangeably to indicate cohabiting couples.
6 Language barriers may also be lowered by foreign language skills of non-migrants, including the lingua franca English. However, as such language skills are not measured for non-migrants in the employed dataset, I refrain from formulating corresponding expectations.
7 “Second generation migrants” – those born in Germany to parents not born in Germany – were thus excluded from the respondents used for analyses, as only some respondents belonging to this group were filtered to the migrant-specific questions on language skills and discrimination. However, including those doesn’t change the following results substantially, likely because the filtering leads to low usable case numbers.
8 Including interethnic partnerships does not change the following results substantially, likely because they are few in number and partner characteristics vary little over time compared to intraethnic partnerships.
9 As the same measure often enters the analyses twice, once as “respondent variable” and once as “partner variable”, concerns about bias in the analyses may be raised. Therefore, an additional robustness check was performed by randomly assigning the individual partners of each couple to two separate groups. Afterwards, the following analyses were performed separately for both groups ( and ). Thereby, each measure either entered the analyses as “respondent variable” or as “partner variable”. Results proofed to be largely robust.
10 From 2013 onward, the question was slightly changed, referring to “persons who themselves or whose parents are not from Germany”.
11 For migrants, age was not included as control, as survey year combined with length of stay perfectly predict age. Length of stay is used as it is considered to be a more important predictor of migrants’ interethnic relationships.
12 Given the emphasis on moderation effects in the theoretical background, an alternative approach could be to employ Structural Equation Modelling (SEM). A Structural Equation Model was calculated for both Non-migrants and Migrants. The main results were largely comparable to those in the regression models, see the Online Appendix for details.
13 Exact results available on request.