Do national identification and interethnic friendships affect one another? A longitudinal test with adolescents of Turkish origin in Germany.

Previous research has repeatedly found a positive association between immigrants' identification with the receiving society and their share of interethnic friends. That is, immigrants with a low level of national identification have relatively little contact with natives, and vice versa. Earlier cross-sectional studies, however, were not able to draw firm causal conclusions about the direction of causality. Theoretically, four different scenarios exist: The causal arrow might run from identification to friends (A), but also from friends to identification (B) or in both directions (C). Finally, the relationship might be spurious, caused by unobserved joint determinants (D). Using three-wave panel data for adolescents of Turkish origin in Germany, I examine these four scenarios. First-difference models with lagged independent variables that account for both time-invariant unobserved heterogeneity and potential reverse causality provide no evidence of reciprocal effects between national identification and interethnic friendships. This finding contradicts common interpretations of cross-sectional studies.