Identity, homophily and in-group bias

Abstract Many instances of social interaction display either or both of the following well-documented phenomena. People tend to interact with similar others (homophily). They also tend to treat others of shared social identity more favorably (in-group bias). While both phenomena involve some degree of discrimination towards others, a systematic study of their relations and interplay is yet missing. In this paper we report the findings of an experiment designed to address this issue. Participants are exogenously and randomly assigned to one of two groups. Subsequently they play a sequence of eight games with either an in-group or an out-group member. In treatment \EXO\ in- and out-group matches are formed exogenously, while in \ENDO\ participants can choose between in- and out-group matches. We find strong evidence of in-group bias in EXO, and strong evidence of homophily in ENDO. In-group biases, however, either decrease or disappear altogether under endogenous matching. We show that self-selection of homophilous agents into in-group matches cannot explain this fact. We also find that homophily is strongly correlated with risk aversion, and we build on this evidence to derive a rationale for both the existence of homophily and the disappearance of in-group biases under endogenous matching.

Many instances of social interaction display either or both of the following well-documented phenomena. People tend to interact with similar others (homophily). They also tend to treat others of shared social identity more favorably (in-group bias). While both phenomena involve some degree of discrimination towards others, a systematic study of their relations and interplay is yet missing. In this paper we report the findings of an experiment designed to address this issue. Participants are exogenously and randomly assigned to one of two groups. Subsequently they play a sequence of eight games with either an in-group or an out-group member. In treatment EXO in- and out-group matches are formed exogenously, while in ENDO participants can choose between in- and out-group matches. We find strong evidence of in-group bias in EXO, and strong evidence of homophily in ENDO. In-group biases, however, either decrease or disappear altogether under endogenous matching. We show that self-selection of homophilous agents into in-group matches cannot explain this fact. We also find that homophily is strongly correlated with risk aversion, and we build on this evidence to derive a rationale for both the existence of homophily and the disappearance of in-group biases under endogenous matching. (C) 2016 Elsevier B.V. All rights reserved.