Gordon B. Dahl

Kellogg Labor Seminar

Abstract:

Does diversity facilitate or obstruct intergroup cooperation? We examine how polarization and fractionalization in a classroom setting affects natives' in-group bias. We leverage the random assignment of students to classrooms within schools to obtain variation in the ethnic composition of a native's peer group. We combine this with an incentivized, large-scale lab-in-the field experiment which elicits in-group out-group bias using an investment game. In contrast to unadorned contact theory, we find an inverse-U shape in cooperation, with the maximum amount of bias occurring when the native and the immigrant groups are of roughly equal size. The bias is strongest in polarized classrooms where the immigrant group is culturally homogenous and distant to native culture, as measured either by religion or linguistic similarity. We corroborate our lab findings with a survey-based measure of in-group out-group trust. We rule out statistical discrimination as a channel and find evidence for both taste-based discrimination and inaccurate stereotypes as mechanisms.