The use of identity primes to explain behavioral differences between groups: A methodological note

Abstract

Although economists are increasingly using behavioral experiments to study variation in preferences among human groups (e.g., males vs. females), relatively little attention has been paid to the inferential limitations of such experiments. Focusing on group identity primes, this methodological note considers the conditions under which researchers are able to identify causes and explain differences in behavior or preferences between groups. Main concern is that group identity priming (or other approaches to strengthening causal inference in such experiments) are not a panacea and that a persistent omitted variables problem continues to hamper identification of specific causes of group differences in preferences. This note develops a framework for thinking systematically about the treatment effects of priming individuals’ group identity and possible approaches to clean identification of specific group-level traits explaining differences in preferences between groups. Unpacking observed effects of group membership on preferences and identifying specific causes of group differences is paramount from the perspective of policy implications and I clarify the framework's usefulness using concrete examples.