A Constructivist Perspective on Empirical Discrimination Research

Contemporary scholars view race as a constructed social category, not a biological fact. Yet most empirical discrimination research treats race no differently than other individual characteristics typically observed in data. This article considers the implications of adopting a constructivist perspective instead. I develop a simple model where agents use observable characteristics to both interpret membership in racial social categories and make decisions. Discrimination is the result of acting based on perceived social identity. The model highlights the need to measure the racial "first stage"-the social identity contrast between individuals-instead of relying on race as coded in data, and draws a novel distinction between race-based and direct statistical discrimination. I illustrate some implications using data on wages, speech patterns, and skin color and conclude with strategies for future research that build on the constructivist model. (JEL J15, J31, J71, K38)