Measuring Discrimination at the Local Level

A robust literature in the social sciences uses experimental games to study discrimination. Often, these games are played between strangers, and players are given little information about the population from which the other players are drawn. We introduce a formal framework and a set of novel dictator games to make nuanced inferences about social discrimination based on differences in knowledge of the social context and the receiver. Specifically, we explore the effects of 1) varying the distribution of the receiver population, and 2) moving from a known distribution of receivers to full information about each individual receiver. We use this framework to study discriminatory behavior in rural Sierra Leone, a context characterized by discrimination based on elite status—a factor widely regarded as an important cause of the civil war. We find that the magnitude of discrimination is driven by a dictator’s knowledge of the receiver, not by her knowledge about the distribution of receivers. Our findings provide important implications for both the measurement and interpretation of discrimination.