Discrimination in the lab: Does information trump appearance?

Using a laboratory experiment, we find evidence consistent with statistical discrimination in a public good and group formation game. In the game, payoff relevant information is presented to subjects, thereby making it costly to discriminate when choosing group members. We find that behavior is correlated with race and people use race to predict behavior. However, race only matters when information on behavior is absent. These results are further confirmed when incentives are in place to encourage behavior that is counter to stereotypes. Not all subjects discriminate in the same way, suggesting unfamiliarity and some in-group, out-group bias. Overall, the evidence points to a lack of information rather than discriminatory preferences.