Correspondence Bias

Abstract

When drawing inferences about a person’s personal characteristics from their actions, “correspondence bias” is the tendency to overestimate the influence of those characteristics and underestimate the influence of situational factors, such as incentives the individual faces. We build a simple framework to formalize correspondence bias, and test its predictions in an online experiment. Consistent with correspondence bias, subjects are, on average, willing to pay to receive the dictator-game givings of an individual with whom they are randomly assigned to play a game that encourages cooperation rather than one with whom they play a game that encourages selfish behavior. We show, further, that experiencing both games oneself, as opposed to playing one and observing the other, reduces the bias, and receiving information about how each of the players behaved in both games, eliminates it.

Keywords: Belief updating, Attribution bias, Incentives, Experiments

JEL Classification: C91, D90