Strategic actors' in situ impressions of systematically versus unsystematically variable counterparts

The covariation model of attribution holds that when an actor's behavior varies across situations, observers make situational, rather than dispositional, inferences about the actor. We conducted four studies to test the hypothesis that situationally variable behavior can also elicit strong dispositional inferences when the behavior follows a systematic if⋯then⋯ situation-behavior contingency. In all studies, participants, who believed that they were interacting with another person in a 30-round repeated prisoner's dilemma game, made strong dispositional inferences about counterparts. However, the specific dispositions they inferred depended upon the type of variability the counterpart displayed: positive dispositions (e.g., rational) when the counterpart's behavior followed a systematic (if⋯then⋯) pattern that made sense given the context; negative dispositions (e.g., irrational) when the counterpart's behavior was unsystematic, or when the if⋯then⋯ pattern was inappropriate for the context. Taken together, these studies begin to identify when behaviors that vary across situations improve versus harm perceivers' impressions.