Three experiments tested whether changes in social category exemplars affect attitude stability, attitude-behavior consistency, or attitude change. In Experiment l, participants displayed greater attitude stability across 1 month, in several social categories, when they named the same rather than different exemplars. In Experiment 2, participants displayed greater attitude-behavior consistency toward each of 2 social categories when they named the same rather than different exemplars at behavior assessment and at attitude assessment. Participants who named a more likable exemplar behaved more positively, and those who named a less likable exemplar behaved more negatively, than their initial attitudes predicted. In Experiment 3, participants changed their attitudes in the predicted direction after estimating the height of an exemplar who was either more or less likable than the one they had earlier named. The results are interpreted as consistent with recent theory and research on attitude introspection, the matching hypothesis, and models of social judgment.

No abstract

Attitude representation theory (C. G. Lord & M. R. Lepper, 1999) explains both attitude-behavior consistency and attitude change with the same principles. When individuals respond evaluatively to an attitude object, they activate and combine assumptions about the attitude object with perceptions of the immediate situation. The assumptions activated can vary across time, even without additional information. Previous research has shown that individuals activate exemplars when answering attitude questions, attitude reports vary with the valence of the assumptions activated, and activating differently liked exemplars reduces attitude-behavior consistency. The present research completed study of the theoretical implications of exemplar stability by showing that individuals with temporally unstable exemplars, whether spontaneous (Experiment 1) or manipulated (Experiments 2 and 3), are more susceptible to subsequent attitude change than are individuals with stable exemplars.

No abstract

Social category attitudes predict behavior worse toward atypical than typical category members. The present study examined whether this typicality effect involves relatively individuated as opposed to category-based impressions of the behavior target. Cognitive load and positive moods are known to increase reliance on heuristic processing strategies such as using stereotypes. Control condition participants' attitudes toward gay males were less likely to predict willingness to help an atypical than a typical gay transfer student. For participants who rehearsed an eight-digit number or were in a positive mood while they read descriptions of the two gay men, however, no such typicality effect occurred. The results support and extend previous theories of heuristic processing, typicality effects, and the attitude-behavior relationship.