Empirical contributions: Predictions of belief homogeneity and similarity following social categorization

In two experiments subjects were categorized into groups and made predictions about the beliefs of other in‐group or out‐group members. It was hypothesized that subjects would predict ( a ) greater homogeneity among out‐group members than among in‐group members, ( b ) greater similarity between themselves and in‐group members than out‐group members, and ( c ) greater similarity among in‐group members in the presence of an out‐group. In addition, assumptions of homogeneity and similarity were expected ( d ) to be strongest on items relevant to the categorization criterion and ( e ) to be accentuated over time. Results from the first study supported all but the final hypothesis. In a second study subjects were given base‐rate information about the beliefs of either in‐group or out‐group members and were asked to predict the responses of another in‐group or out‐group member. Subjects systematically biased their predictions so as to reduce differences between their own opinions and in‐group members' positions and to increase differences between their opinions and out‐group members' positions. In addition, in‐group members were expected to be more similar to the subjects when an out‐group was present than when no explicit out‐group existed. Finally, as in the first experiment, these biases of prediction were strongest for items relevant to the categorization criterion.