Effects of Group Size upon Proportion Judgments of Likable and Unlikable Members and Group Impressions: Small is Bad, not that Large is Good

Abstract

Hamilton and Gifford's (1976) results that illusory correlation (overestimation of correlation) occured between distinctive group (small group) and distinctive members (minority members) have not been replicated in Japan (Shirai, 1979; Sugimori, 1987). Rather, a conceptual replication of Hamilton and Gifford (1976) had demonstrated illusory correlation between small groups and unlikable members. To resolve these inconsistencies of the results, Exp. 1 and Exp. 2 examined proportion judgments of likable and unlikable members in each of the large and the small groups under six proportions of likable and unlikable members. Exp. 1 and Exp. 2 differs in the number of stimulus groups; Ss in Exp. 1 saw member-descriptions of both large and small groups. In Exp. 2., Ss saw member-descriptions of a large or a small group to exclude contrast effects. In both experiments, the number of unlikable members in small groups were relatively overestimated than the number of likable members; whereas in large groups, the number of likable members were relatively overestimated than the number of unlikable members only when the large groups were presented with small groups (in Exp. 1.). These results suggest that the results of Hamilton and Gifford (1976) can be explained by centralizing tendency. It was also found that there is negativity bias on small groups.