An experimental analysis of the contrast effect and its implications for intergroup communication and the indirect assessment of attitude.

Experiments I and II were designed to differentiate between alternative interpretations of the contrast effect. Experiment I examines and rejects the hypothesis that the effect is specific to rating scales, since the effect also occurs when subjects are asked to make CDmparative judgments. Experiment II leads to the rejection of the hypothesis that the effect is purely semantic (i.e., specific to situations in which verbal labels are applied to attitudes); when subjects are asked to write items that would be endorsed by people with whom they agree or disagree, they write much more extreme items for those with whom they disagree than for those with whom they agree. In fact, the items subjects write for people with who::n they disagree are not endorsed by those people, the reason most often given being that the items are too extreme. Experiments III, IV, and V demonstrate that the differences are so clear-cut that such attitude statement writing could be used as an indirect measure of attitude. Judges were able to categorize writers' attitudes rather accurately, but accuracy was no greater when the contrast effect was explained to them than when it was not. The implications Df these findings for intergroup conflict are discussed, particularly the implication of a positive feedback effect that will lead to exacerbation of group differences.