Effects of Arbitrarily Placed Category Boundaries on Similarity Judgments

Abstract

Three experiments examined the effects of arbitrarily placed boundaries on judged similarity between pairs of stimulus persons. Subjects in Experiments 1 and 2 rated the similarity of applicants for a job in which three category labels (Ideal, Acceptable, and Marginal) were superimposed on the composite scores of applicants based on measures relevant to job performance. The exact position of the category boundaries, described to the subjects as arbitrary, was varied across two conditions, allowing a comparison of a given pair of stimulus persons when the members of the pair were in the same category, or in different categories. Whereas the boundary positions were varied across subjects in Experiment 1, they were varied within subjects in Experiment 2. Subjects in both experiments rated pairs as more similar when in the same category than when on opposite sides of a category boundary. In Experiment 3, subjects rated pairs of movie actors on three scales assessing dispositional similarity, based upon the degree of the actor's political liberalism on a percentile scale. The presence of semantic labels, as well as visual boundary markers, was varied, and again confirmed the predicted effects of boundary markers. The presence of a no-categorization control indicated the presence of both intercategory accentuation and intracategory minimization. Moreover, actors separated by objectively large distanceswithina category were seen as more similar than actors objectively smaller distances apart but separated by a category boundary.