You want to appear competent? Be mean! You want to appear sociable? Be lazy! Group differentiation and the compensation effect

Using the two fundamental dimensions of social judgment, warmth and competence, we show that, contrary to general models of impression formation, negative information on one dimension has positive consequences on the way a target is judged on the other dimension. Participants learned about two groups which were either congruent on warmth and competence (one group high on both and the other low on both) or they were compensatory (one group high on warmth and low on competence, the other high on competence and low on warmth). Our results show that in the compensatory condition, the groups were rated more extremely than in the congruent condition and that this was especially the case for the dimension on which the groups were high. Results are discussed both in terms of how they run counter to traditional theories of impression formation and what they tell us about the fundamental dimensions of social judgment. 2008 Elsevier Inc. All rights reserved. Introduction ing (Fishbein & Ajzen, 1975) or averaging (Anderson, 1965) of Impressions of individuals and social groups are fundamentally evaluative. In other words, nearly all attributes along which individuals and groups are seen to vary have a consensually shared positive pole and a consensually shared negative pole. Everyone agrees that it is better to be smart than stupid, better to be honest than dishonest, better to be friendly than unfriendly, and better to be industrious rather than lazy. Although we can think of exceptions to this rule (is it better to be thrifty or extravagant?), valence is fundamental to social perception (Osgood, Suci, & Tannenbaum, 1957). Social attributes tend to covary according to their shared valences. Additionally, because of what has been called the ‘‘halo” error in social judgment, our judgments of others may actually reflect an overestimation of the degree to which similarly valenced attributes co-occur. Thorndike (1920) was the first to define this ‘‘halo” error as the tendency to ‘‘think of a person in general as rather good or rather inferior and to color the judgment of the separate qualities by this feeling” (p. 25). Most models of impression formation and information integration in fact rest upon the assumption that global impressions are fundamentally guided by the valences of the attributes that are integrated into an overall global impression (Anderson, 1965; Asch, 1946; Srull & Wyer, 1989). For some authors, valenced judgments of others are presumed to depend upon the addll rights reserved. ique de Louvain, Department