The complexity of relations between dimensions of social perception: Decomposing bivariate associations with crossed random factors

Abstract

Theories of social perception argue that there are two underlying dimensions of social judgment, variously labeled competence/agency and warmth/communality. How these relate to each other has been the focus of extensive empirical work with research showing both a ‘halo” relation (targets rated more positively on one dimension are rated more positively on the other) and a “compensatory” relation (targets rated more positively on one dimension are rated more negatively on the other). We argue that these divergent findings result from different comparative contexts under which participants judge social stimuli and on the underlying factors that contribute to variance in the resulting judgments. In two studies, we vary the comparative context under which perceivers made judgments and we decompose the variance in such judgments (and their covariance) into components due to the random factors of participants, targets, and their interaction. Halo relations emerge for participant means, regardless of comparative context. On the other hand, the covariance between target means changes signs under different comparative contexts, as does the interaction covariance.