Abstract
A 2-stage model of the construction of explanations for differences between groups is presented. Category norms affect which of 2 groups becomes "the effect to be explained," and stereotypes shape attributions about that group. In 3 experiments, 288 participants wrote explanations for differences between gay and straight men. Explanations focused on gay men who were also judged to have more mutable attributes. However, these effects were not correlated. Participants focused explanations on straight men when explicitly instructed to do so (Experiment 1), Explanations focused on both groups equally when the gay men constituted the numerically larger sample, when gay men were more typical of the overarching category (i.e., people with AIDS) than straight men, or when more straight men were described as performing the behavior (Experiment 2). Stereotype-consistent information prompted more essentialist references and fewer reconstructive references to gay men than did stereotype-inconsistent information (Experiment 3). The relevance of this model for theories of norms, stereotypes, and for the conduct of social science is discussed.