Typicality Can Create, Eliminate, and Reverse the Dilution Effect

Abstract

Nisbett, Zukier, and Lemley presented evidence for a “dilution effect,” in which information nonpredictive of a stimulus person’s behavior “watered down” or diluted the predictive value of categorical information diagnostic of that behavior. Two experiments suggest, however, that nondiagnostic information influences prediction by altering the perceived goodness of fit between the stimulus person and the diagnostic category. The authors conclude that (a) Nisbett et al. found evidence for dilution because their nondiagnostic items decreased the goodness of fit between the stimulus person and the diagnostic category and (b) depending on the typicality of the nondiagnostic items, it is possible to dilute, enhance, or leave unchanged the predictability between category and behavior. The structure of social categories, and the importance of typicality in the stereotyping process, is discussed.