This project seeks to clarify the mechanisms through which the media contribute to audiences’ understandings of social groups. It encompasses two inter-linked studies. One study identifies two dimensions through which audiences evaluate the realism of media characters. It then investigates how one of these dimensions, character representativeness, is associated with the audience’s level of familiarity with the society portrayed in the media text and with their sense of the variability of the represented society. Participants from two different societies, the US and Greater China, evaluated the characters in two film segments, one from a culture with which they were familiar and one from a society with which they were unfamiliar. They then evaluated the homogeneity of the films’ societies. I assessed the participants’ perceptions of the characters’ representativeness through two measures. One measure supported the initial hypotheses of the study. Characters from socially distant societies were seen as significantly more representative than those from socially near ones. The other measure did not provide any support for this hypothesis. There was no consistent evidence that the perceived variability of a film’s society moderates perceptions of the representativeness of the film’s characters. iv The second study investigates whether viewers’ perceptions of the representativeness of a text’s characters shape the strength of the text’s effect on viewers’ perceptions. It also sought to determine whether the activation of particular category structures or viewer attributions of character behavior influenced effects. Volunteers saw a film clip and then completed a questionnaire about the representativeness of the characters and their perceptions of the source society. Before seeing the film clip, half the participants were primed with publication materials designed to activate the category structure of society membership. Neither the representativeness of the characters nor the variability of the film society is associated with the application of the characters’ attributes to the viewers’ perceptions. There are no consistent differences between the priming and control groups on any of the outcome measures. There is no consistent evidence that audience members’ attributions shape the media representations’ effect. Possible reasons for the studies’ failure as well as implications for future research are discussed.