In this study, we examine the influence of discussion frequency, network size, and 3 variables that together entangle the often misunderstood concept of network heterogeneity: discussion frequency with like-minded individuals (''safe'' discussion), discussion frequency with nonlike-minded individuals (''dangerous'' discussion), and diversity of discussion based on the proportion of safe and dangerous discussion. Data were gathered via a postelection random-digit dial telephone survey of residents of a battleground state (N = 600) in November 2004. Three central dependent variables were measured: factual political knowledge, political knowledge structure density, and political participation. The results support the argument that different aspects of political discussion have different implications for different democratic outcomes-and that different conceptualizations and measures of discussion ''heterogeneity'' produce different results.

No abstract

Recent calls for censorship of rap music have demonstrated the need to test the perceptual and especially the behavioral components of Davison's third-person effect hypothesis. The hypothesis states that people perceive media content to have a greater impact on others than on themselves (perceptual component), and that these perceptions lead people to take actions, such as censorship, to prevent the impact (behavioral component). Results of a survey of college students (N = 202) using rap lyrics as the context revealed strong support for both components of the hypothesis. Limited support was found for the social distance corollary of the perceptual component, while the knowledge corollary of the perceptual component was not supported. A new target corollary to the perceptual component was proposed; it predicts that those groups seen as likely targets of a communication will produce larger third-person perceptions than will generalized others.

No abstract

The cognitive mediation model predicts that the impact of a learning gratification for news media use on knowledge of news content is mediated by information processing variables. Specifically, surveillance gratifications seeking should encourage two forms of information processing: news attention and elaboration. These forms of information processing should covary and have a direct and positive impact on learning of news content. The impact of surveillance gratifications seeking on knowledge—expected at the zero-order level—should be approximately zero when these information processing variables are controlled. A secondary analysis of two sample surveys (N = 512 and N = 567) plus analysis of original data specifically designed to test the model (N = 299) provide nearly complete support for hypotheses derived from the model in the context of political learning from the news. Suggestions for expansion of the model are provided.