Intramedia Mediation: The Cumulative and Complementary Effects of News Media Use

Abstract

This study focuses on one often overlooked political communication-based media effect, intramedia mediation, and the indirect effects that stem from relationships that exist among various forms of media use. Data from a 2000 national Annenberg election panel survey are used to assess a series of relationships between television and newspaper public affairs use and how these forms of media consumption affect citizens' knowledge of presidential campaign endorsements. The indirect effects that stem from the relationships that exist among these two forms of media use reflect the cumulative and complementary functions of mass communication consumption across time. An analysis of intramedia mediation in coordination with the study of the direct effects of public affairs media use on this study's outcome variable produces substantially larger overall effects for both forms of news use. Thus, the study of intramedia mediation contributes to a better understanding of the full range of media influences on a given outcome variable over the course of a political campaign. Ramifications of these findings are outlined and future lines of research summarized. Keywords: knowledgeindirect effectsmediationnews media usestructural equation modeling Notes 1. This particular data set also includes a set of individuals who were interviewed in the week just prior to the October 17, 2000 debate between Gore and Bush. These individuals were discarded from this study's analyses because the Time 1 and Time 2 media use measures for many in this group would overlap given that the respective exposure items ask about use of a given media outlet in the last week. In addition, the close time frame between the Time 1 and Time 2 measures for these individuals was not deemed appropriate for this study. Finally, only a random selection of all respondents first interviewed in July or August were asked about their knowledge of campaign endorsements in the post-debate phase of the study. 2. CitationMacKinnon et al. (2004) provide the equations for generating the upper confidence limit and lower confidence limit (see Equations 6 and 7, p. 106). If researchers are working under the assumption that the two normally distributed variables that form the mediation-based relationship are uncorrelated, then the tables provided by CitationMeeker et al. (1981) on pages 129-144 can be used to obtain the desired estimates referenced in the CitationMacKinnon et al. (2004) equations. The desired Meeker estimates are based on the confidence interval the researcher wishes to employ (e.g., 90%, 95%) and the two d estimates that are reflective of the respective path estimates involved in the mediation test divided by their respective standard errors (i.e., da = a/sa and dβ = β/sβ). The Meeker Upper and Lower estimates utilized for the mediation tests performed in this study are as follows: television at Time 1 to television at Time 2, upper Meeker, 1.68, lower Meeker, - 1.61; newspaper at Time 1 to newspaper at Time 2, upper Meeker, 1.61, lower Meeker, -1.65; television at Time 1 to newspaper at Time 2, upper Meeker, 1.72, lower Meeker, -1.55; newspaper at Time 1 to television at Time 2, upper Meeker, 1.78, lower Meeker, -1.48. 3. The zero-order correlation matrix provided for this study points to similar relationships between the respective Time 1 news media use measures (r = .34) and the Time 2 news media use measures (r = .35; see Appendix). However, when the four direct Time 1 to Time 2 media use paths and the relationship between the Time 1 media measures are included in the model, the relationship between the Time 2 media use measures becomes non-significant, and the model fit does not noticeably improve with the introduction of this path. As a result, a decision was made to retain the relationship between the Time 1 media use measures, but not the Time 2 media use measures. Additional informationNotes on contributorsR. Lance Holbert R. Lance Holbert is Assistant Professor in the Department of Communication, University of Delaware