The Indirect Effects of Partisanship and Partisan Media on Knowledge About Same-Sex Marriage Policy: Exploring the Knowledge and Belief Gap Hypotheses

Abstract

This study applies the knowledge gap and the belief gap hypotheses to explore knowledge about same-sex marriage rights in the United States. Whereas the knowledge gap predicts that individuals of higher socioeconomic status will be better informed on political issues, the belief gap suggests that that political ideology is a stronger predictor of knowledge and beliefs than education or income levels. Results from this study indicated that higher income levels were a significant predictor of more knowledge about same-sex marriage policy but education was not related to knowledge. As for the belief gap, political party affiliation was not a stronger predictor of knowledge than education level. However, political party affiliation and partisan media (both conservative and liberal) indirectly impacted knowledge through attitudes toward homosexuality, such that Democratic partisanship and liberal media use indirectly led to more knowledge about same-sex marriage rights and conservative media indirectly led to less knowledge about same-sex marriage rights.

Notes

^1. The questions for this study were formulated by first addressing the Human Right’s Campaign’s (HRC) Frequently Asked Questions page on the HRC website. Questions were inspired by each category created by the HRC and comprised the following categories: weddings and religious implications, parent–child relationships (adoption), social security spousal benefits, income taxes, and employment discrimination. Information from HRC was then verified using government resource sites. The authors then created 12 true-or-false questions to cover each category in the HRC’s Frequently Asked Questions category.

Because we created these items ourselves, we evaluated their underlying factor structure. Using SPSS, we conducted two factor analyses of the 12 knowledge items in order to probe the data. First, we used principle components analysis with varimax rotation, which is one of the more popular approaches to factor analyses (Costello & Osborne, 2005; Sass & Schmitt, 2010). Using this approach, we found one major factor that explained about 20% of the variance, had an Eigenvalue of 2.499, and included six of the 12 knowledge items. There were also three smaller factors with fewer items (one to three items) and smaller Eigenvalues (in the 1.1–1.4 range). We also used a more true factor analysis extraction method, one that will be more likely to generalize to other samples and reflect the nature of the population—maximum likelihood with oblique rotation (i.e., direct oblimin; Costello & Osborne, 2005). This analysis resulted in four factors as well: one factor with the same six knowledge items as the PCA results and the other factors with one item each. In moving forward, we decided to use the six knowledge items that emerged from the factor analyses; these are the items listed in the Measures section. For the sake of parsimony and manuscript length, we decided to use only these six items in the study’s analyses. In other words, we did not use the other knowledge items that composed smaller factors with only one to two items.

Additional information

Notes on contributors

Dyann Diercks

Dyann Diercks (M.A., University of Wyoming, 2015) is a communications instructor in the Department of Rhetoric, Languages, and Philosophy at Front Range Community College. Her research interests include political communication, visual storytelling, and media studies.

Kristen D. Landreville

Kristen D. Landreville (Ph.D., The Ohio State University, 2010) is an associate professor in the Department of Communication and Journalism at the University of Wyoming. Her research interests include communication campaigns, political communication, persuasion, and emerging forms of journalism and new media.