Media Literacy Messages and Hostile Media Perceptions: Processing of Nonpartisan Versus Partisan Political Information

Abstract

Partisans are poor judges of news content, rating neutral content as biased against their views (the hostile media perception) and forgiving biased content when it favors their side. This study tests whether a short news media literacy public service announcement (PSA) appearing before political programming can influence credibility and hostility ratings of the program and program host. Our findings suggest that a media literacy PSA can be effective, but its impact depends on the position of the news program and on the political ideology of the viewers. In this case, the media literacy PSA only influenced conservatives’ evaluations of the political program, improving perceptions of a neutral or congruent (conservative) host while further depressing ratings of an incongruent (liberal) host. Liberals’ evaluations of the program were unaffected by the PSA. Implications for media literacy messaging and information processing are discussed.

Notes

¹This study received approval from George Mason University in December 2013 and from the University of Iowa in January 2014.

²Two types of respondents were cut from the data during cleaning: participants who had insufficient variance in their response (more than 60% of questions were answered with a 4; N =  23) and participants who spent more than 6 minutes on the video page (N =  89). Six minutes was selected as the cutting point for two reasons: (a) It was roughly double the longest video length, and (b) a boxplot indicated that all scores over 365 seconds were outliers.

³These descriptive statistics were similar across schools. T tests suggest there are no meaningful differences in participant age and ideology, whereas a significant (χ² = 9.14, p < .01) difference does exist in gender distribution. School is controlled in all subsequent analyses. We also tested the analyses controlling for gender, but it does not change the relationships described in the article and is therefore excluded from analyses.

⁴This experiment included two additional conditions that were not analyzed in this study. These factors were not crossed with the experimental factors reported here, so participants in these conditions were excluded (N =  133).

⁵The host and two guests were White men. Although we recognize that race and gender could influence responses to the political program and the media literacy PSA, we selected White men to reflect the current political talk show environment and to avoid confounding race and gender with political stance. We encourage researchers to test the possible influence of a different gender or racial profile of the hosts and guests on these perceptions.

⁶We would like to acknowledge the Mass Communication Research Center at the University of Wisconsin-Madison, with whom we worked to create this video, with a special thanks to D. Jasun Carr, Mitchell Bard, Leticia Bode, and Dhavan Shah.

⁷We also tested this operationalization of neutral programming against the definition from the HME literature, which we define on page 5 as programming seen as neutral by impartial observers – in this case, ideological moderates. In an analysis of covariance (ANCOVA) controlling for school and exposure to the PSA among moderates, we see a significant main effect of host role, F(1, 124) = 4.24, p < .05: Moderates rated the neutral host as neutral (M =  3.98, SE =  .17), the conservative host as conservative (M =  3.57, SE =.19), and the liberal host as liberal (M =  4.36, SE =.19).

⁸Copies of the scripts for both experimental manipulations are also available at: http://emilyk.vraga.org/

⁹We also tested whether there was a significant interaction between the two experimental manipulations on the manipulation checks. These interaction terms were not significant for either variable and thus are not reported here.

¹⁰Although neither of the pairwise comparisons are statistically significant, we interpret these patterns of differences given the significant interaction term. For transverse interactions (as between the congruent vs. incongruent host and the PSA), the pairwise comparisons do not need to be significant to indicate valid differences between conditions.

¹¹A series of 7-point scales asked participants to rate whether the advertisement was informative, useful, accurate, relevant, interesting, and credible. These six items were combined into an index (Cronbach's α = .85, M =  4.28, SD =  1.13), with a higher score indicating higher evaluations of the advertisement.

Additional information

Notes on contributors

Emily K. Vraga

Emily K. Vraga (Ph.D., University of Wisconsin-Madison, 2011) is an assistant professor in the Department of Communication at George Mason University. Her research focuses on how individuals process news and political information, particularly in response to disagreeable messages they encounter in digital media environment.

Melissa Tully

Melissa Tully (Ph.D., University of Wisconsin-Madison, 2011) is an assistant professor in the School of Journalism and Mass Communication at the University of Iowa. Her research focuses on the intersection of digital media and civic participation.