How Young, Uninsured Americans Respond to News Coverage of Obamacare: An Experimental Test of an Affective Mediation Model

Abstract

This experiment integrated theory from multiple domains to examine how aspects of news coverage of the Affordable Care Act (ACA) and differences in participants’ cognitive and emotional contingent conditions interact to shape attitudes and behavioral intentions toward health care legislation. Using a sample of uninsured young adults (N = 1,056), we tested an affective mediation model, which assessed the mechanisms through which media frames, exemplar case studies, and individual predispositions affect this type of news consumer. Results demonstrate the complicated pathways through which emotions mediate the effects of news coverage of ACA based on political predispositions, the need for orientation toward the health care issue, and the influence of equivalency framing in the form of example cases. These findings contribute to a more nuanced explanation of the causal mechanisms underpinning framing effects of public policy news coverage on an understudied population. The need for further examination of emotion along with cognition when investigating framing effects of public policy news is discussed, and the importance of exemplar cases as a significant manifestation of the effects equivalence framing is highlighted.

Notes

¹ Scholars have conceptually defined game-framed coverage as news that includes a preponderance of stories about the competitive aspects of politics, such as who is ahead in the polls; who is gaining momentum; who is most effectively executing party strategy; and other similar result-oriented concerns on winning, losing, strategy, and tactics (Cappella & Jamieson, 1997; Iyengar, Norpoth, & Hahn, 2004).

² Our sample includes a higher percentage of White individuals (nearly 72%) than data suggest is represented in the U.S. nonelderly uninsured population (45% White; Kaiser Family Foundation, 2016). Also, overall, according to 2015 U.S. Census data, our sample is fairly comparable to the overall U.S. population, but slightly more White, slightly better educated, and slightly less Republican. As such, we are not claiming our data are representative of the whole population given we are conducting an experimental test of mechanisms and not a public opinion survey.

³ An online pretest experiment (N = 201; M_age = 37.80, SD_age = 12.49, 56.7% male, 85.1% White) was used to test the effectiveness of the frame manipulations and any bias due to manipulation of the stories. Participants were randomly assigned to view one of the 16 stimuli. They were then asked to rate their agreement, from 1 (strongly disagree) to 7 (strongly agree), with four statements related to perceptions of a game versus issue frames, such as “The story focused more on politics than on health care” and “The story was mostly about political maneuvering” (α = .91, M = 4.99, SD = 1.32; higher numbers indicate stronger perceptions of a game frame). Those who saw the game-framed message reported higher (M = 5.15, SD = 1.36) game frame perceptions than did those who saw the issue frame (M = 4.83, SD = 1.26), F(1, 196) = 3.88, p = .01 (analysis of covariance controlling for exemplar condition). The number and valence of exemplars did not require pretesting as they are inherent components of the message (see O’Keefe, 2003). The pretest sample’s demographics differed from the actual experiment’s sample in terms of demographics (overrepresenting male and White individuals); however, because of the focus of the pretest as a manipulation check, and the large total sample size used in the experimental conditions, this difference was not a point of concern regarding interpretation of the results of the actual experiment.

⁴ Of the 1,056 cases in the data set, four contained missing data (each of those was missing data only for one or two variables). To run the path analyses in MPlus software, which requires there not be any missing data, these cases were removed. Four of 1,056 is only 0.38% of the data, far under the 5% threshold of acceptability for missing data (Kline, 2011).

⁵ Although the chi-square was significant, chi-square is a notably poor indicator of model fit for models with more than 400 cases (Kline, 2011), and therefore we relied on additional fit indices such as root mean square error of approximation, comparative fit index, and standardized root mean square residual, in line with the model fit criteria proposed by Hu and Bentler (1999).

Additional information

Notes on contributors

Jason A. Martin

Jason A. Martin (Ph.D., Indiana University, 2011) is an associate professor and chair of journalism in the College of Communication at DePaul University. His research interests include political communication and media law with special attention to developments in mobile political communication that affected underprivileged populations and the First Amendment rights of digital journalists.

Jessica Gall Myrick

Jessica Gall Myrick (Ph.D., University of North Carolina at Chapel Hill, 2013) is an associate professor of film/video and media studies in the College of Communications at The Pennsylvania State University. Her research interests include the role of emotions in shaping audience reactions to health and environmental messages, the impact of interactive communication technologies in this realm with an emphasis on how media use can lead to beneficial and prosocial outcomes, and how to increase civic engagement in processes related to public health and environmental policies.

Kimberly K. Walker

Kimberly K. Walker (Ph.D., Indiana University, 2009) is an assistant professor in the Zimmerman School of Advertising & Mass Communications at the University of South Florida. Her research interests include the ways in which vulnerable groups utilize online support groups for health decisions, how media messages affect health behaviors and decisions, and the impact of health literacy on health behaviors and decision-making.