Disentangling Debate and Media Effects in the 2016 Elections

• In Event: Media Effects

Abstract

The studies that show that exposure to presidential debates can affect voters’ perceptions and voting intentions (Benoit, Hansen, & Verser, 2003) tend to treat debate exposure as independent from media exposure. However, debates are viewed on media outlets and debate segments are the focus of subsequent news coverage, some of it strongly partisan (Stroud, 2011). When scholars assess debate effects through cross- sectional surveys that aggregate data from interviews conducted over multiple days, they risk misattributing media effects to debates. In this study, we use 2016 cross- sectional data to distinguish the effects of each on perceptions and intentions. To do so, we focus on an argument underscored by content hacked from a Democratic account by Russians, released through WikiLeaks, and prominently featured in the second debate as well as in media discussions before and after that debate. At issue was a hacked speech segment from an unreleased address by Democratic nominee Hillary Clinton in which she said it was sometimes necessary for leaders to take one position in private and another in public. Our analysis shows that partisan media exposure and exposure to the second debate each independently predict the belief that Clinton speaks differently in private and public, with partisan media exposure the stronger predictor of the two (see details below).
Using a multi-method approach, we estimated media attention given to WikiLeaks and the private position meme by analyzing online (n = 8372), print (n = 237) and television (n = 192) news items between October 7th 2016 and October 20st. We analyzed data from two cross sectional surveys (n = 2,021) conducted in the weeks after the email release to examine whether exposure to the second presidential debate and exposure to conservative media were associated with: 1) the belief that Hillary Clinton speaks differently in private and public, and; 2) changes in intentions to vote for Clinton.
For each medium, we examined the number of mentions of both “WikiLeaks” and the “private and public position” statement. We found that conservative online news outlets wrote more about WikiLeaks (M = 26.5% of total election articles) than moderate (M = 7.95%) and liberal (M = 7.2%) sites. Similarly, conservative media (M = 3.96%) wrote more about the public and private position meme than moderate (M = .96%) or liberal ones (M = 1.2%). Similarly, conservative print media covered WikiLeaks more often (M = 27.6%) than moderate (M = 5.83%) and liberal (M = 5.53%) media. As for the private position argument, conservative media (M = 0.43%) covered this specific topic more often than moderate (M = 0.22%) media, but less than liberal media (M = 0.66%). Finally, in broadcast and cable news, the conservative channel Fox News discussed WikiLeaks much more prominently in the shows captured (M = 82.3%) than the moderate CBS (M = 23.7%) or the liberal MSNBC (M = 50.0 %). The pattern is similar when looking at the proportion of news items containing the phrases “private position” and “public position;” Fox news used the terms more often (M = 17.6%) than CBS (M = 2.2%) or MSNBC (M = 6.2%).
As for effects, an OLS regression showed that, controlling for intention to vote for Clinton, demographics, and political preferences, the associations of the belief that Hillary Clinton speaks differently in private and public with consumption of conservative and moderate media were significant. Specifically, relying on conservative media was positively associated with this belief (Beta = .15, p < .001) and relying on moderate media was negatively associated with it (Beta = .09, p < .001). Surprisingly, reliance on liberal media was not significantly correlated with this belief (Beta = -.02, p = .29). Watching the second debate was positively associated with the belief (Beta = .04, p = .03). The model explained 23.5% of the variance in the belief.
Finally, a logistic regression (Pseudo R2 using adjusted McFadden = .44) found that when controlling for demographics and political preferences, the belief that Clinton speaks differently in public and private was negatively and significantly associated with intention to vote for her on election day (OR = -.21, p < .001).
To sum, both exposure to partisan media, and to a lesser degree exposure to the debate, were associated with perceptions about Clinton that in turn was associated with intentions to vote. More research is needed to understand the surprising finding regarding liberal media.

References
Benoit, W. L., Hansen, G. J., & Verser, R. M. (2003). A meta-analysis of the effects of viewing U.S. presidential debates. Communication Monographs, 70(4), 335–350. https://doi.org/10.1080/0363775032000179133
Jamieson, K. H. (2018). Cyberwar: How Russian Hackers and Trolls Helped Elect a President What We Don’t, Can’t, and Do Know. Oxford, New York: Oxford University Press.
Stroud, N. J. (2011). Niche News: The Politics of News Choice. Oxford University Press, USA.

Authors

• Yotam Ophir, University of Pennsylvania

• Dror Walter, Georgia State University

• Kathleen Hall Jamieson, University of Pennsylvania