What can sentiment in cable news coverage tells us about the 2024 campaign?


Dr. Gavin Ploger

Howard R. Marsh Postdoctoral Teaching Fellow at the University of Michigan. He studies political communication, political psychology, and public opinion in the United States, especially as they relate to polarization, democracy, partisan identity, and news representations of politics.

Email: gploger@umich.edu



Prof. Stuart Soroka

Professor in the Departments of Communication and Political Science at the University of California, Los Angeles. His research is focused on political communication, political psychology, and the relationships between public policy, public opinion, and mass media. 

Email: snsoroka@ucla.edu


U.S. Election 2024

51. The powers that aren’t: News organizations and the 2024 election (Dr Nik Usher)
52. Newspaper presidential endorsements: Silence during consequential moment in history (Dr Kenneth Campbell)
53. Trump after news: a moral voice in an empty room? (Prof Matt Carlson, Prof Sue Robinson, Prof Seth C. Lewis)
54. Under media oligarchy: profit and power trumped democracy once again (Prof Victor Pickard)
55. The challenge of pro-democracy journalism (Prof Stephen D. Reese)
56. Grievance and animosity: Fracturing the digital news ecosystem (Dr Scott A. Eldridge II)
57. Considering the risk of attacks on journalists during the U.S. election (Dr Valerie Belair-Gagnon)
58. What can sentiment in cable news coverage tell us about the 2024 campaign? (Dr Gavin Ploger, Dr Stuart Soroka)
59. The case for happy election news: Why it matters and what stands in the way (Dr Ruth Palmer, Prof Stephanie Edgerly, Prof Emily K. Vraga)
60. Broadcast television use and the 2024 U.S. presidential election (Jessica Maki, Prof Michael W. Wagner)
61. Kamala Harris' representation in mainstream and Black media (Dr Miya Williams Fayne, Prof Danielle K. Brown)
62. Team Trump and the altercation at the Arlington military cemetery (Dr Natalie Jester)
63. Pulling their punches: On the limits of sports metaphor in political media (Prof Michael L. Butterworth)

Media coverage often provides a valuable indicator of the ‘tone’ of the campaign, capturing some combination of journalists’ assessments of the candidates, changes in the public’s electoral preferences, and the sentiment of the campaigns and candidates as well. Examining how news covers the candidates consequently provides some useful information—perhaps more diagnostic than predictive—about election campaign dynamics. 

We have accordingly been tracking the tone of media coverage of both Kamala Harris and Donald Trump throughout the 2024 election campaign. Our data include all sentences mentioning either Kamala Harris, the Democratic presidential candidate, or Donald Trump, the Republican presidential candidate, in all archived broadcast transcripts from Nexis Uni, for two major cable news channels: CNN and Fox News.

Once sentences are extracted from news transcripts, each sentence is scored for sentiment by counting the number of positive and negative words using the Lexicoder Sentiment Dictionary. Each sentence is assigned a net sentiment score, the difference between the numbers of positive words and negative words. Following past work, it is calculated as follows: log((positive words + .5) / (negative words + .5)). Weekly results then capture the mean of net sentiment in all Republican-candidate sentences, and the mean of net sentiment in all Democratic-candidate sentences. All sentences, regardless of channel, are given equal weight in this measure.

The figure below accordingly shows the sentiment of news coverage of the Democratic and Republican candidates for president on CNN and Fox News during the lead-up to the 2024 election. The x-axis shows the start date for each week. Circles indicate the average sentiment for each candidate during each week, with whiskers showing 95% confidence intervals.

The top panel of the figure shows that CNN coverage has systematically favored the Harris campaign, describing it with more positive language than the Trump campaign. Note that Harris’ relative advantage in news sentiment is driven by highly positive coverage of her, not by negative coverage of Trump. Indeed, as the figure shows, CNN’s coverage of Trump remained stable and near 0—neither positive nor negative—regardless of events in either campaign. That coverage of Trump is, on balance, neutral, is of some significance given that the typical sentiment of news coverage is slightly negative. Coverage of Trump was thus more favorable than the baseline signal of “average” media coverage, despite intuitions about mainstream media criticism and/or bias against him.

The bottom panel of the figure shows that Fox News coverage also included more positive sentiment about Harris, at least at first. Like CNNFox’s coverage of Trump is consistently neutral. While there were certainly differences in news content between the left-leaning CNN and the right-leaning Fox News—in particular, CNN’s coverage of Harris was clearly more positive than Fox News’ coverage—the pattern of each network’s coverage of the two candidates was broadly similar, especially in the early weeks of the fall. This finding underscores that news sentiment about the candidates is not driven entirely by partisan bias and the tone of journalistic commentary. Rather, the sentiment of coverage likely also reflects the tone of the candidates’ own messaging. Harris’ messaging was generally positive, especially before the last two weeks of the campaign. To this point, Harris’ relatively positive coverage on Fox was evident throughout September. By October, however, the sentiment of Fox’s coverage of Harris had declined to 0 (matching its coverage of Trump), likely due to a combination of shifting public attitudes toward Harris and the Harris campaign’s changing messaging. 

In short, election coverage from CNN and Fox News was consistent with the partisan leanings of these networks. Coverage of the Harris campaign was substantially less positive on Fox (and more similar to coverage of Trump) than on CNN. These partisan leanings were not overwhelming for most of the Fall campaign: both networks covered Harris positively (albeit to different extents), and both networks’ coverage of Trump was ambivalent. In the days leading up to the election, the partisan differences in news sentiment became much clearer. CNN viewers received relatively positive content about Harris; Fox viewers did not. 

The electoral consequences of these differences are unclear. Generally positive coverage of Harris throughout the campaign did not appear to translate into meaningful advantages in electoral support: even during the weeks of consistently positive coverage of the Harris campaign in September, the two candidates were essentially tied in most polls. We take this as an indication—in this campaign, at least—that differences in media sentiment were more likely a function of journalistic commentary and campaign messaging than public preferences. Indeed, the positive sentiment of coverage, especially on CNN (and other left-leaning networks) may have masked rather than illustrated the nature of voter preferences during the campaign.