Interaction between Media, Audiences and Economic Pressure Drives Polarization

• In Event: Digital News and Democracy

Abstract

AIM: The origins of political polarization are poorly understood. Whether media contributes to the observed increase in ideological and affective polarization in the United States is currently unclear. This study set out to determine whether and how the U.S. news media contributed to the observed partisan differences in attitudes, beliefs and behaviors during the Covid-19 pandemic. METHOD: We surveyed 29 Twitter accounts of representative left-leaning (n=11), right-leaning (n=11), and ideologically neutral (n=7) news media, which had a combined audience of 220 million individual accounts. The analysis of Twitter communication presents several advantages: 1) we can survey thousands of tweets and collect data on audience engagement; 2) we can survey the real-world reactions of millions of people to each tweet; 3) we can geo-locate the retweets of each post and determine which cities across the U.S. amplify particular aspects of news media coverage. We performed a longitudinal analysis of Covid-19 coverage, from January 2020 to December 2022 to assess the emergence of any polarization patterns among audiences and across U.S. cities. We collected 3 million posts by the 29 news media entities over a 3-year period, of which 407,804 tweets were Covid-19-related. We developed and validated automatic filters to identify the major topics of Covid-19 coverage (masking, vaccination, public health, politics, economy, social issues, race and ethnicity, religion, education, and extremism and conspiracy). Each automatic filter identified 95-99% of the expected topics. We used multivariate regression analyses to model the behavior of audiences, and stepwise regression models to test whether combinations of selected variables were the better predictors of audience behaviors or patterns of Covid-19 mortality. RESULTS: We uncovered a dramatic difference between all conservative audiences, on one hand, and the liberal and politically neutral audiences, on the other hand. Followers of the 11 right-leaning news media accounts (including Fox News, Breitbart, OANN, and Newsmax) were significantly more engaged by Covid-19 content than the followers of liberal and centric media accounts; this stark preference for Covid-19 communication appeared in November 2020 and continued thereafter. We further found that Covid-19 communication by conservative-leaning media was significantly more politicized compared to liberal-leaning and centric news media. Multivariate regression modeling showed that politicized Covid-19 content, as well as extremist and conspiratorial content, were most engaging for conservative-leaning audiences. We also found that segments of conservative audiences that were more engaged by any type of news were also most engaged by the Covid-19 coverage, suggesting that the segmentation of news media audiences and their self-exposure to pro-attitudinal coverage is critical to their engagement by Covid-19-related news. We then asked whether the amplification of news media coverage in U.S. cities influenced local Covid-19 outcomes. Stepwise regression modeling of over 25 million geo-located retweets by the general population showed that amplification (retweeting) of conservative media’s extremist and conspiratorial content was the most significant, positive predictor of Covid-19 mortality among the 1,200 U.S. cities we surveyed. By contrast, amplification (retweeting) of liberal and centric media tweets that debunked extremist and conspiratorial news was the most significant negative predictor of mortality among the U.S. cities. We finally asked whether macro-economic aspects, which strongly fluctuated during the pandemic, interacted with the amplification of media tweets to explain the local Covid-19 mortality. We found that local unemployment was a positive predictor of mortality and interacted with the amplification of conspiratorial and extremist language to explain Covid-19 mortality in conservative cities. CONCLUSIONS: These results uncover a progressive polarization of Covid-19 news preference among U.S. audiences. This observed audience polarization was driven by audience segmentation and exposure to pro-attitudinal coverage. Amplification of Covid-19 news coverage across the U.S. was correlated with Covid-19 mortality. Furthermore, an interaction between local socio-economic pressure (unemployment) and local amplification patterns of Covid-19 tweets was a better predictor of local Covid-19 mortality than amplification of news media Covid-19 tweets alone. We thus propose that a triple interaction between the news media’s political bias, the segmentation of American audiences, and socio-economic pressure is responsible for the emergence and persistence of Covid-19-related ideological polarization among Americans – which may explain the observed polarization of Covid-19 behaviors and the resulting polarized patterns of Covid-19 mortality across America during the pandemic.

Authors

• Wenyou Ye, University of Maryland

• Liviu Aron, Harvard Medical School