Masha Krupenkin
Boston College
“If a Tree Falls in the Forest: COVID-19, Media Choices, and Presidential Agenda Setting”
Abstract: During the COVID-19 crisis, political observers noted the dangers of a misinformed or under-informed population. President Trump’s daily press briefings, often covered live, offer a stark example of the potential exposure to misinformation. We use complete transcripts of national cable news, network news and local news from around 800 local TV channels across all 210 Designated Media Areas to examine the media narratives consumed by the public. We use web search data to examine the effects of media coverage on information search and behavior. We find that while the direct effects of Trump press conferences on contemporaneous internet search are weak, these conferences tend to direct media coverage for the next 24 hours post-conference, which in turn drives greater search volume. These effects are stronger for more polarized topics, including Chinese culpability for the pandemic and the use of hydroxychloroquine as a treatment for COVID-19. Different TV news outlets show disparate responsiveness to Trump’s press conferences, with cable news as most responsive and local news as least. Local news tends to focus on softer news, and eschews the partisan debates prevalent in cable coverage. These results demonstrate the power of the media as a disseminator of public health messaging and a mediator of presidential agenda setting.
Bio: Masha Krupenkin is an Assistant Professor of Political Science at Boston College. Her research uses web search data to understand Americans’ revealed preferences about partisanship, immigration, and media consumption.
Deen Freelon
University of North Carolina at Chapel Hill
“Tweeting Left, Right & Center: How users and attention are distributed across Twitter”
Abstract: More than 48 million Americans use Twitter, one of the most popular social networking platforms, and one used extensively by media and political junkies. This study analyzed more than 86 million tweets posted in 2017 to reveal how users from across the political spectrum engage differently with news issues and major media outlets on Twitter. We assigned randomly-sampled users ideology scores based on who they follow, then divided them into four segments: extreme left, center left, center right, and extreme right. This talk is based on a report commissioned by the Knight Foundation and will discuss its key findings. Read the report here.
Bio: Deen Freelon is an associate professor in the Hussman School of Journalism and Media at UNC-Chapel Hill. These days he is mostly interested in disinformation, hyperpartisan communication, and computational method
Giovanni Luca Ciampaglia
University of South Florida
“Political audience diversity and algorithmic bias in news recommendation”
Abstract: Research shows that people have strong preferences for pro-attitudinal content. These tendencies are exacerbated by social media platforms, which often amplify low-quality content because it generates high levels of engagement among like-minded audiences. An analysis of news source reliability ratings compiled by domain experts and web browsing histories from a diverse sample of 19,282 U.S. citizens suggests that using political diversity of online audience as a metric can help social media platforms identify high-quality media outlets. Observational results indicate that sites with more extreme and less politically diverse audiences have lower journalistic standards, while popularity, measured by the number of accesses a website receives, does not predict higher quality journalistic standards. Partisan audience diversity is thus a valuable signal of higher journalistic standards that should be incorporated into algorithmic ranking decisions. To understand the implications of these finding, we incorporate audience diversity into a standard collaborative filtering recommendation framework and show that our improved algorithm increases trustworthiness of websites suggested to users, while keeping recommendations relevant. The effect is especially strong for groups with stronger selective exposure to misinformation. Our findings suggest a scalable algorithmic method to improving news content circulated online.
Bio: Giovanni Luca Ciampaglia is an assistant professor at the Department of Computer Science and Engineering at the University of South Florida (USF). He is interested in information quality in cyber-human systems, in particular trustworthiness and reliability of information in intelligent systems.