Presentation Abstracts

Friday, Oct. 16th

9:30 – 10:30 Keynote Speaker Lance Bennett + Q&A

10:45 – 12:15 Democratic Resilience Beyond ‘Polarization’

1:15 – 2:45 News & the Changing Public Sphere

3 – 4:30 Technology & the Reordering of Knowledge & Power

Saturday, Oct. 17th

9:30 – 11:00 New CCCR Projects: Methodological, Measurement, and Theoretical Advancements

Friday, October 16th

9:30 – 10:30 Keynote Speaker Lance Bennett + Q&A

 Lance Bennett, “The Neutrality Trap: AI and the Future of Democracy”

A growing body of research shows left-leaning political biases in most of the dominant Artificial Intelligence agents. However, a closer look suggests that those claims of bias are conflated with liberal democratic values and scientific methods. In the current political context, right wing governments, politicians, and many business clients are pushing for “neutrality” in AI. These pressures could result in a “balance as bias” syndrome that opens AI to political manipulation, noise, and chaotic levels of uncertainty. This talk puts the focus on the broader geopolitical forces shaping the guardrails and safety standards of AI agents. A comparative framework examines the sovereign AI regimes forming in the US, the EU, and China, among others, with an eye to how to protect democratic and scientific values in a world where both democracies and science are under pressure.

10:45 – 12:15 Morning Panel: Democratic Resilience Beyond ‘Polarization’

Lilliana Mason, “Nobody Wants This: The Disproportionate Power of an Insurgent Faction of Intolerant Americans.”

The 2024 election provided an illusion that most Americans are on board with the MAGA agenda. But this is not true. The portion of the electorate that agrees with the more intolerant portion of the MAGA movement isn’t a majority of Americans. It isn’t even all Republicans. The group of Republicans who are supportive of this agenda are empirically distinct from the rest of Americans, including their fellow Republicans. This project examines the factional divide in the Republican Party that divides an extremist movement from a more democratically-minded set of voters. 

Jeff Niederdeppe, “Rigorous & Replicated Communication Science to Challenge Conventional Wisdom about Race and Public Policy.”

This talk will make a case for the value of repeated conceptual replications in communication science to advance knowledge claims and challenge conventional wisdom about the impact of messaging about controversial topics. I will use messaging about diversity, equity and inclusion in the United States as a case study on the value of rigorous and rapidly responsive messaging studies to dispel conventional wisdom about apparent racial and partisan divides surrounding politicized public policy issues. Reporting on nine randomized message testing experiments spanning six topics and collecting data from over 40,000 respondents, I will show that messaging about racial inequality in the US is not inherently divisive if theory and evidence-based messaging strategies are used strategically. I will discuss the implications of these findings for democratic governance, communication science, and social science research more generally.

Jianing Li, “Affective Evaluation Networks: Charting the Interdependence of Affects Toward Identity Groups in Communication.”

Affects play a crucial role in communication. While a person’s feelings about various groups are inherently connected, in research practice, affects are often treated as independent rather than interdependent. We introduce an affective evaluation network approach to study how affects toward identity groups are structured in individuals’ minds and contribute to the constitution of shared realities. We show the value of this framework in addressing several concerns with existing approaches. As a demonstration case, we assess U.S. panel data where respondents reported affects toward 16 pluralistic groups. Our approach offers findings that expand prior accounts of polarization and campaign engagement.

Jaime Settle, “Using Local Media to Study Agentic Processes of Subnational Authoritarianism & Democratic Erosion in the United States.”

Understanding the origin, extent, and consequences of democratic erosion necessitates broadening the aperture beyond national-level threats alone to systematically analyze threats to democracy at the more local level. What we lack are systematic and comparable data about the agent-led events that incrementally yet substantially undermine the quality of democracy. The Subnational United States Democratic Erosion Event Database Project provides such an analytical approach by 1) generating a dataset of agentic events that undermine democracy at the state-level in the United States and expanding extant datasets about US national-level erosion events; 2) examining patterns in subnational backsliding as well as the interplay between national and sub-national democratic erosion events; and 3) analyzing the way that structural and partisan biases in media coverage distort Americans’ understanding of the threat of democratic erosion at both the national and sub-national levels.

Dannagal Young, “Individual Well-Being and Democratic Health: The Mediating Role of Self-Uncertainty on Support for Democracy.”

The rise in democratic backsliding around the globe has renewed scholarly attention to the individual-level psychological factors that may undermine support for democratic norms. This study examines how poor well-being—a composite measure encompassing depression, loneliness, poor physical health, anxiety, and low meaning in life – relates to anti-democratic attitudes in the U.S., and whether self-uncertainty mediates this relationship. Using Wave 2 of a three-wave YouGov panel survey of U.S. adults (N = 901), we find that poor well-being (excluding anxiety) is a significant positive predictor of anti-democratic attitudes, while anxiety uniquely predicts greater democratic support. Mediation analyses reveal that self-uncertainty partially suppresses the well-being–anti-democracy link: poor well-being strongly predicts self-uncertainty, but self-uncertainty is itself negatively associated with anti-democratic attitudes. This mediation pattern is asymmetric by party. Among Democrats, self-uncertainty appears to reinforce partisan identity in ways that reduce anti-democratic attitudes—consistent with uncertain Democrats being guided by cues of Democratic elites regarding Trump’s threats to democracy. Among Republicans, we find no evidence of self-uncertainty playing a mediating role. These findings underscore the importance of individual emotional and psychological health for democratic resilience, and highlight the unique role played by anxiety in democratic health. They also point to a potential partisan asymmetry in the psychological pathways through which poor well-being shapes democratic attitudes.

1:15 – 2:45 Early Afternoon Panel: News & the Changing Public Sphere

Danielle Brown, “What If Trust Is The Problem?”

Scholars and practitioners have long identified how journalism has contributed to the misrepresentation and devaluing of disenfranchised communities. To correct these wrongs, journalism has leaned heavily on the cure-all of trust-building. Yet, these efforts can yield lackluster results. What if this obsession with trust building is actually part of the dysfunction? Drawing on extensive statewide surveying and qualitative data collection across Midwestern communities, this talk unpacks the promise and peril of trust, challenging us to rethink our frameworks for media repair, narrative change and democracy building. 

Stephanie Edgerly, “Hope and Action: What Positive News Can (and Can’t) Do.”

News outlets are experimenting with positive news by adding good stories to traditional offerings and launching good-news-only products. But what impact do these efforts have on audiences? How much and what type of positive news is needed to produce democratically desirable effects? This talk presents findings from two experiments that manipulate the mix of negative-to-positive stories and amusing-to-uplifting stories to examine the effects on audiences’ emotions, outlook about the future, and willingness to participate. The results highlight the complexities of boosting people’s positive emotions while motivating them to act. The findings offer actionable insights for news organizations looking to provide some relief in hard times.

Erik Bucy, “When Reality Pushes Back: Narrative Politics and Citizen Witnessing.”

Seminal research has shown that political elites and institutional media—sometimes at odds, sometimes in concert—possess outsized power to shape public understanding of politicized events and legitimize extraordinary interventions (e.g., wars, police shootings, armed insurrections). Yet citizen witnessing can disrupt these efforts. Focusing on the Minneapolis shootings of Renée Good and Alex Pretti, this paper examines when elite attempts to define political reality fail to stabilize into a coherent narrative. In hybrid media environments, bystander video, citizen testimony, journalistic investigation, and networked publics can rapidly challenge official accounts and reshape the evidentiary terrain. The central question surrounding democracy’s imperative to tether politics to reality, then, is not simply how elites construct and impose false narratives, but when and why those constructions fail to hold.

Dominik Stecula, “Ambient Politics in the Post-Institutional Public Sphere.”

In our rapidly changing media environment, we increasingly encounter political opinion leadership in spaces not organized around politics. In this project, we use podcasts as a case study of this new post-institutional public sphere. We analyze data from over 7,000 podcasts, coupled with original survey data fielded by Verasight, as well as interviews with podcasts hosts to show that political discussion extends well beyond explicitly political programs, and reaches audiences with lower institutional trust and weaker partisan attachments. Frequently, listeners don’t even recognize the political content they hear. We ask what this means for democratic citizenship when political opinions arrive with less of the institutional scaffolding that historically helped citizens verify, curate, and organize information. We then consider the implications of this new media environment for American democracy. 

Stuart Soroka, “Tracing the Proliferation of Polarizing News Content.”

There are widespread concerns about political polarization, and about the tendency for mass media to make matters worse. I argue that understanding the role that media play requires an ecosystem-level analysis, tracing the nature of content across sources, markets, and time. In this talk, I first use formal models to make a theoretical case for ecosystem-level analysis. I then consider a combination of dictionary-, unsupervised learning-, and LLM-based approaches to the empirical measurement of polarizing news content across four decades of news coverage.

3 – 4:30 Late Afternoon Panel: Technology & the Reordering of Knowledge & Power

Dave Karpf, “Silicon Valley’s Reactionary Turn.”

Well, that escalated quickly! Not so long ago, the biggest figures in Silicon Valley presented themselves as vocal democracy defenders, a bulwark against democratic erosion and the substrate for prosocial civic engagement. By 2024, those same figures decided that, if they had to choose between a robust administrative state and authoritarianism, they would side with the authoritarians. This paper, based on a chapter from a forthcoming book, discusses how and why the tech billionaires took such an abrupt reactionary turn. 

Erika Franklin Fowler, “Drastic Cuts, Delayed Ramifications, and Democratic Needs: Assessing Strategies to Bolster Support for Science and the Promise of Public Goods.”

Although politicization of science is nothing new, the scope and scale of threats to the scientific enterprise in the United States dramatically escalated in 2025. Unprecedented disruptions to the production of science have since unfolded, including grant cancellations on topics that run afoul of administration priorities, pressures to shift university policies as a condition of funding, and threats to delay and reduce funds universities rely on for research infrastructure. However, we have limited understanding of the public’s awareness of these disruptions and their societal impact, and of how the public might be mobilized in support of science. Drawing on nationally representative surveys and media tracking data, we assess public awareness and perceptions of the impact of these disruptions, their potential connection to media attention and existing patterns of potential for mobilization, and then we test a variety of strategies to bolster support through survey experiments.

Yini Zhang, “An AI-Saturated Political Communication Environment? Elite Production, News Amplification, and Public Perception.”

As generative AI increasingly permeates public and private life, how is AI-generated content used by political elites, amplified by news media, and perceived by the public? This talk draws on two ongoing studies examining these interconnected processes. One study investigates Donald Trump’s use of AI-generated images on Truth Social and their relationship to news attention, showing that synthetic imagery has become a meaningful component of his visual communication. The other examines public perceptions of AI-generated political content, finding substantial overestimation of its prevalence, with politically motivated reasoning further intensifying these misperceptions. Together, the studies point to two emerging features of the political communication environment: the normalization of AI-generated content in elite communication and the gap between actual and perceived prevalence of such content. As uncertainty about authenticity grows, confidence in political information can be further weakened and judgments of credibility can become increasingly shaped by partisan predispositions.

Rebekah Tromble, “Self-Censorship among U.S. Scholars: Preliminary Evidence from Two Surveys.”

Fil Menczer, “AI Good vs. Evil.”

Online manipulation is a battlefield in which AI is being used as a weapon by both attackers and defenders. This talk overviews research exploring both sides. Bad actors are exploiting generative AI to easily and cheaply create fake profiles and content that are hard to recognize. At the same time, AI provides us with tools to mitigate online harms, such as methods to detect inauthentic accounts and coordinated information operations, scale up fact-checking, and support moderation. Even on the good side of the battlefield, we must account for the limits of AI, unintended consequences when people interact with AI, and vulnerabilities of AI to subtle forms of bias and abuse.

Saturday, Oct. 17^th

9:30 – 11:00 Morning Panel – New CCCR Projects: Methodological, Measurement, and Theoretical Advancements

Andrew Trexler, “Measuring Untapped Demand for Accessible Public Interest News.”

Journalists and academics alike have long lamented the public’s limited exposure to substantive news content about public policy. We examine the extent to which this deficit is driven by supply versus demand. Most modern news organizations rely on digital engagement metrics to gauge audience interest and decide what content to produce. Limited exposure to substantive policy content is thus believed to be largely a product of low demand, as outlets prioritize creating content on higher-traffic topics like partisan conflict. However, engagement metrics paint an incomplete picture of demand. Because audiences can only choose from content that is available, audience metrics exclude demand signals over content that is never produced. As such, these metrics cannot capture the preferences of potential audiences. We field studies to test demand for public interest news among traditionally disengaged audiences. Study 1 shows that the less politically engaged demand lower compensation to read public interest news stories than to read stories focused on partisan conflict. Study 2 likewise shows that traditionally disengaged audiences prefer public interest news to conflict-style coverage, but this does not translate to reputational benefits for news outlets that produce more public interest content.

Yoo Ji Suh, “From Blame to Gridlock: Estimating Media Effects Through Content–Survey Integration.”

Why do some people see political compromise as unacceptable while others do not? This project argues that part of the answer lies in how news and social media assign blame. Blame attributions portray an intentional agent as causing harm to a victim, and in doing so mark that agent as morally culpable. Content that casts the other side as blameworthy undermines their standing as legitimate and reasonable partners for negotiation, making compromise feel like concession to wrongdoing. To test this claim, I develop a media content–survey integration design, focusing on the lead-up to the 2024 presidential election when both the volume of blame attributing coverage and the variation in individuals’ exposure to it are at their highest. Building on the linkage tradition in media effects research, I combine content analysis of news and social media with survey data on attitudes toward political compromise. The design proceeds in three steps. First, I build algorithms that estimate individuals’ media exposure from survey data. Second, I analyze media content from the period the survey’s media use questions ask about to construct an index of blame attribution. Third, I merge the two and test whether estimated exposure to blame attribution predicts aversion to political compromise. Beyond its substantive contribution, the design offers a strategy for testing the effects of specific content characteristics rather than outlet- or platform-level exposure alone.

Luhang Sun, “Beyond Silicon Samples: Predicting Voter Choice via Micro-Alignment of AI Digital Twins.”

Large language models (LLMs) are increasingly used to simulate public opinion, yet what information best aligns AI digital twins with individuals, and for whom these simulations work, remains unclear. Using a nationally representative U.S. survey (N = 1,768), we constructed individual-level digital twins to predict 2024 presidential vote choice and optimized prompts across 13 profile dimensions. We found that demographics provided limited predictive utility, while political worldviews and media use were more informative, and additional information did not necessarily improve accuracy. Importantly, LLMs’ predictive performance varied substantially across social groups, particularly by age and party identification. These findings advance a micro-alignment approach that considers both what information enables accurate individual-level simulation and whose behaviors LLMs can reliably predict.

Sean Pauley, “U.S. Latino Television Repertoires.”

Studies of U.S. Latino media consumption often focus on a few specific sources rather than a broader set of sources people watch in combination, which would better account for people’s information environments. Media repertoires, or the combination of sources people watch, are a useful framework for understanding people’s information environments, but repertoires research has generally neglected Latino audiences. We address this gap by focusing on Latino television repertoires through a representative sample of 5,000 Latinos in the U.S. and a novel Bayesian approach to repertoire analysis. We find that Spanish-language networks like Univision and Telemundo play a large role in structuring Latino repertoires. We also find that language, prioritization of Latino identity, connection to Latin American culture, and how recently one’s family immigrated to the U.S. are strong factors in predicting respondents’ media repertoires.

Chris Wells, “Varieties of Cultural Experience: How Taste & Daily Life Relate to Political Attitudes & Involvement.”