About

Christopher Andrew Bail is a Professor of Sociology, a Professor in the Sanford School of Public Policy, and a Professor of Political Science and Computer Science at Duke University. He serves as the Director of the Trinity Social AI Initiative. His research includes studying the impact of social media on public opinion and polarization, exemplified by a project that used bots and social media to expose people to opposing views, revealing that such exposure can increase polarization. He is the author of the book 'Breaking the Social Media Prism,' which explores issues related to social media, public discourse, and societal polarization.

Research topics

• Political Science
• Sociology
• Computer Science
• Law
• Public relations
• Social psychology
• Social Science
• Psychology
• Communication
• Media studies
• Internet privacy
• Economics
• World Wide Web
• Epistemology

Selected publications

• The Discomfort of Being Outnumbered Online: Evidence from a Dynamic Discussion Experiment

  2026-04-20

  articleOpen access

  Research on online political discussion has compared partisan "echo chambers" with balanced cross-partisan environments, emphasizing the benefits of exposure to opposing views. Less is known about discussions dominated by out-partisans, which may arise as politics spills into non-political spaces or online forums evolve. We address this gap using a custom-built mobile application that recreates core social media features and enables experimental control over discussion contexts through researcher-generated content and AI-powered synthetic users. Participants were randomly assigned to co-partisan-dominated, out-partisan-dominated, or balanced discussion environments. Participants outnumbered by out-partisans had worse user experiences and showed greater restraint: they reported less comfort sharing opinions, rated other users and the environment more negatively, and produced fewer reactions. The balanced condition yielded surprisingly mixed results—some depolarization, but also less posting and lower comfort, indicating self-censorship in this setting as well. Our findings suggest cross-partisan conversation is not without costs, especially when opposing voices dominate.

  PublisherDOI

• Can Platform Design Encourage Curiosity? Evidence from an Independent Social Media Experiment

  arXiv (Cornell University) · 2026-01-22

  preprintOpen access

  Social media platforms are often criticized for fostering antisocial behavior rather than prosocial behavior. Yet, testing interventions to encourage prosocial dispositions, such as open-mindedness, has been hindered by researchers' limited ability to manipulate platform features and isolate causal effects in commercial environments. We address this challenge through a randomized controlled trial with 2,282 U.S. adults conducted on a new research platform we developed that uses AI bots to replicate live social media dynamics while enabling controlled experimentation. Participants engaged in 15-minute discussions about energy and climate topics, with treatment groups exposed to curiosity priming either through modified on-platform social norms, interface affordances, or both. Results demonstrate that curiosity priming significantly increased question-asking behavior and textual measures of curiosity in user posts, while also reducing toxicity. Although interventions decreased generic engagement behaviors like liking and commenting, they had no significant negative impact on reported app enjoyment or time spent writing posts and replies. Leveraging experimental control over platform features, our findings suggest that platform designs prioritizing curiosity can promote prosocial behaviors among users without compromising user experience.

  PublisherDOI

• Outnumbered Online: The Consequences of Partisan Imbalance in Online Political Discussions

  OSF Preprints (OSF Preprints) · 2026-04-20

  preprintOpen access

  Much research on online political discussions has focused on social media "echo chambers" of like-minded individuals, but less is known about how people respond to online environments dominated by those who are politically dissimilar. We conduct a preregistered experiment using a mobile application that we developed to evaluate how being outnumbered by out-partisans impacts comfort with sharing opinions as well as perceptions of the platform and its users. Our app mimics a social media platform but provides researcher control over platform features to experimentally isolate their effects and employs synthetic users powered by generative artificial intelligence to create a dynamic content feed for the participant. We find that engaging in a online discussion with mostly out-partisans reduces comfort with sharing one's opinions and lowers evaluations of the platform and its users. These findings shed light on a relatively less explored online setting (i.e., an outnumbered context) and highlight the utility of LLMs in social science research.

  Publisher

• Experiments offering social media users the choice to avoid toxic political content

  New Media & Society · 2026-02-23

  article

  Social media platforms increasingly offer users control over their feeds, promising to reduce toxic discourse. This study tests how the mere offer of algorithmic control shapes user experiences. Respondents evaluated identical posts from a fictional platform, with half given the option to filter out toxic political content. Those given this choice reported greater platform satisfaction. However, those who opted for filtering rated content as more hostile than similar respondents who were not offered the choice. A follow-up experiment showed that exposure to only positive content did not reduce hostility ratings; it heightened them compared to exposure to both positive and negative content. These findings challenge the assumption that user autonomy will improve content experiences. Instead, algorithmic choice raises expectations, prompting users to scrutinize content more critically or attempt to “train” the algorithm to align with their preferences. Platforms must consider how expectations, not just content exposure, shape online experiences.

  PublisherDOI

• Can Platform Design Encourage Curiosity? Evidence from an Independent Social Media Experiment

  ArXiv.org · 2026-01-22

  articleOpen access

  Social media platforms are often criticized for fostering antisocial behavior rather than prosocial behavior. Yet, testing interventions to encourage prosocial dispositions, such as open-mindedness, has been hindered by researchers' limited ability to manipulate platform features and isolate causal effects in commercial environments. We address this challenge through a randomized controlled trial with 2,282 U.S. adults conducted on a new research platform we developed that uses AI bots to replicate live social media dynamics while enabling controlled experimentation. Participants engaged in 15-minute discussions about energy and climate topics, with treatment groups exposed to curiosity priming either through modified on-platform social norms, interface affordances, or both. Results demonstrate that curiosity priming significantly increased question-asking behavior and textual measures of curiosity in user posts, while also reducing toxicity. Although interventions decreased generic engagement behaviors like liking and commenting, they had no significant negative impact on reported app enjoyment or time spent writing posts and replies. Leveraging experimental control over platform features, our findings suggest that platform designs prioritizing curiosity can promote prosocial behaviors among users without compromising user experience.

  PublisherOA PDF

• A research agenda for encouraging prosocial behaviour on social media

  Nature Human Behaviour · 2025-03-10 · 2 citations

  reviewSenior authorCorresponding
  PublisherDOI

• A Design-based Solution for Causal Inference with Text: Can a Language Model Be Too Large?

  ArXiv.org · 2025-10-09

  preprintOpen access

  Many social science questions ask how linguistic properties causally affect an audience's attitudes and behaviors. Because text properties are often interlinked (e.g., angry reviews use profane language), we must control for possible latent confounding to isolate causal effects. Recent literature proposes adapting large language models (LLMs) to learn latent representations of text that successfully predict both treatment and the outcome. However, because the treatment is a component of the text, these deep learning methods risk learning representations that actually encode the treatment itself, inducing overlap bias. Rather than depending on post-hoc adjustments, we introduce a new experimental design that handles latent confounding, avoids the overlap issue, and unbiasedly estimates treatment effects. We apply this design in an experiment evaluating the persuasiveness of expressing humility in political communication. Methodologically, we demonstrate that LLM-based methods perform worse than even simple bag-of-words models using our real text and outcomes from our experiment. Substantively, we isolate the causal effect of expressing humility on the perceived persuasiveness of political statements, offering new insights on communication effects for social media platforms, policy makers, and social scientists.

  PublisherOA PDFDOI

• Towards global equity in political polarization research

  2025-04-16 · 1 citations

  preprintOpen access

  With a folk understanding that political polarization refers to socio-political divisions within a society, many have proclaimed that we are more divided than ever. In this account, polarization has been blamed for populism, the erosion of social cohesion, the loss of trust in the institutions of democracy, legislative dysfunction, and the collective failure to address existential risks such as Covid-19 or climate change. However, at a global scale there is surprisingly little academic literature which conclusively supports these claims, with half of all studies being U.S.-focused. Here, we provide an overview of the global state of research on polarization, highlighting insights that are robust across countries, those unique to specific contexts, and key gaps in the literature. We argue that addressing these gaps is urgent, but has been hindered thus far by systemic and cultural barriers, such as regionally stratified restrictions on data access and misaligned research incentives. If continued cross-disciplinary inertia means that these disparities are left unaddressed, we see a substantial risk that countries will adopt policies to tackle polarization based on inappropriate evidence, risking flawed decision-making and the weakening of democratic institutions.

  PublisherOA PDFDOI

• Towards global equity in political polarization research

  ArXiv.org · 2025-04-15 · 1 citations

  preprintOpen access

  With a folk understanding that political polarization refers to socio-political divisions within a society, many have proclaimed that we are more divided than ever. In this account, polarization has been blamed for populism, the erosion of social cohesion, the loss of trust in the institutions of democracy, legislative dysfunction, and the collective failure to address existential risks such as Covid-19 or climate change. However, at a global scale there is surprisingly little academic literature which conclusively supports these claims, with half of all studies being U.S.-focused. Here, we provide an overview of the global state of research on polarization, highlighting insights that are robust across countries, those unique to specific contexts, and key gaps in the literature. We argue that addressing these gaps is urgent, but has been hindered thus far by systemic and cultural barriers, such as regionally stratified restrictions on data access and misaligned research incentives. If continued cross-disciplinary inertia means that these disparities are left unaddressed, we see a substantial risk that countries will adopt policies to tackle polarization based on inappropriate evidence, risking flawed decision-making and the weakening of democratic institutions.

  PublisherOA PDFDOI

• Designing Social Media to Promote Productive Political Dialogue on a New Research Platform

  2025-06-03

  preprintOpen access

  Social media are frequently blamed for exacerbating ideological divides. We evaluate whetherit is possible to design a social media platform that instead promotes productive, open-mindeddialogue among users. We recruited 1,043 Americans who expressed willingness to test a newsocial media platform by installing a mobile app we developed for iOS and Android devices.Participants were randomly assigned to spend ten minutes on one of two versions of a simulatedsocial media platform. The first was a control condition designed to resemble typical text-basedsocial media sites, and users were told that people whose posts are liked very often receivea “popular user” badge. The second version was identical to the first, but respondents wereinstead told users receive an “open-minded” badge if they respectfully engage with opposingviewpoints. Participants using both versions were asked to join an ongoing discussion abouteither gun control or immigration. Unbeknownst to them, each participant only interactedwith chatbots powered by generative AI. We find that respondents assigned to the experimentalcondition exhibited more intellectually humble language in their posts and comments andreported experiencing more positive emotions. We do not find significant treatment effects onparticipants’ self-reported intellectual humility, affective polarization, or issue attitudes. Theseresults have important implications for future research on social media and social psychology andintroduce new methods for conducting randomized controlled trials using generative artificialintelligence.

  PublisherOA PDFDOI

Recent grants

• Civil Society Organizations, Social Media, and Public Attention

  NSF · $77k · 2015–2016

• A Mixed Method Field Experiment Using Social Media

  NSF · $224k · 2018–2019

• Civil Society Organizations, Social Media, and Public Attention

  NSF · $144k · 2014–2015

Frequent coauthors

• Alexander Volfovsky

  Duke University

  14 shared

• Terence E. McDonnell

  University of Virginia's College at Wise

  14 shared

• Jennifer C. Lena

  Columbia University

  13 shared

• Ann Mische

  12 shared

• Iddo Tavory

  12 shared

• Margaret Frye

  12 shared

• Omar Ližardo

  11 shared

• Friedolin Merhout

  11 shared