Research

Published Papers

The consistency principle: Crisis perceptions, partisanship and public support for democratic norms in comparative perspective
European Journal of Political Research. 2025. [Paper]
With Amanda Driscoll, Jay Krehbiel, and Michel J. Nelson

A growing body of research theorizes that partisanship can undermine democracy as citizens prioritize their political interests over abstract norms and values. We argue that crises might counteract intense partisanship by giving citizens clarity on the threats posed by rule of law violations. Examining the differential application of a law – a breach of democratic norms – we draw on an experiment embedded in representative surveys of Germany, the United States, Hungary and Poland to examine citizens’ sense of appropriate punishment for elites’ violation of a municipal mask-wearing ordinance. We find evidence of partisan bias in citizens’ willingness to support punishment in all four countries. But, in the two consolidated democracies, we find that concern about the Covid-19 crisis diminishes partisan biases in punishment preferences: citizens who are most concerned about the crisis also model the most consistency in their willingness to hold copartisans into account.

Understanding political communication and political communicators on Twitch
PloS One. 2024. [Paper]

As new technologies rapidly reshape patterns of political communication, platforms like Twitch are transforming how people consume political information. This entertainment-oriented live streaming platform allows us to observe the impact of technologies such as “live-streaming” and “streaming-chat” on political communication. Despite its entertainment focus, Twitch hosts a variety of political actors, including politicians and pundits. This study explores Twitch politics by addressing three main questions: 1) Who are the political Twitch streamers? 2) What content is covered in political streams? 3) How do audiences of political streams interact with each other? To identify political streamers, I leveraged the Twitch API and supervised machine-learning techniques, identifying 574 political streamers. I used topic modeling to analyze the content of political streams, revealing seven broad categories of political topics and a unique pattern of communication involving context-specific “emotes.” Additionally, I created user-reference networks to examine interaction patterns, finding that a small number of users dominate the communication network. This research contributes to our understanding of how new social media technologies influence political communication, particularly among younger audiences.

Spatial modeling of dyadic geopolitical interactions between moving actors
Political Science Research and Methods. 2023. [Paper]
With Howard Liu and Bruce A. Desmarais

Political actors often interact spatially, and move around. However, with a few exceptions, existing political research has analyzed spatial dependence among actors with fixed geographic locations. Focusing on fixated geographic units prevents us from probing dependencies in spatial interaction between spatially dynamic actors, which are common in some areas of political science, such as sub-national conflict studies. In this note, we propose a method to account for spatial dependence in dyadic interactions between moving actors. Our method uses the spatiotemporal histories of dyadic interactions to project locations of future interactions—projected actor locations (PALs). PALs can, in turn, be used to model the likelihood of future dyadic interactions. In a replication and extension of a recent study of subnational conflict, we find that using PALs improves the predictive performance of the model and indicates that there is a clear relationship between actors’ past conflict locations and the likelihood of future conflicts.

The effects of an informational intervention on attention to anti-vaccination content on YouTube
Proceedings of the International AAAI Conference on Web and Social Media. 2020. [Paper]
With Omer F. Yalcin, Samuel E. Bestvater, Kevin Munger, Burt L. Monroe, and Bruce A. Desmarais

The spread of misinformation related to health, especially vaccination, is a potential contributor to myriad public health problems. This misinformation is frequently spread through social media. Recently, social media companies have intervened in the dissemination of misinformation regarding vaccinations. In the current study we focus on YouTube. Recognizing the extent of the problem, YouTube implemented an informational modification that affected many videos related to vaccination beginning in February 2019. We collect original data and analyze the effects of this intervention on video viewership. We find that this informational intervention reduced traffic to the affected videos, both overall, and in comparison to a carefully-matched set of control videos that did not receive the informational modification.

Working Papers

The political influence of non-politicized friends: How do social networks affect the spread of protest information in social media?
Under review

How do social networks influence the spread of protest information on social media? This article argues that the political characteristics of accounts sharing protest information affect how that information is interpreted and spread by other Twitter users. Specifically, I suggest that whether Twitter accounts are perceived as overtly political or nonpolitical can shape how users respond to signals about political protests. I hypothesize that nonpolitical accounts may exert more influence in spreading protest messages than political accounts, as they are seen as less biased or more trustworthy. To test this theory, I conducted an online experiment using vignettes that simulate the Twitter environment. Participants were exposed to protest-related Tweets and were asked whether they would retweet or like them, with some accounts presenting political traits in their profiles and others appearing nonpolitical. Contrary to my expectations, the results did not reveal a statistically significant difference in participants’ responses between political and nonpolitical profiles. However, the study revealed unexpected patterns, including the role of education in shaping retweet behavior differently across political groups and the influence of context-specific factors, such as protest types and images, on user engagement. These findings suggest that individual characteristics and content features may interact in complex ways, warranting further exploration.

Delayed takedown of illegal content on social media makes moderation ineffective
With Bao Tran Truong, Erfan Samieyan Sahneh, Gianluca Nogara, Enrico Verdolotti, Florian Saurwein-Scherer, Natascha Just, Luca Luceri, Silvia Giordano, and Filippo Menczer
Preprint arXiv:2502.08841 (2025), Under review [Paper]

Social media platforms face legal and regulatory pressures to moderate illegal content through takedown procedures. However, the effectiveness of content moderation varies widely across platforms due to differences in takedown deadlines imposed by various regulations. This study models the relationship between the timeliness of content removal and the persistence of illegal material on social media. By simulating illegal content diffusion using empirical data from sources like the DSA Transparency Database and Facebook NetzDG reports, we demonstrate that while rapid takedown (within hours) significantly reduces illegal content prevalence and exposure, longer delays (beyond 23 days) render moderation efforts futile. Our findings stress the need for regulatory frameworks with enforceable, short deadlines, such as those outlined in German law, to ensure meaningful content removal. These insights provide critical recommendations for policymakers aiming to enhance online safety and improve moderation strategies.

The rise of Bluesky
With Ozgur Can Seckin, Filipi Nascimento Silva, Bao Tran Truong, Fan Huang, Nick Liu, Alessandro Flammini, and Filippo Menczer
Preprint arXiv:2504.12902 (2025), Under review [Paper]

This study investigates the rapid growth and evolving network structure of Bluesky from August 2023 to February 2025. Through multiple waves of user migrations, the platform has reached a stable, persistently active user base. The growth process has given rise to a dense follower network with clustering and hub features that favor viral information diffusion. These developments highlight engagement and structural similarities between Bluesky and established platforms.

Can LLMs infer political affiliation from general discourse?
With Byunghwee Lee, Yong-Yeol Ahn, Filippo Menczer, Jisun An, and Haewoon Kwak

Political affiliation is easily inferred from explicitly political texts, but what happens when the discourse is about general topics, such as technology, sports, or music? Our study examines whether large language models (LLMs) can accurately infer political affiliations of individuals from general online discourse, given that such capability can be leveraged for micro-targeting, undermining people’s privacy. Using two datasets of user-generated text (Debate.org and Reddit) that contain political and general topics, we find that both GPT-4o and Llama-3.1-8B can reliably infer political affiliation even with only the posts in general domains. The prediction can be improved even more by aggregating multiple text-level inferences—especially when weighted by the self-reported confidence scores from LLMs. We also show that predictions using texts from categories with closer semantic proximity to a political category, as well as those with overlapping user participation, yield higher predictive performance. Finally, we demonstrate that inferring political affiliation from general discourse allows accurate prediction of stances of users on unseen political issues, pointing to the real-world implications of LLM-driven profiling. By highlighting the substantial privacy risk of large-scale inference on political affiliation and the potential for user data exploitation without explicit consent, our findings emphasize the need for stronger regulations and ethical guidelines in developing and deploying LLMs.

Inter-platform differneces in news-sharing behaviors around the U.S. 2022 midterm elections
With Ozgur Can Seckin, Kai-Cheng Yang, and Filippo Menczer

Social media platforms have become central channels for political news dissemination, yet they differ substantially in how they shape the circulation of such content. While concerns about the spread of low-credibility and highly partisan news are well documented, less is known about how these patterns vary across platforms with distinct design features and user cultures. In this study, we present a comparative analysis of political news-sharing behaviors during the 2022 U.S. midterm elections across five major platforms: Twitter, Facebook, Instagram, Reddit, and 4chan. Drawing on large-scale data collected throughout the election cycle, we examine the credibility, partisan leaning, demographic composition, and localness of shared news sources. Our findings reveal substantial inter-platform variation. Reddit features the highest proportion of credible and left-leaning domains, while 4chan exhibits the lowest credibility and a more ideologically balanced mix. Consistent with existing research, right-leaning domains are associated with lower credibility across all platforms. We also observe distinct demographic patterns in the primary audiences of shared domains, with Reddit and Instagram skewing toward more racially and gender-diverse audiences, and Facebook and 4chan featuring domains targeting older, white, and male audiences. Additionally, Facebook and Instagram show greater emphasis on local news compared to other platforms. These results highlight the importance of platform-specific affordances in shaping political information exposure and underscore the need for multi-platform research to fully understand the dynamics of news sharing in the contemporary social media environment.

Work in progress

Twitching, fast and slow: Field experiment in political stream
With Chloe Ahn, Drew Dimmery, and Kevin Munger

Sexism, Support for Violence, and Democratic support: Evidence from South Korea
With Boyoon Lee, Yoonseok Lee