Publications

Testing the Causal Impact of Social Media Reduction Around the Globe

Steven Rathje, Nejla Asimovic, Tiago Ventura, Sarah Mughal, Claire E. Robertson, Christopher Barrie, Joshua A. Tucker, Jay J. Van Bavel, and the Global Social Media Team

Accepted in Principle at Nature, 2026

Pre-Registered Report

Abstract More than half of the world’s population uses social media. There is widespread debate among the public, politicians, and academics about social media’s impact on important outcomes, such as intergroup conflict and well-being. However, most prior research on the impact of social media relies on samples from the United States and Western Europe, despite emerging evidence suggesting that the impact of social media is likely to differ across the globe. Building on the results of pilot experiments from three countries (n = 894), we plan to conduct a global field experiment to measure the causal impact of reducing social media usage for two weeks across 23 countries (projected n > 8,000). We will then test how social media reduction influences four main outcomes: news knowledge, exposure to online hostility, intergroup attitudes, and well-being. We will also explore how the effects of social media reduction vary across world regions, focusing on three theoretically-informed country-level moderators: income level, inequality, and democratic strength. This large-scale, high-powered field experiment, and the global dataset resulting from it, will offer rare causal evidence to inform ongoing debates about the impact of social media and how it varies around the world.

Comparing the Humanness of Machine-Generated and Human-Authored Text

Autumn Toney-Wails, Leticia Bode, Tiago Ventura, Ethan Wilcox, and Lisa Singh.

Forthcoming at ACM Computing Surveys, 2026

PDF

Abstract As chatbots have become more commonplace writing tools, a need exists to understand the breadth of research about the humanness of machine-generated text via techniques that extend beyond the traditional Turing Test, in both dialogue (e.g., conversing with a chatbot) and non-dialogue (e.g., reading a news article) scenarios. To fill this gap and support future work, we synthesize current literature that examines and identifies humanness features of written communication generated with the state-of-the-art generative pre-trained transformer language models, provide a working definition of humanness, propose a text-based humanness taxonomy based on linguistic properties, and identify current research gaps

Misinformation Beyond Traditional Feeds: Evidence from a WhatsApp Deactivation Experiment in Brazil

Tiago Ventura, Rajeshwari Majumdar, Jonathan Nagler, Joshua A. Tucker

The Journal of Politics, 2025

PDF Pre-Registration Replication Materials Cite

Winner, 2024 Paul Lazarsfeld Best Paper Award (Political Communication, APSA 2023) | Winner, 2024 Information Technology and Politics Award (APSA 2023) | Winner, 2024 Best Paper in Political Behavior (Brazilian Political Science Association) | Winner, 2024 Best Overall Paper (Brazilian Political Science Association)

Abstract Most advanced democracies associate misinformation with feed-based platforms (Twitter, Facebook); but Global South countries experience misinformation via messaging apps like WhatsApp. We deploy a multimedia deactivation experiment testing misinformation exposure reduction on WhatsApp before the 2022 Brazil Presidential election. Our intervention significantly reduced false rumors recall. Consistent with mass media minimal effects theories, short-term information environment change didn’t significantly alter belief accuracy, political polarization, or well-being.

Survey Professionalism: New Evidence from Web Browsing Data

Bernhard Clemm von Hohenberg, Tiago Ventura, Jonathan Nagler, Ericka Menchen-Trevino, Magdalena Wojcieszak

Political Analysis, 2025

PDF Preprint Replication Materials Cite

Abstract Online panels are important for political science research, but panelist compensation incentivizes “survey professionals” raising data quality concerns. We explore three US samples donating browsing data via Lucid, YouGov, Facebook (n=3,886). Survey professionalism is common but varies: conservatively 1.7% Facebook, 7.9% YouGov, 34.3% Lucid identified as professionals. Evidence professionals produce lower quality is limited: no systematic demographic/political differences from non-professionals; no more response instability. They are more likely to speed, straightline, and attempt to repeat questionnaires. We conclude professionals don’t distort research inferences despite warranted concerns.

Keep Your Promises, Even When Your Peers Do Not: A Survey Experiment on the Influence of Social Media on Trust

Natalia Aruguete, Ernesto Calvo, Carlos Scartascini, Tiago Ventura

Journal of Information, Technology and Politics, 2025

PDF Pre-Registration Cite

Abstract Two survey experiments (~2,300 respondents each) in Brazil and Mexico measuring the effect of partisan/polarizing social media messages on political trust and trustworthiness using a modified trust game. Results show a statistically significant decline in trust among users exposed to polarizing messages (belief others won’t fulfill pledges) with null effect on trustworthiness (we keep our pledges). The decline is larger if respondents actively like, share, or comment on the message, highlighting active engagement as a mediator in trust diminishment.

The Fact-Checking Dilemma: Fact-Checking Increases the Reputation of the Fact-Checker but Creates Perceptions of Ideological Bias

Tiago Ventura, Ernesto Calvo, Natalia Aruguete

Research & Politics, 2025

PDF Pre-Registration Replication Materials Cite

Abstract Pre-registered survey experiment during the 2021 mid-term Argentina election when COVID-19 was polarizing (n=5,757). We exposed respondents to real tweets with COVID-19 case numbers followed by fact-checking adjudications confirming or refuting. Pro-attitudinal messages increased the fact-checker’s (Chequeado) quality rating and perceived ideological closeness. Counter-attitudinal messages also increase perceived quality with no ideology effect. Fact-checks are reputation-improving and non-backfiring. However, the intervention affects voter perception of the fact-checking organization’s ideology.

Voting for Law and Order: Evidence from a Survey Experiment in Mexico

Tiago Ventura, Sandra Ley, Francisco Cantu

Comparative Political Studies, 2024

PDF Pre-Registration Replication Materials Cite

Abstract Examines the demand-and-supply dynamic of security policies via two informational shortcuts: personal violence experiences and candidates’ profiles. We test our argument via a survey experiment in Mexico modeling voters’ support for candidates with various security proposals and use recent network model developments to measure crime exposure effects. We find higher crime victimization is associated with support for only some iron-fist policies. Results show null partisan advantage effects but reveal non-partisan heuristics (candidate professional experience) play a role in security policy preferences.

Framing Fact-Checks as a “Confirmation” Increases Engagement with Corrections of Misinformation: A Four-Country Study

Natalia Aruguete, Flavia Batista, Ernesto Calvo, Matias Guizzo-Altube, Carlos Scartascini, Tiago Ventura

Nature: Scientific Reports, 2024

PDF Pre-Registration Cite

Abstract Four-country survey experiment assessing the influence of confirmation/refutation frames on engagement with online fact-checks. Respondents randomly received semantically identical content affirming accurate information or refuting misinformation. Despite semantic equivalence, confirmation frames elicit higher engagement and reduce self-reported negative emotions related to polarization. Crucial for designing policy interventions to amplify fact-check exposure and reduce affective polarization.

Training Computational Social Science PhD Students for Academic and Non-Academic Careers

Aniket Kesari, Jae Kim, Sono Shah, Taylor Brown, Tina Law, Tiago Ventura

PS: Political Science & Politics, 2023

PDF Cite

Abstract A guide to CSS training for PhD students based on the literature and collective working experiences. We contend students should supplement traditional training with CSS training in three core areas: (1) learning data science skills, (2) building a portfolio using data science for social science questions, (3) connecting with computational social scientists. We also offer recommendations for departments and professional associations.

Truth Be Told: Cognitive Moderators of Selective Sharing of Fact-Checks on Social Media

Natalia Aruguete, Ingrid Bachmann, Ernesto Calvo, Sebastian Valenzuela, Tiago Ventura

New Media & Society, 2023

PDF Cite

Abstract Survey experiment (2019 Argentina election) measuring the propensity to share corrections to political misinformation randomly confirming or challenging beliefs. We find evidence of selective sharing: individuals prefer sharing pro-attitudinal rather than counter-attitudinal fact-checks. The effect is conditioned by adjudication type; respondents report higher intent to share confirmations vs refutations. Findings supported with RDD analysis of Twitter data and additional experiments.

Winning! Election Returns and Engagement in Social Media

Tiago Ventura, Ernesto Calvo, Silvio Waisbord, Natalia Aruguete

PLOS One, 2023

PDF Replication Materials Cite

Abstract Analyzes social media engagement when elections are adjudicated to one contending party. We extend models of political dialogue explaining differences in engagement (time-to-retweet) when users support the winner or losers. We show users supporting the winner are more engaged with lower time-to-retweet. RDD design with data from Argentina (2019), Brazil (2018), UK (2019), and US (2016).

Network Activated Frames: Content Sharing and Perceived Polarization in Social Media

Natalia Aruguete, Ernesto Calvo, Tiago Ventura

Journal of Communication, 2022

PDF Cite

Abstract Describes how the sharing behavior of interconnected users alters the content frequencies observed by peers. We label this “Network Activated Frames (NAF)” and test its mechanisms with an original image-based conjoint design replicating network activation in three surveys. Results show partisans share more content than non-partisans, and preferences become over-represented. A network of peers with cross-cutting ideological preferences amplifies disproportionate partisan frames. Implemented in Argentina, Brazil, and Mexico.

News by Popular Demand: Ideology, Reputation, and Issue Attention in Social Media News Sharing

Natalia Aruguete, Ernesto Calvo, Tiago Ventura

International Journal of Press/Politics, 2021

PDF Replication Materials Cite

Abstract Retrieves measures of ideological congruence, issue salience, and media reputation explaining news sharing in social media. We describe how our proposed model connects to the news sharing literature. We show that if ideology and salience correlate, ideologues’ preferences become overrepresented in observational data causing heightened polarization perceptions. Tests with data from Brazil, Argentina, and the US.

The Effect of Streaming Chat on Perceptions of Debates

Victoria Asbury, Keng-Chi Chang, Katherine McCabe, Kevin Munger, Tiago Ventura

Journal of Communication, 2021

PDF Pre-Registration Slides Cite

Abstract Field experiment during the September 2019 Democratic Primary Debate. Subjects were assigned to view the debate with or without streaming chatboxes on ABC (no chatbox), FiveThirtyEight (expert chat), and Facebook (social chat). Democratic subjects in the Facebook chat condition reported lower affect towards Democrats and a worse viewing experience.

Connective Effervescence and Streaming Chat During Political Debates

Tiago Ventura, Kevin Munger, Katherine McCabe, Keng-Chi Chang

Journal of Quantitative Description: Digital Media, 2021

PDF Cite

Abstract Examines large samples of comments in social chat feeds during livestreamed 2020 US presidential and vice-presidential debates on ABC News, NBC News, and Fox News Facebook pages. We quantify quality features of political discussion. Results consistent with the quasi-anonymous constrained chat nature: comments are generally short, include substantial toxicity and insults, and differ significantly across platforms. Underscores the importance of studying streaming chat as a potential influence source on political attitudes and behavior.

Partisan Cues and Perceived Risks: The Effect of Partisan Social Media Frames During the Covid-19 Crisis in Mexico

Natalia Aruguete, Ernesto Calvo, Francisco Cantu, Sandra Ley, Carlos Scartascini, Tiago Ventura

Journal of Elections, Public Opinion and Parties, 2021

PDF Pre-Registration Cite

Abstract Survey experiment evaluating effects of social media exposure on health and job risk perceptions during COVID-19 in Mexico. A framing experiment with positive/negative partisan messages shows respondents are sensitive to negative frames regardless of messenger color; incumbent supporters deflect government responsibility when exposed to opposition negative frames.

Do Mayors Matter? Reverse Coattails on Congressional Elections in Brazil

Tiago Ventura

Electoral Studies, 2020

PDF Replication Materials Cite

Winner, Donald C. Piper Award for Best Graduate Student Research Paper, GVPT, University of Maryland

Abstract In federal democracies, parties invest in local politics to improve upper-level performance. Uses the reverse coattails concept to investigate the effects of winning local elections on national electoral dynamics in Brazil. RDD shows parties boost national performance by earning more votes on House elections in districts where members control local offices. Discusses pork access controlled by co-partisan House members and mechanical information gains. Uses Bayesian LASSO to address RDD data sparsity; demonstrates pro-large party bias on coattail effects.

Will I Get Covid? Partisanship, Social Media Frames, and Perceptions of Health Risk in Brazil

Ernesto Calvo, Tiago Ventura

Latin America Politics and Society, 2020

Pre-Registration Replication Materials Cite

Abstract Survey with embedded social media experiment during the early COVID-19 pandemic in Brazil. Descriptive results show pro-government and opposition partisans report different health and job risk expectations; job/health policy became wedge issues. Exploits survey recruitment random variation to show Presidential speech effects on TV on risk perceptions and partisanship moderation. Framing experiment models cognitive mechanisms driving partisan differences in health risk and job security perceptions during the COVID crisis.

Polarization, News Sharing, and Gatekeeping: A Study of the Bolsonaro Election

Natalia Aruguete, Ernesto Calvo, Tiago Ventura

Digital Journalism, 2020

PDF Cite

Abstract Examines how news sharing changes gatekeeping preferences of news organizations. We model users’ news sharing behavior using Twitter data and test using data from the 2018 Brazil Bolsonaro election, examining if polarized users polarize news organizations further.