Study finds Republicans flagged for posting misleading tweets twice as often as Democrats on Community Notes

Lisa Lock
scientific editor

Robert Egan
associate editor

New research from the Oxford Internet Institute at Oxford University, in partnership with researchers at Panthéon-Sorbonne and the MIT Sloan School of Management, reveals partisan differences in which posts get flagged as misleading by the social media platform X's Community Notes program. The study's findings show that misinformation is shared disproportionately across political divides, but this imbalance isn't a result of biases among the users rating the Community Notes on X.
The researchers analyzed crowd-sourced assessments from X's Community Notes program, where users on X can flag posts as misleading and write notes explaining why. These notes are then evaluated by other users, and if enough people agree on their helpfulness, they appear alongside the original post.
Their analysis, in the Proceedings of the National Academy of Sciences, shows that the Community Notes program identifies twice as many misleading tweets from Republicans as from Democrats in their sample.
"Our findings provide strong evidence of a partisan asymmetry in misinformation sharing which cannot be attributed to political bias on the part of fact-checkers or academic researchers," said Professor Mohsen Mosleh, associate professor of social data science at the Oxford Internet Institute, part of the University of Oxford.
"Even on Elon Musk's X, the user-based Community Notes program flags posts by Republicans as misleading much more often than posts by Democrats. This undercuts the logic offered by Musk and Mark Zuckerberg for eliminating fact-checkers on X and Meta, respectively, namely that fact-checkers are biased against Republicans."
Examples of Community Notes included:
Original Tweet: "A candidate must win the popular vote to become President in the U.S."
Community Note: While the popular vote reflects the total votes nationwide, the U.S. presidential election is decided by the Electoral College. A candidate can win the presidency without winning the national popular vote, as occurred in 2000 and 2016.
Original Tweet: "GTA 6 is releasing next week according to this leaked footage!"
Community Note: This video is not official and has been confirmed as fan-made content. Rockstar Games has not announced an official release date for GTA 6.
Original Tweet: (Image of a flooded city with the caption: "This is New York after yesterday's rainstorm.")
Community Note: This image is from a 2015 flood in Jakarta, Indonesia. It has been digitally altered and is not related to recent weather in New York.
The researchers also analyzed the topics covered in Community Notes and find substantial partisan differences in the proportion of tweets rated as helpful on the Community Notes program across the different topics. Republican users were overrepresented in misinformation sharing on health, with 81% of health-related flagged posts written by Republicans, followed by politics, 73%; science, 68%; other topics, 66%; and the economy, at 63%.
"Our results demonstrate that even Community Notes, where diverse groups of users have to agree in order for posts to be flagged, finds far more misleading content being posted by Republicans compared to Democrats," said co-author Thomas Renault, assistant professor, University of Paris 1 Panthéon-Sorbonne. "Our findings also indicate that Republicans will be sanctioned more than Democrats even if platforms transition from professional fact-checking to Community Notes."
The researchers analyzed a dataset of X's Community Notes focusing specifically on "misinformed or potentially misleading" notes written in English targeting English-language tweets on X sent between January 2023 and June 2024.
More information: Thomas Renault et al, Republicans are flagged more often than Democrats for sharing misinformation on X's Community Notes, Proceedings of the National Academy of Sciences (2025).
Journal information: Proceedings of the National Academy of Sciences
Provided by Oxford Internet Institute