Bergeron-Boutin, O., Nyhan, B., Settle, J., Thorson, E., Wojcieszak, M., Brown, T., Crespo-Tenorio, A., Rivera, C. V., Allcott, H., Barberá, P., Dimmery, D., Freelon, D., Gentzkow, M., González-Bailón, S., Guess, A. M., Kennedy, E., Kim, Y. M., Lazer, D., Malhotra, N., Moehler, D., Pan, J., Thomas, D. R., Tromble, R., Wilkins, A., Xiong, B., de Jonge, C. K., Franco, A., Mason, W., Stroud, N. J., & Tucker, J. A. (2026). Untrustworthy sources on Facebook and Instagram in 2020: Concentrated exposure but no attitudinal effects. Science Advances, 12. https://doi.org/10.1126/sciadv.adz6502
Summary
This paper offers platform-scale empirical evidence on how often US adults encountered content from untrustworthy sources — outlets that repeatedly publish misinformation — on Facebook and Instagram during 2020, and whether that exposure moved political attitudes. Drawing on data covering the entire active US adult user base and pairing observational measurement with an experimental reduction of exposure, the authors argue that exposure was low on average, heavily concentrated among a small minority of users, and, when experimentally reduced, produced no detectable attitudinal effects. The work thus challenges the common assumption that widespread exposure to untrustworthy sources is a primary driver of attitudinal change.
Key Contributions
- Platform-scale empirical evidence on the frequency of exposure to untrustworthy sources across an entire national adult user base, rather than convenience or panel samples.
- A crucial distinction between average exposure and its concentration among specific users, sharpening the question of who actually encounters misinformation.
- Causal evidence — via experimental exposure reduction — on the limited attitudinal impact of untrustworthy-source content, informing debates on misinformation harms.
Methods
The authors analyze 2020 platform data covering all active US adults on Facebook and Instagram, measuring exposure to content from Pages, groups, and web domains on Facebook and public accounts on Instagram that were identified as repeatedly publishing misinformation. Alongside this observational measurement, an experimental/interventional component reduced exposure to such content in order to assess associated attitudinal effects.
Findings
- The average Facebook and Instagram user saw relatively little content from untrustworthy sources in their feeds during 2020.
- Exposure was highly concentrated among a small fraction of users rather than evenly distributed across the population.
- Experimentally reducing exposure to untrustworthy-source content produced no measurable effects on users’ political attitudes.
Connections
This paper sits squarely in the recalibration literature that pushes back on inflated estimates of misinformation’s reach and effects, aligning closely with Budak2024-ef on the modest prevalence and impact of misinformation exposure. Its emphasis on concentrated exposure among a small subset of users also complements work reconsidering how misinformation reach is estimated and interpreted, as in Hourigan2026-oc.
Podcast
A research-radio episode discusses this paper: 🎧 MP3 · Spotify