Allcott, H., Gentzkow, M., Levy, R., Crespo-Tenorio, A., Dumas, N., Mason, W., Moehler, D., Barbera, P., Brown, T., Cisneros, J. C., Dimmery, D., Freelon, D., González-Bailón, S., Guess, A., Kim, Y. M., Lazer, D., Malhotra, N. A., Nair-Desai, S., Nyhan, B., de Queiroz, A. C. P., Pan, J., Settle, J., Thorson, E., Tromble, R., Velasco, C., Wittenbrink, B., Wojcieszak, M., Yang, S., Zahedian, S., Franco, A., de Jonge, C. K., Stroud, T., & Tucker, J. A. (2025). The effects of political advertising on Facebook and Instagram before the 2020 US election. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.5259653
Summary
This paper reports a large-scale randomized field experiment examining the causal effects of political advertising on Facebook and Instagram in the six weeks leading up to the 2020 US presidential election. Conducted as part of a collaborative research effort with Meta, the study randomly assigned subsets of 36,906 Facebook and 25,925 Instagram users to have political ads removed from their feeds. The intervention produced no detectable effects across a broad range of political attitudes, knowledge, and behaviors — though the authors are careful to note that the design was statistically underpowered to rule out small effects on turnout and vote choice specifically. The paper contributes rare experimental causal evidence to debates that have long relied on observational or smaller-scale data, and it complicates strong claims about the persuasive power of digital political advertising.
Key Contributions
- Delivers large-scale experimental causal evidence on the effects of social media political advertising during a major US election.
- Documents descriptive patterns in presidential ad targeting — showing ads went largely to co-partisans — and in ad content, where fundraising appeals dominated.
- Informs platform-regulation and persuasion debates by demonstrating null effects on attitudinal outcomes.
- Establishes a methodological template of platform-collaborative randomized ad removal for studying advertising effects in situ.
Methods
- Randomized controlled experiment with 36,906 Facebook users and 25,925 Instagram users.
- Treatment condition removed political ads from participants’ news feeds for six weeks before the 2020 election.
- Descriptive analysis of presidential ad targeting patterns and content categories.
- Outcome measures spanned political knowledge, affective and issue polarization, political perceptions, turnout, and vote choice.
Findings
- Presidential ads were predominantly targeted at co-partisan supporters rather than persuadable voters.
- Fundraising appeals were the single most common type of political ad.
- No detectable effects of ad removal on political knowledge.
- No detectable effects on affective or issue polarization.
- No detectable effects on political perceptions.
- Effects on turnout and vote choice could not be ruled out due to limited statistical power.
Connections
This work sits within the broader study of digital media and elections, sharing an analytical focus on advertising and platform data with Gonzalez-Bailon2024-rq, which likewise examines exposure dynamics on Facebook. Its concern with how political actors deploy and target platform advertising connects to Votta2025-xz and Nenno2025-xa on political ad targeting and content. Most other papers under this topic address non-US or non-experimental contexts and are only loosely related.