A recent study has illuminated the ways in which Facebook and Instagram influenced, and did not influence, political knowledge, polarization, and behavior during the 2020 U.S. presidential election. By deactivating accounts for a portion of participants in the weeks leading up to the election, researchers were able to isolate the effects of these platforms on users’ political engagement and perceptions.
The study, published in the Proceedings of the National Academy of Sciences of the United States of America, offers the most extensive evidence to date on how access to social networking platforms impact attitudes and behaviors during a presidential election season.
Social media has been widely debated as a force affecting democracy, particularly in the United States. Concerns abound that these platforms might foster political polarization by creating ideological echo chambers. Others worry that social media might spread misinformation, thereby affecting voter knowledge and trust in the electoral process.
Additionally, the role of social media in either mobilizing or demobilizing voters, and its potential to sway election outcomes, remains contentious. The study aimed to provide empirical evidence to clarify these issues, focusing particularly on Facebook and Instagram due to their prominence among voters.
“There has been a longstanding debate about how social media platforms like Facebook and Instagram are affecting democracy,” said study author Matthew Gentzkow, Landau Professor of Technology and the Economy at Stanford University. “Experimental evidence on those questions remains limited. We had a unique opportunity here to test hypotheses about the impact of Facebook and Instagram on outcomes like knowledge, polarization, and perceived legitimacy of the election with richer data and at larger scale than had been possible before.”
The researchers conducted a randomized experiment involving 19,857 Facebook users and 15,585 Instagram users. Participants were regular users, engaging with the platforms for more than 15 minutes per day. They were divided into two groups: a treatment group that deactivated their accounts for six weeks before the election, and a control group that deactivated their accounts for just one week. This setup allowed the researchers to compare political outcomes between those who had restricted access to social media and those who did not.
A total of 27% of participants were in the treatment group, while the remaining 73% were in the control group. Participants were incentivized financially to comply with the deactivation periods. The study gathered data on various political outcomes through surveys, directly measured platform data from Meta (Facebook’s parent company), and external records such as voter turnout and campaign donations.
The researchers found mixed results regarding political knowledge. Facebook deactivation slightly reduced overall political knowledge, mainly due to decreased news knowledge. However, it increased participants’ ability to distinguish misinformation from factual news. This suggests that while Facebook exposes users to a lot of news, it also exposes them to misinformation, which can obscure overall knowledge gains.
“I was surprised that we had power to detect impacts on belief in misinformation,” Gentzkow said. “Prior studies — including our own — have struggled to detect such effects, given that most people on social media are exposed to relatively low doses of misinformation.”
Deactivating Facebook and Instagram did not significantly impact political polarization. Measures of both affective polarization (feelings toward political parties and their supporters) and issue polarization (opinions on key political issues) showed negligible changes. Similarly, there was no significant effect on participants’ perceptions of the election’s legitimacy, including beliefs about electoral fraud. This finding held despite the heightened focus on these issues during the 2020 election.
The study found that deactivating Facebook and Instagram reduced online political participation, such as signing petitions and posting about politics. However, it did not significantly impact voter turnout, suggesting that while social media might influence how people engage politically online, it does not necessarily translate to changes in actual voting behavior.
The researchers observed a slight reduction in support for Donald Trump compared to Joe Biden among participants who deactivated their Facebook accounts. This means that those who were not using Facebook in the weeks leading up to the 2020 U.S. presidential election were slightly less likely to report supporting Trump. However, this reduction was statistically insignificant, which means that the observed difference was small enough that it could have occurred by chance.
“This effect was not quite statistically significant, so we need to take it with a grain of salt,” Gentzkow said in a news release. “But if it’s real, it’s big enough that it could impact the outcome of a close election.”
Together, the findings indicate that “the impacts on political attitudes and beliefs were remarkably small,” Gentzkow told PsyPost. “Staying off Facebook and Instagram had little or no effect on people’s political views, their negative opinions of opposing parties, or beliefs around claims of election fraud.”
The study has some limitations that should be considered when interpreting the results. For instance, the study only examined the effects of short-term deactivation (six weeks), so the long-term impacts of social media use remain unclear.
“The time horizon we are looking at is just six weeks,” Gentzkow noted. “We’re looking at the impact of being on social media in the period leading up to the election. That’s an important question, but it’s not the same as asking how the world would have looked if Facebook and Instagram had never existed. For example, we are not ruling out the possibility that Facebook and Instagram contribute to polarization over longer time horizons.”
Future research could explore the long-term effects of social media deactivation to understand how sustained absence from these platforms might influence political outcomes. Studies could also investigate the impact of different types of content, such as high-quality news versus misinformation, to determine what specific elements of social media are most influential. Additionally, examining the effects on different demographic and political subgroups could provide a more nuanced understanding of how social media affects various segments of the population.
“We want to continue bringing rigorous experimental methods to bear on critical questions about the impact of social media and new technologies on democracy,” Gentzkow said.
The study, “The effects of Facebook and Instagram on the 2020 election: A deactivation experiment,” was authored by Hunt Allcott, Matthew Gentzkow, Winter Mason, Arjun Wilkins, Pablo Barberá, Taylor Brown, Juan Carlos Cisneros, Adriana Crespo-Tenorio, Drew Dimmery, Deen Freelon, Sandra González-Bailón, Andrew M. Guess, Young Mie Kim, David Lazer, Neil Malhotra, Devra Moehler, Sameer Nair-Desai, Houda Nait El Bar, Brendan Nyhan, Ana Carolina Paixao de Queiroz, Jennifer Pan, Jaime Settle, Emily Thorson, Rebekah Tromble, Carlos Velasco Rivera, Benjamin Wittenbrink, Magdalena Wojcieszak, Saam Zahedian, Annie Franco, Chad Kiewiet de Jonge, Natalie Jomini Stroud, and Joshua A. Tucker.