Credit: Andrey Tikhonovskiy via Unsplash
The unreliable part of social media, including fake news reports, conspiracy theories and doctored images, may not be as nasty as some think.
Removing misinformation from social media feeds does not change what people think, a large-scale field experiment has found.
The study, published in Science Advances journal late July, reduced exposure to content from untrustworthy sources on Facebook and Instagram by about 70% among more than 15,000 users over three months.
The research found no measurable effect on beliefs about false claims, trust in mainstream media, attitudes or political polarisation.
The finding also held among users who had consumed the highest levels of misinformation content before the study began.
"Our results demonstrate that a feasible platform intervention can successfully reduce exposure to content from untrustworthy sources but suggest that these changes are unlikely to have immediate effects on attitudes and beliefs," the authors said.
The research, part of the US 2020 Facebook and Instagram Election Study, was conducted by a team of more than 25 academics from institutions including Dartmouth, Stanford, Princeton and New York University, working in collaboration with Meta researchers.
A second finding challenges an assumption that social media users are routinely exposed to high volumes of misinformation.
The study found that content from untrustworthy sources made up just 1.1% of what the median Facebook user saw. On Instagram, it was just 0.1%.
A small number of users were exposed to a heavy concentration of untrustworthy content.
Almost a quarter (23%) of Facebook users and 11% of Instagram users were responsible for 80% of all exposure to untrustworthy content on those platforms, representing 53 million and 22 million accounts respectively.
Among the top 2.5% of users by exposure level, content from untrustworthy sources made up 14% or more of what they saw on Facebook and 10% or more on Instagram.
Most exposure came from sources users had chosen to follow directly, rather than from algorithmic recommendations or reshared content.
The experiment ran from September 24 to December 23, 2020, over the US election period.
Untrustworthy sources were defined as Facebook Pages, groups and domains, and Instagram accounts that had received two or more strikes from Meta based on third-party fact-checker ratings of misinformation.
The treatment reduced average daily views of content from untrustworthy sources from 5.6 to 2.6 on Facebook, and from 22.9 to 10.6 on Instagram.
Despite those reductions, researchers found no statistically significant effects across ten preregistered outcome measures covering false beliefs, polarisation, trust in media and election attitudes.
The researchers note their findings have taken on new relevance following Meta's January 2025 decision to stop third-party fact-checking on Facebook and Instagram in favour of a community notes model.
That policy change means the strike system underpinning the experiment no longer operates, making a source-level intervention of the kind tested in the study no longer possible on Meta's platforms.
The authors said their findings provide limited guidance on the effects of removing fact-checking and warned that exposure to problematic content could increase considerably in the absence of formalised enforcement.
"Without credible scientific evidence about the effects of eliminating source-level penalties for misinformation, it is impossible for companies or policymakers to effectively evaluate the benefits and harms of social media platforms," the authors said.
The study was funded in part by Meta, which covered participant fees, recruitment and data collection costs.
Academic funding came from sources including the Democracy Fund, the Guggenheim Foundation and the Knight Foundation. The lead academic authors retained final editorial control and Meta could not block publication of results.
Several authors are current or former Meta employees.
Have something to say on this? Share your views in the comments section below. Or if you have a news story or tip-off, drop us a line at adnews@yaffa.com.au
Sign up to the AdNews newsletter, like us on Facebook or follow us on Twitter for breaking stories and campaigns throughout the day.