During the US presidential election, many voters read the surprising news that Pope Francis had endorsed Republican Donald Trump. Others read the even more shocking revelation that the Clinton Foundation was tied to a pedophile sex ring operating out of a pizza parlor. Often these articles were shared via social media, and though the stories have been thoroughly debunked, Trump’s poll-defying victory forced the heads of technology companies to face accusations that they helped spread “fake news” about the candidates that swayed voters in Trump’s favor.
Facebook COO Sheryl Sandberg said she didn’t believe fake news swung the election, yet the company has since launched efforts to label false stories. Google CEO Sundar Pichai speculated that fake news could have changed the outcome because Trump’s margin of victory was “very narrow”—fewer than 100,000 votes in Wisconsin, Michigan, and Pennsylvania.
But while fake news on social media received a lot of attention, there’s reason to doubt it handed Trump the election. Fake articles would need to have been 36 times as persuasive as TV campaign ads to change the election results, according to research by Hunt Allcott of New York University and Matthew Gentzkow of Stanford University.
Most fake news favored Trump. The researchers aggregated three separate lists of fake-news articles to create a database of 156 stories that appeared in the three months before the election and were labeled false by reputable third-party sources. (In the study, they define fake news as “news stories that have no factual basis but are presented as facts.”) They find that 115 of the fake stories were pro-Trump/anti-Clinton—the researchers considered a story positive for one candidate if it was negative for the other—while only 41 favored Hillary Clinton, and that the average pro-Trump story was shared more on Facebook (30 million total shares, compared to 7.6 million for the pro-Clinton stories). The average fake headline was shared 386,000 times.
Despite the flurry caused by outlandish headlines, most voters considered social media—the primary, though not exclusive, distribution venue for fake news—less influential than more-traditional media outlets as a provider of election news. In an online postelection survey of 1,200 US adults, Allcott and Gentzkow find that only 14 percent named social media as their most important source of election news. Cable TV was the most valuable source for more than 23 percent of those surveyed, while more than 19 percent considered network TV their most important source.
The researchers also find that previous polls may have overstated how much fake news voters saw on social media. To test this idea, they created a set of plausible fake-news headlines that were not actually published. An example: “Leaked documents reveal that the Trump campaign planned a scheme to offer to drive Democratic voters to the polls but then take them to the wrong place.” (The researchers refer to these as “placebo” stories.)
In the survey, 15 percent of respondents said they recalled seeing the average fake news headline, and 8 percent said they believed it. Yet the results are statistically identical for the placebo headlines, suggesting that survey participants were dramatically overstating their recall. Allcott and Gentzkow estimate that just over 1 percent of people truly recalled seeing the average story.
Based on these findings and studies on TV campaign ads, the researchers write, “for fake news to have changed the outcome of the election, a single fake news story would need to have convinced about 0.7 percent of Clinton voters and non-voters who saw it to shift their votes to Trump, a persuasion rate equivalent to seeing 36 television campaign ads.”
These results may be a relief to tech-company CEOs, but the findings still indicate significant and troubling voter biases. Heavy social-media users were more likely to believe fake headlines, as were those with no higher education. Republicans were four to eight times as likely as Democrats to report believing pro-Trump headlines; Democrats, meanwhile, were 50–100 percent more likely than Republicans to believe pro-Clinton headlines. People whose social networks were considered politically segregated—with three-quarters or more of their friends preferring the same candidate—also were more likely to report seeing and believing fake news. If your Facebook feed is populated solely by those who share your political leanings, it may be time for a reality check.