Communications of the ACM
Social Media Has Small Impact on Americans' Beliefs, Researcher Says
Social media had only a small influence on how much people believed falsehoods about candidates and issues in the last two presidential elections, a pair of national studies found.
And Facebook—which came under fire for spreading misinformation in the 2016 campaign—actually reduced misperceptions by users in that election compared to those who consumed only other social media.
The results suggest a need to put the dangers of social media spreading misinformation in perspective, says R. Kelly Garrett, professor of communication at Ohio State University and author of "Social Media's Contribution to Political Misperceptions in U.S. Presidential Elections," published in the journal PLOS One.
"Given the amount of attention given to the issue, it may seem surprising that social media doesn't have a larger impact on Americans' belief in falsehoods," Garrett says. "It is an issue that we should be concerned about, but it is not the main driver of why so many people believe false information about issues and candidates."
In previous research, "Troubling Consequences of Online Political Rumoring" published in Human Communication Research, Garrett found evidence suggesting that email contributed to the spread of false information in the 2008 election, before social media was as popular as it is today. Garrett specifically designed these studies to gauge the role of social media in what Americans believed in the last two election campaigns.
Reliance on social media for political news has increased rapidly. In 2012, about two in five Americans reported using social media for political purposes, according to the Pew Research Center.
In 2016, more Americans named Facebook as the source they used for pre-election political information than any other site, including those of major news organizations, according to the newly published PLOS One study.
"This study began long before 'fake news' became as popular a topic as it is today. But the questions that drive this study are very much in keeping with our concerns about how disinformation is spread online," Garrett says.
During both the 2012 and 2016 election seasons, groups of more than 600 Americans filled out surveys online three times, indicating their social media use at each point as well as their beliefs in confirmed falsehoods.
The PLOS One study's examination of the 2012 election involved misperceptions about the two presidential candidates, Barack Obama and Mitt Romney. Participants rated on a five-point scale how much they agreed with eight falsehoods, including "Barack Obama is Muslim, not Christian" and "As governor of Massachusetts, Mitt Romney signed a health care law providing taxpayer-funded abortions."
Overall, Republicans tended to hold less accurate beliefs about President Obama than Democrats did, while Democrats held less accurate beliefs about Romney than did Republicans.
Results showed that increasing social media use reduced participants' "belief accuracy" about Obama falsehoods, although the effect was small.
In the most extreme case, someone using social media to get political information could have an accuracy score concerning Obama falsehoods almost half a point lower on the five-point scale than someone who did not use social media at all.
Social media use did not influence belief in the Romney falsehoods, Garrett says. One important reason may be that the rumors about Romney were much less known than those about Obama.
The PLOS One study's examination of the 2016 election focused on false beliefs about four campaign issues. The misperceptions studied were: Repealing the Affordable Care Act would reduce the national debt; Most Muslims support violence against Western countries, including the U.S.; Immigrants are more likely to commit violent crimes than individuals born in the U.S.; and Human activity has no influence on global climate.
After considering more than a dozen potential issues, Garrett selected these four because they were referenced most frequently on the campaign trail and received extensive media coverage, and because of evidence that Americans were at least occasionally mistaken about them.
Results show that, overall, Republicans beliefs tended to be less accurate than those of Democrats, which made sense because the falsehoods were a prominent part of the Republican campaign strategy, Garrett says.
Participants with higher levels of education held more accurate beliefs.
Unlike in 2012, participants surveyed on the 2016 election were asked which social media platforms they used during each of the three waves of the study. Facebook was the most popular social media platform for following news among study participants, followed by YouTube and Twitter.
Overall, social media use was not related to participants' belief accuracy on the four issues.
But the influence of using social media was different for people who used Facebook than for people who only used other platforms. Among the heaviest social media users, those who used Facebook were about a half point more accurate on the five-point scale, on average, than those who didn't.
"It is not a huge difference, but it does call into question the conventional wisdom that Facebook had an especially harmful influence on campaign issue beliefs," Garrett says. The small effects found in the study don't mean the problem of fake news shared on social media should be ignored, but its role in Americans' belief in falsehoods should be reevaluated, he says.
"We have evidence that foreign powers have tried to sway American elections by sharing falsehoods on social media and that is profoundly troubling. If it has an effect, even a small one, we should be deeply concerned," Garrett says.
"But we need to have a broader perspective on the problem. We know that Americans hold beliefs that are not accurate, with frightening regularity. And if social media aren't the primary driver of this, we really should invest more energy into finding out what else is going on."
The study was based on work supported by the U.S. National Science Foundation.
No entries found