Facebook could help reduce animosity between Republicans and Democrats by driving its users to news outlets with political leanings that differ from their own, finds a new study published in the American Economic Review.
The study finds that even though people gravitate toward news that matches their political views, they are willing to read news articles featuring opposing views when they appear on their Facebook feeds.
Reading about differing views does not necessarily change a person’s stance on an issue, but it can change attitudes toward people who hold those other views, Ro’ee Levy, a postdoctoral associate at MIT, writes in the paper, “Social Media, News Consumption, and Polarization: Evidence from a Field Experiment.”
In fact, during the course of the two-month experiment, participants’ attitudes toward members of the opposing political party improved. Levy says the drop in “affective” polarization, or negative attitudes toward the opposing political party, surprised him and represent a reversal of the ever-deepening rift separating Republicans and Democrats that American National Election Studies surveys documented from 1996 to 2016.
“A small nudge was able to offset a substantial portion of the change in affective polarization over the past two decades,” he notes.
Levy’s research underscores the power that Facebook, a leading source of news for one-third of U.S. adults, has in reinforcing or changing public opinion. When self-described liberals saw news from traditionally conservative news organizations in their feeds, they sometimes clicked on those articles — and then read them, Levy says. The same held true for self-described conservatives exposed to news from liberal-leaning news outlets.
The results of this study differ from earlier research that found Republicans became more conservative when their Twitter feeds filled with messages reflecting opposing political ideologies. There are key differences between the two studies, however. For that study, published in PNAS in late 2018, people who identified as Republican were exposed to a steady flow of tweets from liberal elected officials, opinion leaders and others over a month.
When Levy asked the U.S. adults who participated in his Facebook experiment questions about how they felt about people holding opposing political views, those who had randomly visited news websites offering opposing views tended to express a more positive attitude toward those individuals. He chose eight leading news organizations — four liberal-leaning outlets such as MSNBC and four conservative-leaning ones such as Fox News — for the experiment.
“My theory for why attitudes change is that consuming news from the other side allows [people] to understand the other side better, even if they completely disagree with those positions,” Levy explains.
He adds that after reading news from a different perspective, people might say to themselves, “‘I might not agree with that, but [people holding other views] are not trying to hurt me. They have a different set of values or priorities.’”
Levy recruited participants in February and March 2018 via Facebook ads that offered participants a chance to win Amazon gift cards. The final sample comprised 37,494 adult Facebook users living in the U.S., 17% of whom identified as Republicans and 38% of whom identified as Democrats. Participants were 47 years old, on average, and just more than half were women.
After participants completed a survey that asked about their position on policy issues and their feelings toward people with opposing political views, Levy randomly assigned them to one of three groups — a liberal treatment group, a conservative treatment group and a control group.
He asked those in the liberal treatment group to like the Facebook pages of at least one but no more than four liberal-leaning news outlets. He asked those in the conservative treatment group to like between one and four conservative-leaning outlets. Members of the control group were not asked to like any news outlet Facebook page.
Because of the way Facebook’s algorithm works, Levy says, he knew that if participants liked an outlet’s page, its news coverage would begin to appear on the participant’s Facebook feeds.
The four outlets that Levy characterized as conservative are The Wall Street Journal, Fox News, The Washington Times and National Review. The four outlets he characterized as liberal are The New York Times, MSNBC, Slate and HuffPost. Levy also offered more than a dozen other leading conservative- and liberal-leaning news outlets as alternate choices, in case participants had already subscribed to those provided as primary choices.
He left participants to decide on their own, without prompting, whether to read the news outlets’ posts, share the posts, click on links within those posts or unsubscribe from the outlets. Almost 2,300 participants agreed to install a browser extension, allowing Levy to collect data from their Facebook feed and on their news-related browsing behavior in exchange for a small reward. These individuals could choose to receive a $5 gift card, enter into a lottery for a $200 gift card or get a copy of the study results.
At the end of the experiment, participants took another survey designed to measure changes in opinion about public policies and members of the opposing political party. They were asked questions such as how they would feel if one of their children married someone from the opposing party, how difficult it is to see things from each party’s point of view and how warmly they feel toward each party.
Levy discovered that participants who had been asked to like “counter-attitudinal” websites — those providing a different perspective on political issues — had more positive attitudes about people holding opposing views.
“The effect of the social media feed on news consumption implies that any change to the feed, due to new subscriptions or a change in the algorithm, can drastically change one’s news diet,” he writes in his paper.
Levy notes that the short study period is one possible reason why participants’ opinions on policy issues did not change after consuming news from outlets with opposing political leanings. However, two months was long enough to change participants’ attitudes toward members of the opposing political party.
“Intuitively, participants may have learned some of the opposing party’s arguments and thus understood better why that party supports certain positions,” he writes. “This led to more positive attitudes but did not change political opinions as long as participants did not find these arguments particularly important.”
Levy says his findings indicate that social media companies and journalists can do more to reduce political polarization in the U.S.
Social media platforms could change their algorithms to expose their users to a wider array of perspectives, he argues.
“Suggestions include making algorithms more transparent, nudging users to diversify their feed, and modifying algorithms to encourage serendipitous encounters,” Levy writes. “The experiment described in this paper essentially measures the effect of one such intervention and shows that a simple scalable nudge can effectively diversify news exposure and decrease polarization.”
Meanwhile, he says, journalists should more fully explain Democrats’ and Republicans’ stances on controversial issues and the reasoning behind their positions. Research indicates newsrooms contribute to political polarization by focusing on extreme politics and featuring interviews with politicians with extreme views.
A study published in December 2019 in the Journal of Communication, “As Seen on TV? How Gatekeeping Makes the U.S. House Seem More Extreme,” finds that major network and cable TV news outlets have given the most airtime to members of Congress with the most extreme views, creating a perception there is greater division among elected leaders than actually exists.
According to an earlier study in Journalism & Mass Communication Quarterly, members of Congress with the most extreme views receive more attention from print media outlets than political moderates. Those researchers also found that far-right Republicans get more coverage than far-left Democrats.
Says Levy, “If outlets and journalists can at least put more emphasis on explaining the positions of the different sides, maybe that can decrease polarization. That seems like something that’s feasible. Just explain where they’re coming from.”
Expert Commentary