Here is a fact about Facebook: Once largely used to share family photos and cat memes, it’s now a platform where people encounter political news with varying levels of accuracy.
And here is some fake news about Facebook: It is responsible for two of the most surprising political results of recent times, Brexit and the election of Donald Trump.
The desire to ascribe this power to Facebook is understandable. After all, it may be comforting to blame a Big, Bad Tech Company for political outcomes one finds unpleasant.
But is Facebook really that persuasive?
It would be easy (and tempting) to look to media coverage and determine that Facebook is not only powerful but also that the information people obtained on the social network may have swung the 2016 election. But despite headlines that suggest that the influence of Facebook can be clearly and simply identified, there are considerable challenges to connecting what people see on Facebook with their political opinions and behaviors.
The difficulty with studying whether Facebook changes people’s political behavior is that we can’t just ask them about it. If a person tells a researcher that he saw a lot of pro-Trump information on Facebook and that he’s a strong support of Donald Trump, the causal connection isn’t clear. In other words, this person could have become a supporter of Trump after encountering the pro-Trump information on Facebook. He also could have seen and believed the Facebook information because he was already a Trump supporter.
Facebook…is influential in some ways (like making some stories seem more important than others) but may not change political behavior.
There are other challenges in studying Facebook’s influence. People, research suggests, have difficulty remembering their own media habits. And people also have a tendency for what scholars call “expressive survey responding,” which means a person who says he fully believes everything positive he’s ever heard about Trump could be trying to make a larger political point.
Understanding what role (if any) Facebook plays in political persuasion, then, is a difficult process. Studying Facebook’s potential for influence means monitoring what people see on their feeds and then analyzing whether this information changes people’s political thoughts and opinions. A recent study published in the peer-reviewed political science journal Political Research Quarterly attempts to do just that. Facebook, this new study finds, is influential in some ways—like making some stories seem more important than others—but may not change political behavior.
To consider if (and how) seeing information on Facebook affects people, one of us (Feezell) designed an experiment that leveraged Facebook’s groups feature. When users join a group, the administrator of that group can post information directed to all group members and track which members see the information on their feeds.
In the study, this feature was used to create two groups: a political group, which received some political news, and a nonpolitical news group, which received no political news.
Students at a university received invitations to join one of the groups, but these invitations were random: Some students were invited to join the political group, but others were invited to join the nonpolitical group.
Over the course of 75 days, both groups received news stories. Interspersed among stories about technology and science, the political group got news about current issues at the time—immigration, climate change, Ukraine, and local crime, all from a variety of mainstream sources.
The nonpolitical news group also received the same set of stories about technology and science, but instead of the additional political stories they got stories that had no explicit political content (e.g., this story on handwriting).
Both groups received about 50 stories, about one story every one to two days.
To gauge how Facebook influences opinions about the news, Feezell conducted two surveys with participants: one before they joined the Facebook group and one after their 75 days as a group member.
In both surveys, people answered questions about the types of political issues they found important and interesting, including topics covered in the articles sent to the political group.
While Facebook could be setting the agenda for our national conversation, it is probably not causing erstwhile Clinton supporters to buy red #MAGA baseball caps or leading Trump supporters to pledge their support to the Democratic party.
Before joining the Facebook groups, people in both the political group and the nonpolitical group had similar opinions about which political issues were and were not important. Many of the respondents in both groups did not think immigration, climate change, Ukraine, and local crime were important issues. By the time of the second survey, of those people in the nonpolitical group who didn’t think any of the issues were important, 29 percent shifted to believing that at least one these issues were important. Even though these people didn’t see the information as part of the Facebook group, it is reasonable to assume they encountered it somewhere in their day-to-day lives given that these were big news stories during the period of study.
But in the political group, of people who didn’t think the issues were important in the first survey, 43 percent now thought these issues were important. Presumably, seeing these stories on Facebook led these people to believe that these were the most important issues of the day at greater rates than people would have had they not been part of a political Facebook group.
The effects were most striking among people who had reported they had relatively little interest in politics. Of those with little interest in politics who saw the political stories in their feed, 46 percent changed their mind and reported that these were now the most important stories of the day. In contrast, of those with little political interest who didn’t see the political stories, only 24 percent changed their minds about the importance of the political stories. There is no such difference between the group with high political interest: Seeing political stories in their newsfeeds has little discernible effect on whether people who are interested in politics perceive these stories to be important.
This is evidence of a process that scholars call “agenda setting”—the idea that new information can affect what people believe is a salient issue.
So is Facebook somehow responsible for what people think and feel about politics?
Facebook is not unusual in its capacity to suggest to people which issues they should find important. Agenda setting has existed for decades, long before Mark Zuckerberg started the social network. For instance, the stories producers chose to include in the nightly news routinely determined what Americans believed to be the key issues of the day.
But agenda-setting is not evidence of opinion change: A person can believe an issue is more or less salient to politics without changing their position on it.
Though Feezell’s study gives us reason to believe Facebook can set the agenda, it gives us no evidence that the effect of the information people encounter on Facebook is any different than the effect of the information you might see on the nightly news.
Though this study gives us reason to believe Facebook can set the agenda, it gives us no evidence that the effect of the information people encounter on Facebook is any different than the effect of the information you might see on the nightly news.
This view of Facebook’s influence as limited is supported by other work. A recent study by Andy Guess, Brendan Nyhan and Jason Reifler, for example, shows that while fake news was prevalent on Facebook, the people most likely to encounter and read fake news were people who already had strong political positions. Even more broadly, decades of political science research suggest that true reversals of political opinions are rare.
So while Facebook could be setting the agenda for our national conversation, it is probably not causing erstwhile Clinton supporters to buy red #MAGA baseball caps or leading Trump supporters to pledge their support to the Democratic party.
This distinction between agenda setting and opinion change means that it is difficult to draw a clear link between what people see on Facebook and their ultimate political opinions and behaviors. But complexity is to be expected. And how people come to form the political opinions that eventually guide their behavior has been a topic of analysis for decades.
Opinion formation is a nuanced interaction between things like people’s geographic and social contexts, the political cues they receive, their political knowledge and the types of opinion questions they are asked (to name just a few). What’s more, the relationship between opinion and new information is likely cyclical due to a long-studied process scholars term “selective exposure”: People are more likely to focus on and seek out information that supports their existing opinions. As the Guess, Nyhan, and Reifler study suggests, rather than only asking how Facebook affects opinion, a better question may be to ask how people’s opinions influence what they see on Facebook.
Facebook may be powerful and, as Mark Zuckerberg’s experience on Capitol Hill suggests, Facebook probably knows a good deal about the likes and dislikes of its users. But to understand how people come to have likes and dislikes in the first place, we cannot focus on Facebook alone.