Some Real Data on Fake News

Mark Reinstein, Corbis via Getty Images

Fake news is rampant, and it’s poisoning our politics. Many people are saying this — on the right, on the left, and on whatever remains of the center.

But is that truism, itself, fake news?

Here’s the truth: Political scientists are only beginning to understand the mechanics of fake news and the havoc it has (or hasn’t) wrought on the American political system. Wounds are fresh, and research is scant. Andrew M. Guess, an assistant professor of politics and public affairs at Princeton University, is part of a team — which also includes Brendan Nyhan, of Dartmouth College, and Jason Reifler, of the University of Exeter — that has brought new research to bear on the question of how fake news spreads and whom it infects.

They did it by collecting web-traffic data from the computers of 2,525 willing citizens during the final month of the 2016 presidential campaign, plus a week after it ended. Their conclusion? About 10 percent of the people — mostly fans of Donald J. Trump, the surprise victor — accounted for nearly 60 percent of the visits to fake-news websites. And many people got to those sites by clicking links on Facebook.

The data are open to interpretation, of course. But as far as we can tell, at least, the researchers didn’t make it up.

The Chronicle talked with Mr. Guess about the study and what it means for politics and education. The interview has been edited for length and clarity.

Q. What is fake news? Am I fake news? Are you?

A. Ha! No. We try to define fake news as narrowly as possible — and also as removed as possible from the way that the term has become used in political discourse today, which is kind of as an all-purpose insult against news sources you don’t like. We tried to steer as far away as possible from any kind of semantic debates about whether something is hyperpartisan legitimate news that is slanted, versus just completely fabricated. And so we hope that the list of fake news sources that we came up with is as uncontroversial as possible.

Q. So, just to be clear, your study is not counting Breitbart or ThinkProgress articles as fake news.

A. That’s right.

Q. People call fake news an epidemic, and there is a sort of public-health flavor to the way that it’s been covered. What kinds of people are most at risk of being exposed to fake news and not knowing it?

A. We do find some clear patterns. And I think they probably fit into roughly two categories. One would just be people who just don’t have very high levels of online media literacy or web literacy. So, we see this suggestion of an “age effect” where regardless of which candidate they support, older people just seem to be consuming more fake-news content, or they’re more likely to come into contact with the fake-news article.

The other is a partisan component; if people supported Donald Trump they are more susceptible to being exposed to fake news. Whether that was because there’s just something about Trump supporters, or it was just the particular political conditions of the 2016 election, most of the news content was pro-Trump.

Q. I think when a lot of people think about fake news they think about ecosystems online that are populated by hyperliterate denizens of social media. But it sounds like you’re saying the people who are most at risk of being infected by fake news are the same people who would be most at risk of sending money to some email scammer saying that they’re a Nigerian prince.

A. To be clear, we can’t directly test whether this “literacy” explanation is really what’s driving the age effect. I also just want to be clear that I don’t think this is the only type of person that was coming into contact with fake news. There’s also just a pure partisan component in which the more traditional pattern of people with greater political knowledge or interest are just more likely to consume pro-attitudinal content in general, including fake news.

Q. How big is the problem? How many people are being misled by fake news, in your estimation?

A. We’re talking about roughly one in four Americans who visited a fake news website over this five-week period. That means they clicked on or consumed at least one article. That’s potentially a lot of people.

But then the question is, What does that mean? On the one hand, it seems significant that millions of people came into contact with blatantly false information, and they may or may not have believed what they saw, and it was in the context of a close and contentious election. On the other hand, if you think about all the other content that people are coming in contact with during the same period — and even through the same channels, such as Facebook — then it’s also easy to think about how this was just a drop in the bucket. So, I think it just depends on your perspective. Is one fake news article just too much? Or do we think that, compared to all the other stuff that people are getting, maybe the problem is not that severe?

Q. Can we know for sure how many minds an item of fake news might have changed?

A. Yeah, that’s the question everybody wants the answer to. We can’t answer how persuasive any given piece of news content was. When we were collecting this data, we had no idea that this thing was going to become such a prominent feature of the post-election narrative. And so we kind of realized we were sitting on a gold mine, but that also meant that we couldn’t ask these questions about whether people believed what they were seeing, or whether it was changing their minds contemporaneously.

Q. One of the more disheartening findings in this study is that the work of fact checkers rarely reaches the same people who read fake-news articles. Can you imagine any kind of intervention, beyond fact checking — some sort of antidote or vaccination, sticking with the public-health metaphor — that could be applied to the fake-news problem?

A. I think there are two types of approaches that are currently being tried, and I don’t think we know enough about whether they’ll work. One is you’re seeing these efforts by educators to try and boost online literacy and media literacy at the front end. So, when kids are still in school, trying to kind of inoculate them from future exposure to misinformation.

The other general approach is what Facebook is now trying to do: surfacing Snopes and other types of debunking-of-fake-news articles in the “related articles” segments when someone shares something. To me, that actually has a lot of potential.

Q. You didn’t study the specific exposure or vulnerability of college students to fake news. But you are an educator, and I’m curious about whether you think college students are more or less susceptible to fake news than the general population.

A. I’d certainly like to believe that a university education can provide people with the tools they need to counteract fake news, or to resist its allure. That said, we tested for an education effect, and we don’t specifically find one.

I think it’s just important to realize that anyone is potentially susceptible to misinformation. The reason fake news worked as a business model during the election campaign was that there was enough demand for it, and some of that demand comes from people who were kind of highly knowledgeable about politics and just have this insatiable thirst for information, even if it’s not true.

Q. And people of all ages, from retirees to college students, have very strong partisan feelings that might make them a little bit blinkered to the possibility that what they’re reading might not be true.

A. Yeah. Partisanship can affect anyone, and it can be a very powerful filter. And I think it’s difficult especially today, when the feelings that people have for their own party, and the other party — it’s just historically high levels. That overall environment seemed to be a precondition for the spread of this type of misinformation.

Q. Colleges are in the business of creating a reliable information record and disabusing people of incorrect beliefs. Can you see a role for higher education in fighting the scourge of fake news, whether that’s in the classroom or outside the boundaries of their campuses?

A. Yeah, that’s a really good question. I’m probably the worst person to ask about this because I specifically teach classes about politics and the media, so it’s easy for me to talk about these topics. One thing that people can do is kind of demystify journalism and the way it works, and give students an opportunity to get into the habit of consuming mainstream journalism and building that trust.

Q. I’m thinking of a professor at the head of the classroom telling students, “These are the news sources that I will accept as citations in your papers in this class, and here are some examples of ones I absolutely will not, and here’s why.” Is that the sort of work you’re talking about?

A. So, for example, last semester I asked all of my students to follow both a mainstream daily newspaper — The New York Times, The Washington Post, or The Wall Street Journal — and, in addition to that, a partisan publication. The idea is to give them some experience with what legitimate journalism from two different modes looks like, as compared to fake news or misleading information that they might encounter from their friends or relatives.

Q. When people talk about the effects of social media on the spread of fake news, they tend to think of all social media as variations on the same unreliable pathway to good information. Are all social-media platforms equally bad in terms of leading people to unreliable news sources?

A. We actually found that Facebook was by far the biggest driver of traffic to fake-news websites. We tested that against Google, Twitter, email, Reddit, and anything else that we could think of. It does seem that there’s something special about Facebook.

Q. What is it?

A. Well, we can’t test that directly, but I think there are a couple of important features that distinguish Facebook from, for example, Twitter. On Twitter, in theory everything you post is public, so there might be a reluctance to publicly share links to fake news articles. Maybe people are hesitant to be openly partisan. On Facebook, people are connected to the people who they know in real life that could be family members, close friends.

I do think that there is something important here with social ties that’s potentially driving the sharing and consumption of fake news. It’s sort of the dark side of these social ties.

Q. What about people who study politics, like yourself. Does the prevalence of fake news, and delivery systems like Facebook, change how academics should be looking at the political landscape?

A. I think there’s been an appreciation of the role of factual misperceptions in informing political beliefs. But in terms of the role of algorithms and social media and potentially supercharging some of these dynamics, the challenge is going to be how to study the determinants of the spread of misinformation. A lot of this stuff is happening on proprietary platforms, like Facebook, where academics don’t always have the kind of data access they might want.

Q. Do you think that somebody can study politics effectively if they are not studying what’s going on on Facebook?

A. I think as more and more people turn to Facebook and other social-media sites as primary source of information, it’s going to be even more important for all of us to appreciate their role in formulating people’s political beliefs. I think this is only going to get more important.

Steve Kolowich writes about ordinary people in extraordinary times, and extraordinary people in ordinary times. Follow him on Twitter @stevekolowich, or write to him at steve.kolowich@chronicle.com.