If you’re convinced your grandparents or crazy old Uncle Larry share a lot more dubious posts on Facebook, your intuition is likely correct. It’s primarily people over the age of 65, identifying as conservative or Republican, who shared the most fake news articles during the 2016 election, according to a new study published in .
But the overall number of people who shared fake news at all is actually very low: less than nine percent.
“In that sense, I’m happy to confirm what everybody thought,” said co-author Andrew Guess, a political scientist at Princeton University. “It’s always nice to have the data to back it up.”
And the age factor holds steady even after accounting for other factors, like political ideology or party affiliation, which surprised Guess and his co-authors. “I might have thought the age finding would be driven by other things, like older people being more likely to be conservative or Republican, but when we account for that, there’s still an independent effect of age,” he said.
Let’s take a moment to define “fake news,” since the term is so widely misused these days (deliberately or otherwise). We’re not talking about news from professional media outlets with facts or a message that tells people something they don’t want to hear, therefore they lob the term “fake news” to give them a reason to ignore it. Actual fake news is false or misleading information written to resemble a valid news article to fool unwitting readers. The researchers relied on a list of domains put together by Buzzfeed News journalist Craig Silverman, along with lists of stories debunked by fact-checking organizations, to classify links as fake news or not.
“We didn’t magically foresee that everyone would be talking about online misinformation after the election.”
The study started as a long-planned survey to uncover how people learned about the campaigns and the issues discussed by the candidates via social media. Over the course of the 2016 election, Guess and his collaborators at New York University conducted a panel survey with polling firm YouGov about what people shared on Facebook.
“We didn’t magically foresee that everyone would be talking about online misinformation after the election,” said Guess. “It didn’t become clear to us until later that we were sitting on a treasure trove [of data] that would help us answer questions about something everyone was talking about.”
Of the 3,500 Facebook users who took the survey, around 1,300, or 44 percent, also agreed to share their personal information with the researchers. That’s a surprisingly high response rate, but, at the time, according to Guess, privacy concerns surrounding Facebook profiles hadn’t yet surfaced as a major issue. “I am worried that it might be more difficult to do this kind of research in the future, since concerns about privacy have only increased,” he said. “You want to do good science, and you have this rich data trove, but you also want to protect people’s privacy.”
There has been a media narrative in certain sectors that fake news was rampant during the election and may even have helped influence the outcome. Guess . found just the opposite in a separate study published last year focusing on the consumption side of the fake news phenomenon, finding that Trump supporters over the age of 60 read the most fake news. The latest study focuses on the articles individuals chose to share during the campaign. But the conclusion aligns nicely with its predecessor.
“The overall prevalence [of sharing fake news] is lower than you would expect, given the popular narrative, more concentrated than many people thought,” said Guess. Rather, there was a large skewing of the results toward age and ideology. “It isn’t the case that fake news was being spread by people who were undecided or moderate, which I think would need to be the case in order for this to have big effects, for people to be swayed by this content,” he said. Education, income, and gender were not significant factors.
So who did share the most fake news articles? Pretty much the same demographic that consumes the most fake news: people over 65, mostly conservative Republicans. Eleven percent in that demographic shared fake news, compared to just 3 percent of users aged 18-29; 18 percent of Republicans shared fake news, compared to less than 4 percent of Democrats. But before liberals rush to congratulate themselves on being more media savvy than their conservative counterparts, Guess . point out that when they separated articles that were pro-Clinton (or anti-Trump) versus pro-Trump (or anti-Clinton), that effect vanishes.
“That’s one indication that it’s the slant of the stories that might be driving this ideological asymmetry,” said Guess. “In an alternate universe, or maybe even in the next election, if the political content were different and more of the fake news produced had a pro-Democratic stance, we might see more liberals and Democrats sharing than conservatives.”
That finding has implications for the kinds of intervention strategies and demographics targeted in efforts to combat fake news, which to date have focused primarily on teaching digital skills to kids in schools. That might still be beneficial, but Guess suggests it might be more effective to target the aging population more. “We’re in the middle of this huge interaction between massive technological change and this large cohort of people that are entering retirement with lots more free time around the same time,” said Guess. “We’ve never really experienced what the ramifications might be at a societal level.”
Guess acknowledges that studying fake news is very much a moving target. These two studies are based on 2016 data—an eternity in online culture. He and his colleagues are continuing to collect new data, but the amount of fake news has decreased considerably since 2016, along with the sharing of it on Facebook. Maybe this is because the platform is cracking down, or people have become more savvy about what they share, or there’s simply less fake news produced in a non-presidential election year. “It’s an arms race,” said Guess. “Who knows what the next form is going to look like?”
The two studies Guess . have published so far looked at consumption and sharing, respectively, but because they were separate, with different groups participating, it’s not clear if these are the same people. “Maybe we have people who are not actually reading content, and they’re inadvertently spreading it,” he said. (This is a reasonable assumption, given the degree to which people share articles online after only reading the headline.) “And we still don’t know very much about whether people actually believe the content they’re seeing. At this point, we’re still very much at the beginning of understanding the phenomenon of online misinformation.”
DOI: , 2018. 10.1126/sciadv.aau4586 (About DOIs).
