GRAND RAPIDS, Mich. (WOOD) — A study done at London’s Oxford University took a look at election-related posts on social media Goliath Twitter and found that a big chunk of campaign-related news shared in Michigan was of the fake news variety.
Researchers at Oxford say they started out with 22 million tweets posted between Nov. 1 and Nov. 11 by people who self-identified as living in Michigan. They were then able to narrow down those tweets to election-related by looking for keywords like the names of the candidates or campaign-related slogans and hashtags like “make America great again” or “I’m with her.” They then divided those tweets up by content, deciding which were purporting to share news.
“We’re only talking about when people were sharing links, so if they put, ‘Love her, Hillary Clinton,’ it doesn’t go in. This is only when people are sharing links to other information,” said Gillian Bolsover, an Oxford Internet Institute researcher and one of the authors of the study released last week.
She and her colleagues found that fake news accounted for a quarter of the campaign-related links shared, equal to and often exceeding the links shared from legitimate news sources.
“These are not things that adhere to the professional standards of journalistic reporting,” Bolsover said.
>>PDF: The Oxford study
Bolsover said sources like NBC News, The New York Times and Fox News are examples of legitimate sources.
“But in the days sort of right before the election, the proportion that was professional news dropped,” Bolsover said.
Infowars, Breitbart and Truthfeed are cited by report authors as junk or “fake news” sites.
“They’ll often use these techniques that we’ve seen in sort of propaganda, persuasion and advertising techniques,” Bolsover said.
The study gives a glimpse into what was going on in the election via a major social media platform.
“I think this is really the first piece of academic research, or any research really, that’s been done to try to estimate how much of the content that was being shared was fake news,” Bolsover said.
The results may surprise some and merely confirm suspicions for others.
“I’m surprised it’s that low, to be honest,” said Len O’Kelly, an assistant professor of journalism at Grand Valley State University.
O’Kelly teaches multimedia journalism — which these days means Twitter.
“That’s where many of us go to get our news first now,” O’Kelly said.
The Oxford study looked at Michigan because it was a battleground state that unexpectedly voted Republican — albeit by a narrow margin — for the first time since President Ronald Reagan was in office.
“I would hypothesize that this was common across all battleground states, so there’s nothing bad about users in Michigan,” Bolsover said.
A study done by others found that German elections also saw a lot of fake news shared, but far less than in the United States.
“It really is incumbent on the consumer to take some time and figure out: this one I can trust, this one I can’t trust,” O’Kelly said.
While the study showed pro-Trump tweets vastly outnumbered pro-Clinton tweets, Oxford researchers say next they want to drill down to see whether they can determine if there was any kind of coordinated effort.
Perhaps the unanswerable question is whether the tweets had an impact on the outcome of the vote.
“The 2016 campaign was unlike anything we’d ever seen. We are in uncharted territory,” O’Kelly said.
He says it’s hard to say how influential tweets would be so close to an election.
“We give a lot of attention to what’s going on on Twitter when I think in a lot of cases people aren’t reading everything that’s there,” he said.
It also leaves open the question of what to do about it.
“There’s an important balance to be made between wanting to control misinformation and allowing information to flow freely on the Internet,” Bolsover said.
So what comes next?
“Technology’s moving so fast and society’s moving so fast that there’s not particularly a reason to believe that the problems in four years are going to be exactly the same. We have to be thinking about the underlying, large trends,” Bolsover said.
“That gives us pause for the next election cycle, that if you think this is a large amount of tweets, wait until 2018,” O’Kelly predicted.