Dragan Radovanovic / Business Insider
- The Trade Desk, an advertising-technology company, took a dive into the cookie pools of readers of two "fake news" stories, one targeting readers on the left and the other on the right.
- It calculated "relevance scores" for a variety of characteristics for each of the two articles, getting an idea of how the audiences for the posts differed from the population at large.
- It found that fake-news readers interested in a pro-Trump story were more likely to be older, white, and in the military or work for the federal government. Those lured by a fake story about the Standing Rock protest by Native Americans were more likely to be affluent, college-educated, and working in the financial-services industry.
Who reads fake news anyway? We all have that uncle or colleague who shares dubiously sourced articles from websites that sound like they belong to national news organizations but don't.
While the focus these days is on fake news that targeted conservative voters - thanks to the specific allegation that the Russian government propagated this kind if misinformation on Facebook to sway last year's presidential election - it's important to note that liberal readers were also lured into fake stories that fit their political mind-set.
Ad-tech company The Trade Desk took a closer look at the readers of fake-news stories of both kinds to build a profile of who took the bait - both from the right and the left - using the same tools a business might deploy to target their
One story was from a site called American News titled "BREAKING: Congressional Plot To Bring Down Donald Trump Escalates ... They Have Momentum Now," while the other was a story titled "Police Raid Standing Rock Camp, Dismantle Tipis And Are Burning What Remains" from the Alternative Media Syndicate.
The fact-checking site Snopes.com lists a number "false" stories from American News, including this one about a plan to add President Barack Obama to Mount Rushmore. The Standing Rock story was specifically outed by Snopes as one that was dubbed "fake" by one of the protest organizers.
The Trade Desk used the "cookie pool" - or the list of other websites viewed within an hour of reading one of the fake stories - of users clicking each story to try to home in on profiles of who these readers are. It created a "relevance score" that shows the ratio of the likelihood that a reader of an article has a certain trait, like an age bracket or a gender, compared to the prevalence of that trait in the overall US population.
We're describing this as a "glimpse" into fake-news readers because there's an obvious limitation to the study. The analysis looked at only two stories, and we don't know what readers did next (like share the story on Facebook). That said, TradeDesk is doing these analyses as the industry tries to work out how to keep advertisers away from this kind of content. Plus, it's an important reminder that liberal readers (and voters) are also susceptible to the lure of a too-good-to-be-true headline.
"The most surprising finding of our study is that fake news affects both the right and the left, the educated and uneducated," The Trade Desk's CEO, Jeff Green, told Business Insider. "There was also a strong correlation between what people read and how they voted. Of course, we expected people's voting to be influenced by what they read; however, the truth was secondary to the content and volume of what they read."
Here are some of the characteristics of the readers of the right-wing American News article, according to The Trade Desk's estimates. The numbers below show how much more likely these readers are to match a certain trait than the general population:
According to The Trade Desk, fake-news readers on the right were more likely to be older, white, and in the military or work for the federal government.
They were about 56 times as likely to be Gen-Xers, 22 times as likely to be white, and 21 times as likely to be male than Americans as a whole. They were also 39 times as likely to be in the property industry than the US population at large, and 27 times as likely to be over 65.
On the other hand, here are some of the characteristics of readers of the left-wing article from Alternative
Fake-news readers on the left were more likely to be affluent and college-educated, wealthy, and in financial services.
Specifically, they were 54 times as likely to be politically influential, 34 times as likely to be college graduates, and 31 times as likely to be "positive idealists" than the overall US population.
The Trade Desk also looked at relevancy scores based on geography. States that are more conservative unsurprisingly tended to attract more readers to the right-wing American News article:
The map bears some resemblance to the 2016 Electoral College map. States that had higher relevancy scores for the right-wing American News article were more likely to vote for Trump last November:
Ultimately, the goal of the study was to help protect advertisers from being caught up in the world of fake news, said Green. Once advertisers become aware of such publishers, they may be discouraged from running fake stories, he said.
"Understanding fake news helps us protect advertisers from being there. We want to defund fake news," he said. "It's good for all reputable players. We think changing the economics or moneymaking prospects of fake journalism is the fastest way to end it."