The 448-page investigative report that Special Counsel Robert Mueller released in March makes it clear that Russian operatives used social media to interfere with the 2016 U.S. presidential election.
Now new research forthcoming in the journal Computers in Human Behavior offers insights into how the Kremlin’s elections-interference operation played out on Twitter. The study finds that Russia’s Internet Research Agency — a troll farm — had a dual strategy: to sow division and weaken America’s two main political parties while also bolstering voter support for Republican presidential candidate Donald Trump.
Lead author Darren Linvill, an associate professor at Clemson University’s College of Behavioral, Social and Health Sciences, said the IRA wasn’t using Twitter to circulate so-called fake news, as some people have believed. It was “pushing the extreme version of narratives we already created ourselves,” Linvill tells Journalist’s Resource.
“They were trying to push both sides to extreme positions … but they’re doing so in a way that very much also pushed for the election of Donald Trump,” he says. “You can very easily do both at the same time. In fact, doing both at the same time was a good [effective] mechanism.”
Linvill and his research team examined tweets originating from the IRA in the month before the 2016 election. The researchers used data that Linvill and co-author Patrick Warren, an associate professor of economics at Clemson, had collected for an earlier study. That study looked at about 3 million tweets from 3,841 Twitter handles that the U.S. House Intelligence Committee had reported as being associated with IRA activity.
For this study, titled “‘THE RUSSIANS ARE HACKING MY BRAIN!’: Investigating Russia’s Internet Research Agency Twitter Tactics During the 2016 United States Presidential Campaign,” Linvill and three colleagues analyzed a random sample of 4,200 tweets written in English. They find that the tweets supported and attacked both Trump and his Democratic challenger, Democratic presidential candidate Hillary Clinton.
Some of the IRA’s tweets also attacked the news media or public institutions such as the U.S. judicial and electoral systems. The majority of tweets — 52.6% — were neutral. The researchers categorized them as “camouflage.”
“These tweets had no clear or overt connection to IRA agenda building activity,” the authors write. “Such tweets may serve to help give credibility to the IRA account or to build Twitter followers and make connections with a potential audience.”
Of the 4,200 tweets examined, 12% “primarily attacked left leaning ideas, ideals, and/or candidates” and most of them targeted Clinton or her campaign. Meanwhile, 7.4% were in support of liberal ideas and candidates, including Clinton.
The researchers find that 5.4% of the sampled tweets attacked conservative ideas, ideals or candidates and that the majority of those focused on Trump and his campaign. But 7% “primarily supported right wing ideas, ideals, and/or candidates. The majority of these tweets explicitly supported Donald Trump or the Trump campaign,” the authors write.
Another 2% of tweets attacked the news media. “These tweets primarily attacked the validity of mainstream media outlets. They often attacked a perceived liberal bias in the media …,” the authors write. “But there were also some attacks on the media from the left. These tweets often charged the media with racial bias, including @BLMSoldier’s retweet, November 2, 2016, ‘White man kills two cops n #Iowa and media or Trump give no attention. Only matters if a minority doing killing I guess. @NPR #Elections2016.’”
Nearly 7% of tweets could not be categorized because they either no longer could be viewed or contained a link that was no longer active. About the same percentage criticized civil and governmental institutions in the U.S.
The study underscores the importance of taking a broad view of social media — looking at a range of Twitter accounts — to understand the IRA’s agenda-building process.
“Had we analyzed a narrower data set we would have received a blinkered view of IRA activity,” Linvill and his colleagues write. “Looking only at the account @southlonestar, for instance, may lead one to believe the IRA was actively engaged in theTexas secessionist movement. Similarly, looking solely at the account @blmsoldier could lead one to infer the IRA was passionate about police violence. It is only by looking at the messaging as a whole that we can understand IRA tactics and how they work together to influence differing audiences.”
As Linvill awaits the study’s publication in the October edition of Computers in Human Behavior, he and Warren continue to investigate IRA tweets to learn more about how they are used to influence Americans. The two researchers have launched two studies that will build on the findings of this one.
One of the new studies aims to understand who engages with IRA tweets and whether this has changed over time. The other focuses on how Americans engage with them. For example, do they retweet them or attack them? Do they reply with a message back?
“Organizations like the IRA — it’s public relations and marketing, it’s not spycraft,” Linvill says. “It’s PR and marketing that they’re really good at.”
If you’re covering Mueller or the Mueller report, check out this tip sheet from Mueller expert Garrett Graff. Looking for research on voter turnout? Check out our write-ups on how turnout is affected by negative political ads, voters’ race and voter health.