Twitter Posts Didn't Affect the 2016 Election
At least that's what a peer-reviewed study published on Tuesday indicates.
Many folks, including some who read this newsletter, believe its settled knowledge that Russian disinformation campaigns swayed the US presidential election in 2016. I have taken some heat for saying that's an assumption not something that has been studied. So I was happy to hear that scientists had looked closely at what effect the disinformation campaign on Twitter in 2016 had on the US electorate.
Tuesday a study was published that found that disinformation campaigns on Twitter in 2016 had little to no effect. This contradicts what a lot of folks believe so let's look at that study a little closer. The study was led by the New York University Center for Social Media and Politics, and published in the journal Nature Communications. It was conducted by scientists from the University of Copenhagen, Trinity College Dublin, and Technical University of Munich.
The study looked at disinformation spread by the Internet Research Agency, a group with alleged links to the Russian government, as well as a few smaller campaigns linked to groups in China, Venezuela and Iran. To identify those accounts, they used Twitter's identification of foreign influence campaign accounts.
For measuring people's attitudes and exposure, they used data collected by YouGov, from 1500 representative US voters. Surveys measured their attitudes and beliefs and the respondents approved access to their Twitter account to measure exposure to information. Surveys were conducted in April 2016 and October 2016. A third survey was conducted after the election to ask if they voted and if so, for whom. The study only looked at individual attitudes. It did not attempt to measure other effects that may have occurred, like effects on belief in electoral integrity.
The main finding was that Russian disinformation campaigns on Twitter in 2016 reached few users and there was "no evidence of a meaningful relationship between exposure to the Russian foreign influence campaign and changes in attitudes, polarization, or voting behavior."
In fact, exposure to information from US media and politicians vastly outnumbered exposure to disinformation. Study respondents were exposed to an average of 4 disinformation posts from the Internet Research Agency per day during the period, while they were exposed to 106 posts per day from news media and 35 a day from US politicians. That's 25 times as much exposure to news media than disinformation posts. The study found 70% of exposures to the disinformation reached 1% of users. Most of those users identified as "Strong Republicans.
Article on this from The Intercept
The study did not look at Facebook data because Facebook does not make it easy to access and study its data. If you recall, back in August of 2021, Facebook shut down personal accounts of NYU researchers who were trying to study the spread of misinformation on the platform. However Facebook has said previously that it estimates that 126 million users had the potential to view disinformation over a two year period around 2016. The paper estimates about 32 million US Twitter users were exposed to such posts in the 8 months before the 2016 election. Facebook in 2016 had around 3.5 times as many US users as Twitter. So the reach may have been similar on both platforms.
There needs to be more studies like this but it isn't the first. A study published in 2019 in PNAS found "no evidence that interaction with IRA accounts substantially impacted 6 distinctive measures of political attitudes and behaviors over a 1-mo period."
Thanks for reading. And to thank folks who have a paid subscription, there are a couple more paragraphs with my thoughts.
Keep reading with a 7-day free trial
Subscribe to Tom Merritt Tech Newsletter to keep reading this post and get 7 days of free access to the full post archives.