Did Fake News Tip The Election? A Research Redux
Although the actual effect of fake news on the 2016 election will probably never be fully understood, recent research on this topic should tamper the culpability of the largest social media platforms in shaping the election.
First, social media sites aren’t as central to the typical political conversation as people would think. Social media is the second least likely place people get their news, according to Pew. Instead of going simply to digital platforms, people get their news from all kinds of sources, including the radio, local news, cable news, and websites. And each of these sources covers a larger percent of the population than social media platforms like Facebook and Twitter.
Pew has also found in surveys that it is rare for users to discuss, comment, or post about politics online. And those that do go online to discuss politics are nearly guaranteed to be more partisan and thus unlikely to change their position. Indeed, social media networks aren’t as ideologically homogeneous as some have worried. The echo chamber effect is overstated. Most political conversations arise in casual settings with acquaintances.
As for fake news, it is worth consulting the numbers. As Facebook’s General Counsel, Colin Stretch, said under oath before the Senate Judiciary, the totality of the Russian misinformation scheme “equals about four-thousandths of one percent (0.004 percent) of content in News Feed, or approximately 1 out of 23,000 pieces of content.”
Facebook Vice President of Ad Product Rob Goldman also thinks that swaying the election was not the main objective of Russia’s campaign. Since the majority of Russian ad spending occurred after the election, their goal seems to have been sowing discord, much as the NYT detailed late last year.
In the year or so since the election, research into news habits suggests that fake news was concentrated among the most conservative users. And to be fair, these aren’t exactly swing voters. They news junkies, not low information citizens trying to make up their minds before voting.
Still, one needs to take a symmetrical look at this problem. That is, if news does have a huge effect on voter turnout, then the extensive sharing of polling prediction models by 538 and the New York Times should have had a dampening effect. That is, the perception that Clinton was doing well in the polls should have lead to fewer people turning out.
Political scientists are worth consulting on this topic because they tend to take a more nuanced approach to news as a driver of voting decisions. A recent meta-analysis of campaign persuasion confirms that predilection, since the study finds that the average effect of outreach efforts like in-person canvassing and mail in general elections is zero.
In “Ten Things Political Scientists Know that You Don’t,” Hans Noel summarizes what political science has come to accept: “It is not that campaigns do not matter at all in presidential elections. It is that after decades of searching, we have found so little evidence that they do and so much evidence that the fundamentals matter more.”
In a mea culpa, Nate Silver confirmed this view: “Accounting for where the candidates spent their resources makes almost no difference, it turns out, once you’ve controlled for one or two major demographic categories and the 2012 vote [using regression analysis].” As Silver noted, reporting largely ignored the importance of economic conditions and the fundamentals that suggested a far closer race between the candidates.
Indeed, voter prediction models based in political science seem to have been better at divining the outcome. Of the major models, three predicted Trump would win and the other three predicted only a very narrow Clinton victory. Among the most consistent feature in these models is they emphasize change in party. There’s good reason for this. Since Harry Truman, political parties have only been able to hold the presidency for three consecutive terms once, and that was the Reagan-Bush run.
As a final note, the perception of the problem of fake news is likely skewed as well. In communication studies, there is a well supported theory known as the third-person effect. According to this theory, individuals believe that messages have greater effects on those outside of the group than either themselves or their own group members. This survey of nearly 1300 people suggests that the third party effect is live and well when it comes to fake news.
All of this being said, fake news is a problem in that it undermines trust in the system. It also changes the bounds of acceptable debate since leaders can claim something is fake news, sidestepping the need to properly defend their actions. Still, if we want to solve this problem, we need to be rooted in evidence, not merely overblown hype.