ROBERT SIEGEL, HOST:
False, fake, fabricated news - we'll look at the impact online fakery had on our recent election and what other countries are doing to prevent it from affecting theirs. It's All Tech Considered.
(SOUNDBITE OF MUSIC)
SIEGEL: There's been a lot of talk about whether those stories spread through social media helped Donald Trump win, one professional fake news writer said as much to The Washington Post. Well, economist Matthew Gentzkow of Stanford University has tried to fact check that claim. In a new working paper he calculates the influence that those stories had on the election. And, Matthew Gentzkow, welcome to the program, and what did you find?
MATTHEW GENTZKOW: We found that in order for fake news stories to have changed the outcome of the election, seeing one fake news story would need to be as persuasive, have as large a chance of changing people's votes, as seeing 36 TV commercials.
SIEGEL: Tell us about your methodology, then. After you had identified some headlines that had been called out as fake news, how did you measure what impact they had on people?
GENTZKOW: So we collect this database of as much fake news as we can find. We then use a new survey online to estimate how many people saw those fake news stories, and then putting that together we can benchmark the persuasive impact that fake news would have needed to have against something we do know something about which is the effect of television commercials in campaigns.
SIEGEL: Interestingly for this experiment you used placebo headlines, which is to say fake fake news, things that people couldn't have seen because you made them up.
GENTZKOW: So this - the question we wanted to ask people was, do you recall seeing a particular fake news headline prior to the election? We had the suspicion that if we did that, the number of people saying they had seen it would be inflated because people might misremember and say sure, I saw that headline, when in fact they didn't. So to try to guard against that possibility, we came up with a set of what we called placebo stories, fake fake stories that we invented that did not actually circulate before the election. And so by seeing how many people recalled seeing the placebo stories, we could get a sense of the size of that false recall and control for it.
SIEGEL: Yeah. The very odd finding that you made was that the same number of people who recalled reading the let's say real fake news stories and the fake fake news stories was the same.
GENTZKOW: Yeah. The number of people who saw them was almost the same. It was just about 14 or 15 percent of people recalled seeing the fake news stories, and just about 1 percent less recalled seeing the placebo stories. And if you just sort of, you know, read the news online or watch cable TV, you get the impression that we were inundated by a flood of fake news. And, you know, the fact that the average voter saw one of these kind of crazy made up stories during the election is still pretty striking, but it's way short of a world where this was a wave that crashed over everybody in advance of the election and we were all seeing dozens of these stories every day.
SIEGEL: Economist Matthew Gentzkow of Stanford University, thanks for talking with us.
GENTZKOW: Thank you so much.