Did fake news affect the election? Criticism of Allcott-Gentzkow study

Recently, a study was conducted by economists in an attempt to measure the impact that fake news had on the election. According to their research, the effect was minimal.

Quote:

Originally Posted by 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.

However, I have some criticism of their work. Mainly, I feel like input from a psychologist would have helped. I'm no expert myself, so I will highlight some specific concerns, and hopefully someone more knowledgeable than me will let me know if I'm on track.

1. Ignoring Response Bias^WP

Quote:

Just 14 percent of Americans deemed social media the primary source of their campaign news, according to their research.

People do not always respond accurately to surveys. Respondents likely under-reported how using Facebook as a primary source for news, because this is perceived as socially undesirable.

2. Assuming that social media is all that matters
Even if their numbers are right, Facebook is not the only source for fake news. Here is one prominent example spread by Donald Trump's own campaign manager. When this happens, it shows up on TV, gets discussed by coworkers, etc.

3. Overemphasis on recollection
This is the one that really stood out to me.

Quote:

Originally Posted by 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.

It seems that the researchers are operating under the belief that if someone cannot recall a specific headline, then either they did not see it, or it is not affecting their judgment. Our beliefs are affected by all sorts of stimuli and information, even when we can't remember precisely how we came across the info.

If I'm misunderstanding something, or making some bad inferences, please let me know.

Citations:
http://ift.tt/2llpCWC
http://ift.tt/2llk7re
Original study (pdf): http://ift.tt/2iJUf6O

via International Skeptics Forum http://ift.tt/2m4FIZ1