Thank you for your response. I appreciate that your points are germane to the topic, even if I disagree with some of them. Yes, I am arguing first that exposure to fake anti-Hillary news did not change the votes of people already opposed to her. I am arguing much of the traffic in anti-Hillary fake news is between people who fall into this category. Second I am arguing that unless you can produce individuals who can say in all honesty they switched their votes because they saw some anti-Hillary news, then you have suppositions, but no actual evidence, that fake anti-Hillary news had any influence on the result of the election. Take your story about the Pope endorsing Trump. Get a legitimate polling organization to go out and interview a few thousand people who viewed it. Hone in on the subpopulation that say they changed their preference at some point in the Presidential race. Ask them to estimate when they changed their mind. If it happened after exposure to the fake news, then you arguably have someone who was influenced fake news. Even better would be finding someone who said they recall changing their mind as the result of seeing the phony “Pope” endorsement news. Trouble is neither you nor anyone else has been able to find a single voter who has said that. To go from speculation to a scientifically valid hypothesis, you do need to find such individuals or alternatively find another way to approach it so that your assertion is capable of being disproved by data.
Your hurricane story is interesting, but not terribly relevant to analysis of the influence of fake news on the election. Exposing people to a phony story about something new where they have no fixed prior opinion is far different than exposing them to a story trying to influence them to change their minds when they already have a strong opinion. The new “news” is either taken as validation of prior opinion or is ignored as being wrong because it is inconsistent with prior opinion. I have friends who do consumer research on the formation, intensity, and persistence of brand preferences that I think is more relevant.
Finally I think there are other real problems such as how do you distinguish responses to real news versus fake news, and how do you distinguish late deciders who did not follow any news from voters who changed their mind in response to real or fake news. It might be interesting to start with studying the impact of some real news, for example, trying to find people who changed their minds due to Comey’s late October revelation about additional e-mails.