Well, before the election, I knew very few people (in a heavily "red" state) that would fess up to wanting Trump. Based on that observation, I suspected they were probably too ashamed to admit that to an anonymous pollster and possibly to themselves. I figured *if* Trump was within 10-15% of Hillary poll-wise, he'd probably win, and he was within 10% at the final day... Not a terribly scientific observation, but it turned out a good guesstimate. Somehow have to factor in the psychology of people who vote not FOR a candidate, but AGAINST a candidate. (which seems to be most of the votes submitted this cycle were either AGAINST Hillary OR Trump, but not actually FOR either and might not admit to voting for either) . Maybe they need to change the question - like "which candidate(s) would you NOT vote for?", and not even worry about the "who are you voting for?" questions.