2016 Postmortem

event in the last few weeks of an election.

Think about it this way. You get to the end of a campaign. Most folks have not only decided for whom to vote, they are heavily invested in their candidate. They really don't like the "other party's" candidate. Then something comes out about your candidate that makes you wonder if you should vote for them. You still hate the other candidate. You aren't going to vote for them but you don't feel as enthusiastic about voting for your candidate either. 95%-98% of the folks who think like this, by the way, are still going to vote and vote for the original candidate they chose, but 2-5% will stay home. When polling agencies call to make their last surveys, they may miss a lot of these stay home folks when calling their small slice of the electorate. If they do reach one or two of them, they still have a preference for the original candidate, the only difference is the likelihood of them voting, but that may not come out in the survey. The polls may show nothing or may show the slightest of dips but it will not be as pronounced as what is really going on.