It seems to have become an article of faith among Democrats and progressives, even statistically-oriented ones like Nate Silver, that the Comey letter on Clinton’s emails, released on October 29, 2016, swung the election to Trump. Given how close the election was, almost any minor event could have swung the election. But the evidence that the letter had a significant effect, e.g., swinging the popular vote by 3 or 4 points isn’t there. Consider my Facebook post from Monday, October 30, 2016:
As of today, with polling through yesterday, HRC up two in IBD Tracking, one in ABC tracking, and minus 2 in the (leans Trump) LA Times poll. But each of these polls is a 7 day poll, meaning only one day post-email news has been taken into account. And all have been trending to Trump even before the emails. No poll that concluded Thursday or after shows HRC with more than a 3 point lead. A poll conducted today would probably show a dead-even race, maybe a small Trump advantage. This may very well be fleeting, but the media doesn’t seem to be catching on that HRC no longer holds a lead.
So as of when the Comey letter was released, Clinton had a small lead in the popular vote, with the polls trending toward Trump. When the votes were tallied, Clinton emerged with a small lead in the popular vote. Analyses like Silver’s simply don’t take into account the fact that Trump had significant momentum before the Comey letter’s release, momentum that didn’t fully show up in analyses that relied largely on days-old polls and polls that reflected a week’s worth of polling.