I don’t really have time to walk through the analysis one more time, so let’s go straight to my final 2016 prediction for the presidential race, which is an educated guess modified by a few gut feelings; I don’t pretend to have a model at this point, but it’s put up or shut up time. I don’t have much to lose here by sticking with my longstanding prediction that Trump loses, but at this writing, it seems it won’t be the blowout many of us expected as recently as a few weeks ago.
National Popular Vote: Hillary 48, Trump 45, Johnson a shade under 4, McMullin 2, Stein 1.
Electoral Vote: Hillary 303, Trump 235. Looking at the RCP poll averages, the race looks like a nail-biter of 1876/1960/2000 proportions, with six states divided by two points or less (four for Trump, two for Hillary), four states by less than one, and a 272-266 Hillary win on the strength of carrying New Hampshire by 0.6 points and Pennsylvania by 1.9:
However, much as I’m suspicious of over-relying on early voting figures, I’m also skeptical of some of the polls RCP is using (different poll aggregators have a less sunny projection for Trump). That said, I suspect we’ll see turnout that is not as bad for Trump as you might have predicted. My prediction flips Nevada and Florida to Hillary on the strength of heavy Hispanic turnout, and New Hampshire to Trump, but otherwise tracks RCP. Those three and North Carolina are likely to be the closest states: