OK, y'all, Nate Cohn's comments on the last NYT survey should be appreciated in full. And I will add one more example to the shift he describes: the Times poll had Trump running about 9 points better in NYC than in previous years. Thus:
***--- National Polls Do Not Matter ---***
Because the race is so close in the swing states, campaigns are paying little to no attention to other less competitive states. And Trump is running wild in places where his base support is strong. Good for him. It's pushing his national polling numbers up. It does nothing for him in the EC.
National polls are only relevant when there is a predictable partisan lean for the swing states. But that partisan lean is not predictable this year. And pretty much everything I've read says to trust the state polls over the national polls this late in the race -- well, the swing state polls because there are so many of them.
We do not need a popular vote cushion. Trump could win the popular vote and lose the EC. I don't know why people are assuming that's impossible. Remember, the main factor in creating the Dems' EC deficit was that the most populous state has been a blue landslide state, but the third most populous state was a pink state just out of reach. OK, knock about 5 points off Dem performance in CA and add 5 to FL and the EC issue starts to dissipate real quick.
Pretty much everyone is acknowledging this now. It was only Nate Cohn a month ago. The swing states are largely unmoored from the national trends, at least at the closeness levels that we care about. If the national polls move by 7, I'd expect the swing states to move and there's no real difference if they move by 5, 7 or 9. But if the national polls move by a point, and the swing states move by a fraction of that -- which is what we care about -- then it's nothing.