Most of the final national polls seem to be in and Biden is at +8.6 in 538. If you apply "how red/blue was each state compared to Clinton's +2.1 in 2016?" you get:
(Winning everything up to Georgia = 350 electoral votes for Biden)
If you look at *state polling averages* and compare to the 2016-replay estimates, here's how polls suggest each state has changed in terms of partisan lean.
FL ⟶R+4
PA →R+1
NC →R+<1
GA ←D+<1
AZ ←D+<1
IA ←D+1
TX ←D+1
OH ←D+1
MI ⟵D+2
WI⟵D+3
If Trump doesn't pull it off in the end, this will be a finding to remember when thinking about what this election was about (in the way everyone remembers how Trump won people who disliked both candidates in 2016).
We talk about "swing," "independent" or "moderate" voters, but often the difference-makers are probably best called "cross-pressured."
As an aside, if Biden pulls off a landslide, it will be funny when political scientists show the result was ultimately consistent with the negative economic story, but exit polls show the majority of voters still approved of Trump on the economy.
If the polls are right nationally and in Florida it suggests something fairly unusual (but foreshadowed by 2018) has happened there specifically
I mean there will new not unusual explanations, but it won’t be “Florida is always close” it will be “Florida has gone from 3-4pts redder than America to 8-9 points”
The Upshot poll is more evidence of Florida “close” e.g. shifting meaningfully to the right, even allowing for Biden’s relative strength with older voters (and a solid Biden lead with Hispanics). But we will have to see.
Unreal. The fast growing Travis county is at 64.74% turnout among registered voters—higher in percentage terms than 2016 (63.8%) but also higher than every other election since 1994, except juuust short of 2008 (65.05%). Before Election Day.
One of the factors in turnout is competitiveness. And Texas learned at some point in 2018 it was competitive. Seems to be leaving a mark...
It will be interesting to also see what this does for Hispanic turnout in general. Note the expert below, which is based on GOP pollster Echelon Insights’ turnout estimates for 2020. A very large portion of the Hispanic vote (38%) live in two states just becoming competitive.
Wisconsin is a perfect storm of annoying for predicting partisanship. The vote there is not *that* polarized by education, or age (relatively speaking); there is no party registration; and everyone is white.
The broader point here is to “predict” partisanship of a person (an important tool for avoiding issues like non response bias) you need either person’s party registration or *other info highly correlated with partisanship.* Both present a particular challenge in Wisconsin.