This is not my original analysis. I am piggybacking on the efforts of this guy:
It’s a good video that looks at existing voting patterns and ballots returned and tries to extrapolate them to see what kind of turn out Trump needs on election day to win. He also has made it Google Sheets model public, so for nerds like me, you can download it and play with the numbers yourself.
He references very interesting data here:
The guy shows Biden winning in every battleground state at the moment, usually with estimated leads of 3-4%. This is an oversimplification, but for Trump to win, he would need to win Election Day by about that much in reverse.
What’s interesting is when you combine it with data like the following two links, which show a huge difference between parties on who will vote on Election Day.
In PA, this Emerson survey showed 81% of Trump voters will vote on Election Day, while only 36% plan of Biden voters plan to vote on Election Day.
In FL, this Emerson survey showed 53% of Trump voters will vote on Election Day, while only 17% plan of Biden voters plan to vote on Election Day.
emersonpolling.reportablenews.com/pr/florida-2020-biden-holds-slight-edge-three-weeks-out
Now, there are some huge assumptions here:
Unaffiliated Voters will split 50/50
People will actually vote early/late as declared
People will more or less vote their declared party
The TargetSmart estimates are good
But if those assumptions are sound, my numbers show the following:
Turnout is huge.
Trump is going to crush Biden.
My FL numbers show a stupid blowout, greater than 55% to 45%.
There is so much variability and so many assumptions, that it’s not worth focusing on the exact numbers, but the fact is that Biden should have much bigger lead in Democratic submitted ballots than he has now if Election Day is really a dominant Trump day.
I have similar results even in leaning blue states like NV and NH. Biden just doesn’t have much of a lead there at the moment.
If this is true, states like NM might even be in play.
h/t Zalinsky