Monday, November 2, 2020

Some 'final predictions'

Tim Miller, writing at the offers Final Predictions With caveats and life lessons. Here are his predictions and his reasons for them. I’ve reprinted them here without the usual block quotes but with added emphases. Jumping right in …

I come to you as somebody who has looked at the data closely, considered my blind spots, and tried to fight against any priors that might shade my perspective.

The following predictions come with three basic caveats:

(1) The turnout for this election is going to be off the charts. And while JVL has been promising that means the polls will turn out to have been more accurate, I am slightly wary that it will result in some wonky outcomes, because it is hard to model things like 114 percent of 2016 voters in a county showing up before we even get to election day. If that breakdown is disproportionately made up of white people, for example, that might help Trump’s Electoral College math a bit.

(2) If there is a polling error in Trump’s favor that narrows the margins, the chicanery with the early vote in places like Harris County, Texas and the mail-in ballots in Pennsylvania is deeply concerning.

(3) The gold-standard Selzer poll in Iowa showing Trump up 7 is a little freaky and is moving in a different direction than what we are seeing in other states. It includes a downright crazy-seeming result showing the GOP +15 in the first congressional district which flipped to the Dems in 2018. If things get hinky on Tuesday, that number will be the canary in the coal mine.

All that said … after the longest throat-clearing wind-up in the history of predictions, here is the Halloween candy y’all came here for:

Electoral College

Over the past week we’ve seen a bevy of Florida polls largely moving the same direction as the Republican Voters Against Trump internal poll, which showed undecideds in the state breaking for Biden: Monmouth had Biden +6, Marist +4, Harris +3. By the time you get this newsletter, my sources say there will be two other public Florida polls showing similar numbers. Meanwhile ABC is an outlier moving the other direction. With the RVAT poll that makes 6 out of 7. Is it possible that all these pollsters are wrong and ABC nailed it? Sure. It’s possible. But when in doubt, look to the averages and the trends, which are pointing to a Biden win. And if Biden wins Florida, all he needs is one of the other swing states and he’s the president.

The story in North Carolina is similar: CNN has Biden +6, Marist +6, NYT +3 in the past week.

In Pennsylvania there’s the aforementioned concern about the mail-in ballots, but if everything goes relatively smoothly the polls show a stable lead even greater than in Florida.

In Arizona there hasn’t been as much quality polling in the final stretch so it’s possible we are blind to something that is happening there, but Biden has led the entire year and it is hard to imagine a reason for it to move a different direction than these other states.

As for Michigan and Wisconsin, Biden has held clear leads there throughout this race and Trump has all but thrown in the towel in those states.

And then there’s Georgia. A state that hasn’t been a presidential target my entire life. Biden is sending Barack “The Closer” Obama there on Monday and it’s not just for kicks. Trump is spending a boatload of money on TV to protect it. The 538 average has Biden +1.5 in the state. If you look at the final 2016 results you will see that Georgia was actually much closer to a swing state than it was to the next batch of “stretch states” for the Democrats. (Trump won North Carolina by 3.5 percent, Georgia by 5 percent, and then the next closest state was Ohio at 8 percent.)

In that context it shouldn’t really be a surprise that Georgia would be in a position to go slightly for Biden if he is to improve on Hillary’s national margin by 5 percent as the polls currently suggest.

And thus … I think you know where this is heading …brace yourself …

My dispassionate projection is that Joseph Robinette Biden Jr. wins all of the states mentioned above—Florida, Arizona, Pennsylvania, Wisconsin, Michigan, Georgia—for a total of 350 electoral votes. And This means he takes NE–2 and comes up a bit short in Ohio, Texas, ME–2, and Iowa.

Added bonus: If the numbers bear out as we expect, Biden will win Florida and North Carolina clearly enough that the election will be over Tuesday night or Wednesday morning and spare the country the weeks of legal wrangling over Republican voter-suppression tactics.

So there it is. And now I must ask something if you…You all are Bulwark+ paying members. (and Scriber is!) We are pals now. If things go south, I want everyone to bleach bit this newsletter from their hard drive and if anyone asks just say that I provided the best data-driven analysis that I was capable of with a traumatized lack of confidence.

P. S. Miller displays a stunning map illustrating his predictions. It’s interactive and my editing software does not play nice with it. I guess y’all will just have to cough up the cost of a subscription to The Bulwark.

Scriber suggests that you check back Tuesday evening and Wednesday morning.

No comments:

Post a Comment