A Probabilistic Look at The Presidential Election: 14 September 2024

Drifting Towards Harris

This is my first update on my new schedule. I downloaded the data from ElectoralVote.com yesterday, the 14th, at around noon and I’m processing it today on Sunday. With luck, I’ll have this out tonight.

The State of the Race

It’s been a bizarre 10 days! Monday seems like a lifetime ago. It was on Monday when I clicked on a “Cats for Trump” hashtag on Twitter or whatever the hell it’s called now. It struck me as pitiful. It seemed like Trump’s people were out of ideas and were trying anything to move the needle or slow Harris’s momentum.

Tuesday was the debate. It was a solid and clean win for Harris. We got the Trump we saw in the first debate and his furious firehose of falsehoods. That should have been disqualifying then, but we and the media were too busy realizing that President Biden is an octogenarian. In the second round, we got more of the same from Donald Trump with the added bonus that Kamala Harris was really able to piss him off. That led to some crazy, incoherent rants. It turns out it’s pretty easy to push that guy’s buttons as Vladimir Putin and a bunch of others figured out long ago. The only people I saw arguing that Donald Trump won the debate were on right-wing websites where they probably didn’t have the option of saying anything else.

Unfortunately, there are so few undecided voters at this point that the debate didn’t move the needle immediately. As things get played over and over over on the teevee machine some stuff will reach critical mass and start to sink in. A debate bump does not need to be instantaneous.

Oh. Also, the debate and a crudely made AI likeness inspired Taylor Swift to finally endorse Kamala Harris. Trump responded by saying that he hates her while making a knockoff of one of her concert T-shirts. This will eventually be hilarious, but not for Trump.

The latest news is that there was gunfire this afternoon near Trump’s golf course in West Palm Beach. I knew that neighborhood or did 20 years ago. It’s fine, but gunfire isn’t all that surprising. The Trump campaign tried to color this as a second assassination attempt. I guess we’ll see.

Results:

We ran our usual 40,000 simulations of the election. If the election were today it looks like Harris would win about seven out of 10 times. Trump’s chances are close to three out of 10 and as usual we’re seeing about one percent of the results turn out to be ties.

That seems to track. Harris’s collection of “safe” votes is growing and is nearly twice as large as Trump’s. That’s a big driver of the probability. Make no mistake, there are enough toss-up votes that this thing could go either way.

Trends:

We can go back to tracking the individual changes because we’re on our second week of taking polls back 15 days. We have apples-to-apples comparisons. I’ve split the shifts into the moves toward the left and the moves toward the right.

Colorado and Arkansas don’t really show us any recent changes. Those states have probably been in the strong columns for months. It’s just that they haven’t been polled in a long time. We’ll see the same thing when we finally get new polls from Oregon and Indiana.

State	Change
Colorado	Leaning Democratic to Strongly Democratic
Virginia	Leaning Democratic to Strongly Democratic
Michigan	Toss Up to Barely Democratic
Ohio	Strongly Republican to Leaning Republican

Michigan and Wisconsin don’t tell us too much either. These are small shifts on the edges of the categories. Light blue, light red, and taupe are all mostly inside the margin of error.

State	Change
Wisconsin	Leaning Democratic to Barely Democratic
Arkansas	Leaning Republican to Strongly Republican

Virginia and Ohio on the other hand are pretty big shifts and are mostly in Harris’s direction. Ohio doesn’t look likely to turn blue, but if it shifts into a place where it’s competitive and Virginia settles into being safely blue that could inspire a potentially game-changing shift of resources.

Last Words

Kamala Harris having a seven out of 10 chance of winning the election if it were held today seems pretty reasonable at this point. Nate Silver this week has those probabilities almost exactly backward, with Trump having about a seven out of 10 chance of winning. On the other hand, I spent some time this afternoon listening to Christopher Bouzy on Spoutible, revealing his new electoral map. That one seems wildly optimistic in Harris’s direction.

But both of those guys are trying, something more robust than a polls-only estimate. They’re trying to factor in other information, which is more subjective. That might explain why these two projections differ so strongly. The two approaches should theoretically converge as we get closer to election day. In the meantime, I should take a dive into their reasoning to see if I can assess it. Right now I’m comfortable being in the middle.