Here’s my “official” take on the U.S. election for September 1. Sure, I admit I look at the polls almost every day. But I figure writing down the numbers and puzzling over them a bit once a month helps me to filter out some of the noise. So here goes. I still lean on the “Silver” numbers as probably reflecting the most well-thought-out adjustments of poll numbers to something close to reality. The 538 numbers are interesting to give a sense of how much small decisions about these adjustments matter, and the RCP numbers show what unadjusted (i.e., heavily biased) numbers would look like.
| STATE | 2020 RESULT | Silver Bulletin (August 1) | Silver Bulletin (September 1) | 538 (September 1) | RCP (September 1) |
| Arizona | Biden +0.4% | Trump +2.7% | Trump +0.6% | Harris +0.3% | Trump +0.5% |
| Georgia | Biden +0.3% | Trump +2.2% | Harris +0.9% | Harris +0.5% | Trump +0.2% |
| Wisconsin | Biden +0.6% | Harris +0.4% | Harris +3.2% | Harris +3.2% | Harris +1.4% |
| North Carolina | Trump +1.3% | Trump +2.2% | Trump +0.4% | Trump +0.3% | Trump +0.6% |
| Pennsylvania | Biden +1.2% | Trump +0.2% | Harris +1.3% | Harris +1.2% | Harris +0.5% |
| Michigan | Biden +2.8% | Harris +2.6% | Harris +1.9% | Harris +2.4% | Harris +1.1% |
| Nevada | Biden +2.4% | Trump +2.2% | Harris +0.9% | Harris +0.7% | TIE |
So going with the Silver numbers, the electoral college would be Harris 292, Trump 246.
Both Arizona and North Carolina have been in the Harris column during the month of August and flipped back over, while the numbers in Pennsylvania (>:-() seem like they might have tightened over the past two weeks. On the other hand, Georgia and Nevada are huge wins for the Harris campaign if they come through. Move Nevada and Georgia back into the Trump column and Harris still wins 270-268, with recount hilarity likely to ensue of course. This happens to match the RCP polling results above, if you give the Nevada tie to Trump. Surprisingly, she could lose Pennsylvania and still win the electoral college if everything else in the map above were to hold. But these things tend to be correlated and any event that moved Pennsylvania a whole point toward Trump would tend to move other states too. Unless we are talking some serious voter suppression or outright cheating by people in Harrisburg with pointy white hats in the back of their closets.
In the betting markets, PredictIt has Harris at a 56% chance of winning (the electoral college) vs. 47% for Trump (well actually $0.56 to $0.47 with about $0.08 given to other candidates, so apparently they don’t intend for these to add up close to 100%). Polymarket however has Trump at 51% to 48% for Harris. So whoever is betting on that site thinks they know something the rest of us do not.
So, my overall verdict is things look pretty good for Harris at the moment with two months to go. I think this election is hers to lose.