It’s about 4pm the day before the election, and while Nate Silver may yet push through one more additional run of his model, I thought this would be a good time to memorialize the state of the race going into tomorrow, as his model sees it.
Right now he has Trump as the narrowest of favorites, with a 50.4% chance to win 270+ EV, versus a 49.2% chance for Harris to win 270+ EV, and a 0.4% chance of a 269-269 tie. This is in spite of Harris having a 2.3% projected edge in the national vote share, as well as a 5 EV edge in the weighted-average outcome, 271.5 to 266.5.
As I had done back in August, I’ll start by listing the “very safe” states, namely those for which Silver’s model projects at least a 95.5% (two-sigma) chance of victory for one party or the other. In most cases, these states’ probabilities are closer to 1.0 than they had been in August, presumably reflecting a reality that the extreme things that would have had to happen to get those states to break the other way haven’t occurred. Italicized states are ones that have moved into “very safe” territory between August and now; none have defected from “very safe”.
Very Safe Democrat (191): California (100.0%), Colorado (97.3%), Connecticut (99.7%), Delaware (99.9%), District of Columbia (100.0%), Hawaii (96.4%), Illinois (99.9%), Maine-1 (99.3%), Massachusetts (99.8%), Maryland (100.0%), New Jersey (99.4%), New York (99.8%), Oregon (97.2%), Rhode Island (99.3%), Vermont (99.8%), Washington (99.7%)
Very Safe Republican (116): Alabama (99.8%), Arkansas (99.6%), Idaho (99.9%), Indiana (99.9%), Kansas (99.8%), Kentucky (99.9%), Louisiana (99.3%), Missouri (99.9%), Montana (98.0%), Nebraska at large (99.8%), Nebraska-1 (99.8%), Nebraska-3 (99.9%), North Dakota (99.2%), Oklahoma (100.0%), South Carolina (98.7%), South Dakota (99.2%), Tennessee (100.0%), Utah (99.1%), West Virginia (99.7%), Wyoming (99.9%)
Now, from Trump’s perspective, the path to get from 116 towards 270 starts like this, in descending likelihood order:
- Ohio, 95.1% (133)
- Texas, 95.0% (173)
- Mississippi, 94.9% (179)
- Florida, 92.5% (209)
- Iowa, 83.2% (215)
- Alaska, 76.9% (218)
- Maine-2, 76.2% (219)
Whereas from the Harris perspective, the path from 191 towards 270 starts like this:
- Virginia, 93.9% (204)
- Nebraska-2, 93.0% (205)
- Minnesota, 90.7% (215)
- Maine at large, 87.3% (217)
- New Mexico, 87.0% (222)
- New Hampshire, 73.5% (226)
As such, before we get into the 7 canonical swing states, if everything breaks according to expectation Harris would have a 226-219 lead.
The 7 swing states, ordered from Trump’s perspective:
- Arizona, 71.7% (230)
- North Carolina, 61.3% (246)
- Georgia, 61.0% (262)
- Nevada, 53.1% (268)
- Pennsylvania, 52.8% (287)
- Wisconsin, 40.9% (297)
- Michigan, 38.5% (312)
And the same 7 states, but ordered from the Harris perspective:
- Michigan, 61.5% (241)
- Wisconsin, 59.1% (251)
- Pennsylvania, 47.2% (270)
- Nevada, 46.9% (276)
- Georgia, 39.0% (292)
- North Carolina, 38.7% (308)
- Arizona, 28.3% (319)
For 6 of the 7 swing states, the expected margin of victory from Silver’s model is 1.2% or less; the exception is Arizona, where Silver now sees Trump as a 2.4% favorite. As such, if we posit a 1.5% “uniform swing” towards Harris relative to the model’s expectation, Harris would win 308-230 (with Trump’s only swing-state victory being Arizona). On the other hand if we instead posit a 1.5% uniform swing towards Trump, then Trump would win 312-226 (with Harris losing all 7 swing states).
I’ll end with a fantasy scenario for Harris… Silver’s model has Trump as a 5.2% favorite in Iowa, whereas the Des Moines Register poll has Harris by 3 in Iowa. If there’s something systemic, rather than Iowa-specific, behind that result, then we’d be talking about roughly an 8% uniform swing towards Harris relative to the model; and at that point she’d be winning not only the swing states and Iowa, but also Florida, Maine-2, Ohio, and Texas for a 413-125 landslide. As much as I’d like to believe Harris ought to be competitive in all of those states, that seems pretty rich.