Harris is in much better shape than Biden. But she has one big problem.
It’s the Electoral College.
The Silver Bulletin presidential election forecast model relaunched today, and there’s some good news for Democrats: Kamala Harris is in a considerably better position than Joe Biden, who had only a 27 percent chance of winning when he exited the race — and that was probably generous given Biden’s increasingly evident inability to run a normal campaign.
Harris will give Democrats a fighting chance. In fact, she’s a slight favorite over Donald Trump in the popular vote, which Democrats have won in all but one election since 2000. If an election were held today, we’d enter the evening with a lot of uncertainty about the outcome, both because the polling in the pivotal Blue Wall states (Pennsylvania, Michigan and Wisconsin) has been close since Biden dropped out of the race and because there’s some intrinsic uncertainty about where the race stands given how much news there’s been lately.
However, she’s a modest underdog to Trump in the Electoral College, risking a repeat of the popular vote-Electoral College split that cost Democrats the 2000 and 2016 elections. Harris isn’t unique in this regard: Biden also had a large Electoral College-popular vote gap in 2020, barely winning several tipping-point states despite winning the popular vote by 4.5 percentage points. But this is still a problem for Democrats, and we show Harris as having a slightly wider popular vote-Electoral College gap than Biden had in his version of the forecast.
So let me walk you through the model’s logic. The rest of this will be relatively straightforward. I’ll explain how the model gets to its answer, and then I’ll present the bull case and the bear case for Harris (or if you prefer, the bear case and the bull case for Trump). I’m going to avoid too much methodological detail here, which is covered in a separate post that will go up later today or tomorrow1, although the DNA of the model is about 99.5 percent the same as in the Biden-Trump version. And again, for the model itself, just click here: this is a permanent landing page and will update roughly once per day.
How the model thinks about the race
The model starts by creating a “snapshot” of the current state of the race based on state and national polls. Although state polls are ultimately more important in the model’s logic, they’ve been less plentiful than national polls since Biden dropped out, so national polls are probably the place to start.
Harris trails Trump by only 0.4 percentage points in our national average, closer than Biden had been at any point since we turned the Biden-Trump model on in late June. Notably also, she’s getting 44.1 percent of the vote, considerably higher than Biden, who had been hovering between 40 and 41 percent. Fewer Americans are taking a pox-on-both-houses mentality with this matchup, as evident in the declining undecided and third-party vote.
As you can see, the polling average — which we’ve retroactively calculated back to July 1 — has been bouncing around a lot. That’s partly because of all of the crazy political news, and partly because we decided only to use polls conducted fully since the Biden-Trump debate on June 27, the moment at which Biden’s campaign entered a downward spiral and Harris’s candidacy went from a weird hypothetical to highly plausible. It’s also because of sample size: there wasn’t a lot of high-quality polling testing the Harris-Trump matchup until recently. However, we can pretty clearly see that Harris’s numbers have improved since Biden exited the race and she became the presumptive Democratic nominee.