Although we wouldn’t advise worrying too much about the difference between a 52/48 race one way versus a 48/52 race the other way — it’s not a big difference — this wasn’t a good day for Kamala Harris in our model, as Donald Trump is the slight favorite for the first time since August 3.

There’s one big reason for that — Pennsylvania, which is the tipping-point state more than one-third of the time and where it’s been quite a while since we’ve seen a poll showing Harris leading (including two new polls today).

The model is also applying a convention bounce adjustment to Harris’s recent numbers, who has made gains in national polls, and you could argue about whether that’s the right assumption. But the bottom line is that the model has the Electoral College/popular vote gap opening up again, a concern for Harris all along. There’s now a 17 percent chance she wins the popular vote but not the Electoral College, the model estimates.