It's been a long time between posts, but I'll be back on a fairly regular schedule now. The New York Times had a story by Nate Cohn on the upcoming midterm elections which noted Trump's low approval ratings. It said "the president’s party almost always fares poorly in midterm elections, but especially when the president’s approval rating is low." That sounds reasonable, but they didn't actually look at the relationship, so I will. If you regress the number of seats held by the president's party in the House of Representatives after the elections on the number held before and the president's approval rating in October (1946-2022 midterms), you get:
Constant -59.4
(45.0)
Before 0.91
(0.13)
Approval 107.5
(46.7)
The predicted value for the 2026 elections is 177 Republican seats, which is below most other forecasts (the consensus seems to be a little over 200). Looking more closely:
The most encouraging example for the Republicans is 2022, when the Democrats lost only a few seats even though Biden had low approval ratings. Probably that was partly due to negative polarization--there were people who had negative views of Biden but just wouldn't vote for a Republican. That undoubtedly still applies today, although in the other direction. The residual for that election is +28, so if that applies in 2026 it brings us near the consensus forecast. But of course, there were idiosyncratic features of the 2022 elections that don't apply today. Also, I suspect that the linkage between presidential approval and midterm results is becoming stronger (ie, people make less distinction between the president and his party). So I'll guess a residual of +10 or +15, giving the Republicans about 190 seats.
