Contrary to popular belief, the current house map has a substantial pro-Republican bias - Trump won 52.3% of house seats with 50.7% of the vote did 1.5% better in the pivotal 218th seat than he did overall.

It *is* true however that in 2024 congressional Democrats got 49.2% of seats with 48.7% of the two-way vote. But this was because R's ran a bunch of terrible candidates in swing house seats - you can see this effect clearly in gov/senate races too!

@cwarshaw You can explain a lot of it literally with candidate quality indicators in regressions, but the bigger picture thing is that the swing incumbent vs not thing happens consistently for house/senate/gov

@cwarshaw This proves a lot IMO because the set of swing states for governor/senate were pretty weird - hard to construct demographic coalition shift that would make you do better in both swing gov/senate seats *and* swing districts but *only* in incumbent races

@chrislhayes @cwarshaw

@chrislhayes @cwarshaw If this only happened in 2022 it'd be one thing, but then basically the exact same thing happened again in 2024 even though the set of senate/gov states was totally different. It's pretty clearly caused by Trump intervening in swing district primaries imo.

@chrislhayes @cwarshaw I listed this here as one of the things I was most surprised/wrong about from the last five years - it's the big reason why Trump doesn't have 57-60 senators rn.

To get *really* nerdy about why I think it's dumb to use congressional vote as a baseline for partisan gerrymandering, if you regress congress ~ lagged_potus + lagged_congress_minus_lagged_potus you get coefficients of ~ 1 for lagged_potus and 0.6ish for lagged_congress_minus_lagged_potus among incumbent seats pretty consistently for the last 30ish years of congressional elections. For open seats you tend to get 0 for lagged_congress_minus_lagged_potus. So over a multi-cycle timeline just from a predictive perspective any four or six year out prediction would be essentially all potus.

@robertisnthere

@robertisnthere I think it's really not tenable to use anything but potus as a baseline becaue everything else's deviations from potus are highly mean reverting.