This week I sat down with G. Elliott Morris — who ran FiveThirtyEight before ABC shut it down, and who’s built something better with Strength in Numbers and FiftyPlusOne — to get his read on where the midterms stand and what mainstream polling is missing.
The upshot on the midterms right now, according to Elliott, is that “the cake is pretty much baked” on Republicans losing control of the House, while Republicans only have even odds or worse of keeping the Senate. (I suggested flipping the frame to how Republicans are going to do, since focusing on how Democrats will do encourages a hyperfocus on their messaging instead of understanding how much this will be about rejecting what Trump and MAGA are doing.)
We also covered:
How to decipher polls about competitive red-state Senate races. Just because many Republicans might be disillusioned right now (and thus less likely to want to take a survey) doesn’t mean they won’t vote in the midterms —especially once hundreds of millions are spent on ads reminding them that, say, a vote for Talarico is a vote for a Democratic Senate. (And Republican Super PAC money will unfortunately go a lot farther now, thanks to the Roberts Court.) I argued that any credible survey of competitive races right now should highlight two factors that mostly go ignored: 1) which party respondents want to see control the Senate, and 2) vote history — specifically, what percentage of the modeled electorate voted in previous election. A few percentage points difference in turnout from newer voters, especially young people, will make or break these key red-state races.
How the questions in most polls are written to fit the frame of stories political reporters want to write (who’s up and who’s down, which party is more trusted on which “issue”) rather than to genuinely understand what drives voters’ motivations and decisions. A classic example is how pollsters asked people’s opinions about Trump on “immigration” in 2024 without asking about the horrifying specifics of his proposals that had already been reported on, and that fueled massive backlash once they were actually implemented. Elliott’s own polling experiment on the Kilmar Abrego Garcia case predicted the size of the later drop in public support for Trump’s brutal deportation regime.
The apparent fatal flaws in the famous political science finding that politicians only respond to the preferences of the rich, which Elliott unpacked in a recent interview with political scientist Peter Enns. The good news is that it really does matter what the majority of us want and fight for really does matter; the complicated news is that the ultra-rich absolutely do distort our politics in ways that are obvious if you just look out the window.
The limitations of saying things like “9 in 10 odds” in reference to election results, which don’t have fixed probabilities like dice. There’s an interesting parallel between the limits of historical election data and the breakthrough in weather forecasting depicted in the movie Pressure.
Elliott’s new baby! Congratulate him on fatherhood by subscribing to his publications if you don’t already.
