Joel Wertheimer

Democrats absolutely cleaning up about irregular voters in Clark County.

So I did bad data analysis earlier trying to recreate this but with vote history. I deleted all the posts about it because what was getting captured in was just the large secular decline in Dem RV advantage in NC. So I redid it and I’m hoping I can get earlier data to improve but…

I sort of think the actual outcome is more interesting. EV data in aggregate doesn’t tell us much at all and EV data from regular voters doesn’t tell us much bc definitionally they eventually almost all turnout. But there is signal in irregular voter turnout starting next week at least in NC!

Funny enough I'd also been building my own NC dashboard. I focused on splitting voters into propensity quartiles and doing this as rate of registered voters. The high propensity voters will eventually turn out regardless, but you can learn something from the low propensity ones I think.

My firm belief is that Susan Collins is going to significantly overperform the fundamentals in Maine and lose.

This is completely correct. In a world where Democrats can do everything they want on permitting in the future (trifecta, kill filibuster) then the relevant question about the current permitting bill is only whether it's better than the status quo, not an imagined better bill.

Like, imagine Republicans and Democrats agreed to a bill tomorrow to increase SNAP funding that Trump would sign. On what planet would people say we need to wait because Democrats might have a trifecta in the future and can do what we want?

Unfortunately Maine does not give public voter file full history to see how low propensity voters are doing, but somewhat comically more Democrats have voted already this year than in 2024 at the same time. Additionally, we can see Dems have a lot more early voters who never voted absentee/early.

Will try to build this out some only have ME and NC right now, but basically the goal is to monitor how well each party is turning out low propensity voters. The 3/3s make up the bulk of EV but don't tell us much. joelwfllp.github.io/early-vote-t...

I am perfectly willing to believe NY is continuing to move right relative to country but I am struggling with the electorates that the top pollsters are pumping out. They seem to be targeting a 2022 electorate and I don't really understand why. Here's recent elections and RVs vs Marist and Siena

What we seem to be hearing from pollsters is that they're having a really hard time reaching rural Republicans so we might be getting too many (sub)urban GOPers and too many rural Dems but it's pretty unclear to me which way that cuts.

I am perfectly willing to believe NY is continuing to move right relative to country but I am struggling with the electorates that the top pollsters are pumping out. They seem to be targeting a 2022 electorate and I don't really understand why. Here's recent elections and RVs vs Marist and Siena

Similarly, if you had too many educated Republicans and too few educated Democrats you missed something, even if you had the right number of college ed/non-college and GOP/Dems on the topline. Getting those wrong make it more likely you'll miss swingy demographics.

I am perfectly willing to believe NY is continuing to move right relative to country but I am struggling with the electorates that the top pollsters are pumping out. They seem to be targeting a 2022 electorate and I don't really understand why. Here's recent elections and RVs vs Marist and Siena

Of course, lots of polls in 2020 weighed to 2016 recalled vote and still got things very wrong, so what the hell happened there? Well, let's say you were a Trump voter in 2016 and you were very COVID cautious or a Clinton voter who hated COVID lockdowns? You were among the most likely switchers!

I am perfectly willing to believe NY is continuing to move right relative to country but I am struggling with the electorates that the top pollsters are pumping out. They seem to be targeting a 2022 electorate and I don't really understand why. Here's recent elections and RVs vs Marist and Siena

(3) update the probability of the person voting based on their vote history and demographics using self professed likelihood of voting (this varies by demographics, e.g. young people overestimate more). Read @bfschaffner.bsky.social et al. academic.oup.com/poq/article-...

I am perfectly willing to believe NY is continuing to move right relative to country but I am struggling with the electorates that the top pollsters are pumping out. They seem to be targeting a 2022 electorate and I don't really understand why. Here's recent elections and RVs vs Marist and Siena

Ok here's what I think is happening: pollsters are all taking shortcuts to guess at the electorate because they're afraid to model it themselves anymore basically even the ones using a voter file who could do a good job of it. bsky.app/profile/chic...

Thought this was very good

I have been seeing a lot of people not trusting the polls in the Midwest, which is totally fair given recent history. But if all you knew were Trump's national net approval, there would be absolutely nothing weird about any of the polling in any race right now. wertwhile.substack.com/p/on-polling...