It was supposed to be a split between trust in the military and support for military spending, I thought. So owls trust the military but want lower spending, and gulls don’t trust the military but want more spending. One would hope there aren’t a lot of them!
Sex and the City meme, "And I couldn't help but wonder who assigned so many things to grade?"
Democrats in NJ were definitely traumatized by past presidential elections- hence the worried vibes.
(And that underestimation resulted entirely from the high weight given to 2024 turnout in the LV model)
The issue in NJ was the composition of the electorate: we were basically averaging the electorate from 21 and 24, and wound up with a LV model that was too Republican. This isn’t an issue in VA, where they had a statewide last year to draw from.
The vibes were that it was a coin toss: the high quality polls (like mine) had Sherrill up by 8-9. We underestimated her support, but the vibes were completely wrong.
I think it’s entirely plausible that the Secret Service killed Kennedy (completely by accident: hung over, no safeties on their weapons, etc).
We had a similar split in NJ last year: polls showed one thing, vibes another.
The polls were right. The high quality ones generally are.
They really seem to think that “we denied the allegation and they printed it anyway” is the same thing as actual malice.
This happened here in New Jersey, when the state government had to update their unemployment benefits software.
Which was in COBOL.
In 2020.
It’s easy to say that gender is a performance, but it’s only when those performances get really extreme, especially among men, that we start to notice them.
He also published my first book- giving it perhaps more support than it deserved. He’ll be deeply missed.
My friend and colleague, Harry Keyishian died recently: his refusal to sign a McCarthy-esque loyalty oath led to a major legal victory for academic freedom.
I need to dig into the trust/optimism/utility questions on AI: lower trust/higher skepticism among non-whites makes sense as a function of social trust.
I’ve been working with gen pop samples, and the dynamics on AI adoption and use are real weird. Big effects of gender and sex, and I’m not sure how those interact with technical interest/skill (have an upcoming meeting to get that data). Doesn’t fit neatly into the narrative I was expecting.
This is also about the only example of a gender equal policy coming out of the administration: both men and women are going to extreme lengths to alter their appearance and look a certain way to appeal to Trump.
There came a time when the old gods died! The brave died with cunning! The noble perished, locked in battle with unleashed evil! It was the last day for them! An ancient era was passing in fiery holocaust!
I don't think anyone expects this race to be very competitive in November - worst case, I think it's Mejia +10, rather than Mejia +20. I think the plan is to trade voters who don't like her for otherwise marginal or less excited voters who do.
The second I heard the first "signature move" tiny canon about Bob Ross, I knew that this one had to go in.
(Also, I'm now 1 for 5 in canon/nonac/tiny canon submissions, soon to be 1 for 6).
They’re also likely to try and use them against any candidate nationally where they might plausibly stick. So, I hope voters liked those ads, because they’re going to be seeing a lot more of them.
Maybe the most important thing here is that strategists – Democrats and Republicans alike – are going to look at these results and conclude that the attacks on Mejia worked, and ramp them up in November.
And, at the same time, lost the votes of some voters who are registered Democrats, but didn’t like her in particular. I’m not convinced that any of the other plausible candidates for the Democrats would have done better, even if they had won a bigger proportion of Dems in Millburn & Livingston.
If we had good independent polling of the race, or exit polls, we could tell more about what was driving vote choice, but without that, we’re having to make a lot of guesses. I think it’s likely that Mejia turned out some cohorts of Democrats and independents who otherwise wouldn’t have voted.
This means that we're not looking just at different numerators, but different denominators. A candidate who did much better in areas with large Jewish populations would have turned out a different electorate, and maybe won by more, or by less. There's not a clear comparison.
(This is tricky to model, but is literally the textbook example of a nested logit). So while our hypothetical Dem candidate who won all of the Dems, and didn't face any attacks on Israel, might have won over some voters, they also would have turned out a very different set of voters.
What people are missing is that *given the same electorate* is a load-bearing assumption that makes absolutely no sense. People don't decide, sequentially, to vote, and then decide who to vote for. Rather, they look at the options, and then decide whether or not to vote.
Either way, she's winning, but a hypothetical Dem *given the same electorate* who won all of the Democrats (and a proportional split of the indies) would have done better, trashing Hathaway by ~25, rather than by 20.
So, it'a true that Mejis performed much worse in towns like Millburn and Livingston relative to the balance of Democratic and Republican voters who turned out. The electorate in Livingstone (by reg) was D+34, but she won 51-49, for instance.
Folks are getting upset about data showing Mejia underperforming in areas with large Jewish populations (esp Livingston and Millburn). Given the composition of the electorate, a generic Dem who somehow won all Dems would have done ~5 points better.
This is true, and also kind of silly.
