Times/Siena Poll: We're Doing This Shit Again?
On A Poll That Is Assuredly Wrong
If you believe the New York Times/Siena Generic Ballot poll, Democrats are doing 6% worse with white voters without a degree, 24% worse with Hispanics, and 14% better with whites with a degree.
If you believe the NYT/Siena poll, this is the Global Fucking Realignment on steroids, because whites with a degree – aka, those most likely to have socially liberal views on abortion and gay rights – are running to the left, whereas those with socially conservative views – whites without a degree and Hispanics – are trending right. It’s a pattern accelerated from 2020, where Hispanics bolted right, Democrats made serious inroads with white degreeholders, and the GOP don’t have that much more room with non-degree holding whites, because they are close to their (short-term) ceiling with those voters.
If you believe the poll, of course – which you shouldn’t, because as the headline says, my first reaction to this poll was simple.
“We’re doing this shit again?”
…
In 2020, the Times/Siena poll in Florida had Joe Biden winning whites with a degree by 4%. They had Biden winning whites with a degree by 19% in Iowa. Biden lost them by 13% in Florida won them by 4% in Iowa, a 17% miss in Florida and 15% miss in Iowa. They had the direction of travel correct, but they had the scale completely wrong, a huge miss that accounted for the single biggest source of 2020 polling error. They were not alone in this – not at all alone in this, and in Florida, they were one of the least bad pollsters around.
I’ve said it before that the problem with the polls now is that they can’t poll white people, and it’s 1000% true. What happened in the past was that pollsters have never been able to poll Hispanics, but their errors were consistent – polls underestimated Democrats for years and years in Nevada because of bad Hispanic samples, and so what ended up happening was there was a confidence and a comfortability with the idea that they could “guess” the polling miss. In the Midwest, what would happen in the Obama era was if a pollster was good at their jobs, they’d correctly poll Black voters, and bad pollsters would underestimate turnout and margins with that group. It was a tidy little consensus.
Now? They still can’t poll ethnic minorities for shit – as the 2018 Times/Siena disaster that was their Texas polling shows, as did the Georgia runoff polls that had Ossoff outperforming Biden with white voters but in a close race because the GOP got ~20% of the Black vote in that poll – but they also can’t poll whites anymore. And given that’s the case, you end up in a situation where the right answer might be not using polling anymore.
I’m comfortable with my solution to the problems of polling minorities – I have a lot of faith that I will avoid the sorts of huge errors that wildly volatile subsamples can provide by assuming a minimal change from 2020 for Hispanics and Black voters. But if white voters can’t get polled properly, then it’s highly likely that any sort of analytical approach will fail, and fail easily.
If you just take the average of the post-Roe Generic Ballot polls that released vote intention by race data, Democrats lose white voters by 10%. They lost them by 12% in 2020. Those same polls disapprove of Biden 34/60, a 24% net disapproval. If you think that level of split is sustainable, I don’t even know what to say.
If you use the white approval data alone, it points to a R+5.9 environment. If you just use the white generic ballot data, it points to a D+5.6 environment. If you believe the state polls, it’s a D+5. And this is why any effort to suggest what will be the environment in November is a complete fucking guess, because none of the metrics point the same way. (To be clear, I think the environment will be substantially more Republican than this poll. My issue with it is that it is absolutely overstating Democrats.) And all of this is just giving me 2020 flashbacks, because I know for all the discourse – this column included – we’re tilting at windmills. Like, this poll should be catnip for me – it proves the Global Fucking Realignment! Whites with a degree keep swinging left! Forsyth! Southlake! I’m vindicated! – but it’s also wrong, because it’s wrong in the same way the polls in general and this specific pollster was wrong the last time we saw this.
There is a saying that my civil servant father taught me years ago – SSDD, or Same Shit, Different Day. It was his way of saying that nothing much exciting happened, because it was the same shit every damn day. And that’s where the polls in the US are – it’s just Same Shit, Different Day.
At the end of the day, this poll isn’t a remarkable sign of what the commentariat thinks it is, which is the Global Fucking Realignment by another name, but a remarkable experiment in how willing to believe anything called a poll so many are, and it’s why my only reaction is we’re doing this shit again?