In the 2022 generals, the consensus was that we were headed for a red wave. The polls suggested that we’d lose in places like PA and GA and that we’d lose the house by a dozen seats or more. There were others (including me — see here , here, here, here, here, here, here and here for receipts) who noted that what was really going on was that there were a half dozen rightwing partisan pollsters who flooded the zone and shifted the averages.
So far, polling for 2024 seems more dire, at least on the surface, because it’s not one or two pollsters suggesting a Trump win, it’s pretty much all of them. Even the recent piece in The Economist that made everyone so optimistic, the one that says that the highest quality polls show a tighter race, still have Biden down by a point or two in the popular vote, roughly 6 points worse than 2020 — and a deficit that would destroy us in the electoral college.
It’s still early, I suppose, but by this point in 2020, Biden was ahead by about 4 points, and maintained that lead or higher throughout the cycle. Likewise, Clinton was leading Trump in roughly 80% of the national polling taken in Feb 2016. So it’s early, but it’s not that early.
On the other hand, what if there’s something systematically wrong about all polling? It’s easy to hypothesize that, for example, you have a large-ish group of people who are strategically responding to polls to try to push the Dems to nominate someone else, either because of Biden’s age or Gaza or something else, but who will absolutely vote for him in the general. Or that (like 2016 and 2020, but in the opposite direction), the polls are systematically off because there are specific subgroups who are under-responding.
I’ve already noted (here and here) that when you look at a lot of the polling, it doesn’t pass the sniff test. Black voters, for example, have given Dems a 70+ point margin in every single election since 1984, but we’re now supposed to believe that the guy who just got 96% in the South Carolina primary, got his big bump in South Carolina in 2020, and has a biracial VP is going to get only a 40% advantage against a virulent racist (even by Republican standards)? I don’t really think so.
Rather than just wish-casting that there may be something wrong, I think we now have some serious evidence in the form of the NY-03 results. First, here’s all of the public polling:
Three pollsters, three polls, definitely some names you’ve heard of. A prediction of Suozzi +3 (or so). And as of this writing, he’s winning by 17 with about half the votes in, and it seems likely that he’ll win by double digits in the end. This is not a small win.
Or consider the generic ballot, including from November:
For the last year, Dems have never led by more than a point or two in the GCB, and by the November election, the 538 average had the generic ballot tied.
Ignoring special circumstances like the Kentucky governorship (for which you can credit Beshear personally) or the abortion referenda in places like Ohio, even in generic races like in Virginia we won both houses, and in Pennsylvania, a statewide race for state supreme court (as generic a race if there ever was, especially since it wouldn’t flip control), the Dem won by 5 points. More generally, as our own DKos tracker has noted, we’ve improved by an average of 3.3% since the 2020 presidential election — an election, I remind you, that we won by 4.5% nationally. That’s consistent with a roughly 8 point lead nationally (not including the PA State House and the NY-3 seat being decided tonight).
We’re even seeing similar effects in the primaries. Take this pre-election polling for the Dem primaries:
In New Hampshire, Biden was polling at 56, but actually got 64 as a write in candidate. In South Carolina, the one poll had him at 69, with a 64 point lead — but he actually got 96! In Nevada, again, only 1 poll, with him at 78, with a 76 point lead, but he actually got 90 (though to Williamson’s credit, she too smashed her 2% polling by getting 3%).
In other words, it’s not just Dems generically, but Uncle Joe specifically who seems to be performing better — much better — than he’s polling.
And on the Trump side, it’s the inverse. Leaving aside Nevada (which given the caucus/primary mess is basically impossible to disentangle), we’ve got:
Iowa, in which the polling had Trump at 53, and a combined Haley and DeSantis at 34. On caucus night, it was Trump 51, H+D=40. Not a huge underperformance, but an underperformance nevertheless. (And mind you, this was during a huge snowstorm, and a caucus, in which Trump’s zealots were expected to come out in relatively greater force).
New Hampshire, in which the polling had Trump at 54, Haley at 36, an 18 point margin. But while Trump got actually did get 54, his margin over Haley was only 9.
Look, I don’t know what November will bring, but I think it’s foolish to think that today’s result in New York is just a blip. We’ve been seeing months and months of Dems (and more specifically Biden) significantly overperforming polls, and Republicans (and more specifically Trump) underperforming theirs.