I woke up very early today.1 Believe it or not, I get more sleep than you might assume — and the extra hour helped — but I’m no longer as much of a night owl as I pretend to be. One thing I’ve always had trouble doing, though, is sleeping when I have some sort of non-negotiable commitment in the morning. Like, if I have to wake up at 6 a.m. to catch an 8 a.m. flight at LaGuardia2, my body clock might wake me up at 4:45 instead.
Maybe it was because I was anticipating a new round of New York Times/Siena University polls, which are sometimes published at 5 a.m. The morning turned into an anticlimax, without much new data. But NYT/Siena released their presumably final set of polls this evening, and they showed what can only be described as fantastic numbers for Democrats, putting them ahead — sometimes narrowly, sometimes not — in basically every Senate battleground and in the associated governor’s races, including in Texas:
These polls weren’t quite in the territory of FLIPR’s officially “weird” scenarios, but they're close to that boundary. Even if Democrats were to lose a couple of these races — and there’s actually some bad news for them here in Alaska, I think — absolutely no Democrat is going to complain if they only flip the top five races in this chart, especially if it comes along with the Texas governorship.
The Times/Siena polls are also not that surprising, given that the NYT shows a D +11 on the generic ballot and that, in most cases, the trends in their polls follow the patterns FLIPR has previously detected. Although the model trusts the NYT/Siena polls a lot — they’re among our more highly-rated pollsters — it does adjust every survey firm’s numbers for house effects. In other words, it knows that NYT/Siena has tended to have good numbers for Democrats this cycle, and that’s priced into its forecasts, including to some extent even in races that the NYT hasn’t polled.
The other reason I didn’t sleep so soundly last night, I’ll admit, is because I have a fair bit of anxiety about the election. In the Senate, FLIPR has sort of been teetering on the boundary between a “Lean D” and a “Likely D” forecast — formally, this is defined as a 75 percent chance of winning — in a way that felt slightly uncomfortable.
I don’t want to get too “meta” here — there’s been enough of that in the newsletter — but we were sort of in an ambiguous place where it wasn’t quite clear whether pollsters and modelers should “own” the fact that they were projecting a big night for Democrats — and own the risk of another embarrassing systematic polling error. The NYT/Siena polls resolve that, in my opinion, in Democrats’ favor, for better or worse. Polls show that Democrats will win the Senate, perhaps soundly. They might be wrong or they might predict slightly the wrong states to flip. No currently R-held state other than North Carolina counts as “Solid D”.
But Democrats could actually absorb a modest-sized polling error and perhaps still prevail given the latest round of surveys, including some that came in after our deadline tonight and so haven’t even hit FLIPR yet.
It was just barely — the NYT/Siena polls actually had less of an impact on our model than I expected — but for the first time tonight, Democrats’ Senate odds improved to a number that begins with an “8”.
It’s something of a point of pride for Silver Bulletin that we rarely focus much on individual surveys. To be fair, this isn’t exactly an individual survey. The Times surveyed seven states, after having already polled some of the more borderline battlegrounds, namely Minnesota, Florida, South Carolina and Nebraska. NYT/Siena had conspicuously poor numbers for Dems in the Cornhusker State, with Republican incumbent Pete Ricketts up by 6 points, higher than even in recent GOP internals. And NYT/Siena isn’t biting on Florida, where they had Republican Ashley Moody ahead by 6 points. Otherwise, though — yes, we’ll go through another express tour of the seven states the NYT polled tonight — their numbers have been very strong for Democrats.