11:17 PM: Over the last couple weeks I’ve noted many times that we would not only have political winners and losers tonight but also pollster winners and losers. And well, we have ’em. The mainstream pollsters’ numbers look pretty solid. The numbers from GOP pollsters like Trafalgar, Insider Advantage, Rasmussen, etc. are taking it on the chin pretty hard. But it goes beyond differences of methodology. There was always a strong assumption from a lot of knowledgable people that those outfits had shoddy verging on bogus polling practices. So you take the mainstream number, tack on 5 points and that’s your poll. If there’s a major polling error for most pollsters you look pretty smart. But it’s sort of like a broken clock being right twice a day. Like I said, many thought this. But they’d gotten closer to the mark in 2016 and 2020. So it was hard to make the argument too confidently.

11:09 PM: Earlier in the evening I mentioned those three Virginia House seat bellwethers. We now have results in each. Luria, the most vulnerable, has conceded defeat. The other two held their seats. In advance this result seemed to presage Republicans winning the House but by a slender margin. And that’s where we seem to be.

10:57 PM: A huge bummer but not unexpected. NBC calls Ohio for JD Vance.

10:48 PM: Don’t want to assume anything else. But Lauren Boebert seems to have a race on her hands.

10:39 PM: I’ve been cautious about this. But The New York Times’ Nate Cohn has been looking at county totals in Pennsylvania and Fetterman is beating Biden’s 2020 numbers basically everywhere. This looks like a pretty strong signal in the noise. We need to see a lot more. But that’s pretty telling.

10:34 PM: NBC calls Virginia’s 7th district for Abigail Spanberger. [10:43 PM: Now AP too. This one’s done.]

10:28 PM: I think we can say with some confidence now that this is not a GOP wave election. But it’s still true that the GOP did not need a wave to win the House or even the Senate. We know a lot more about the outlines of this than we knew a few hours ago.

10:13 PM: To make the point more directly, I’m literally not looking at statewide totals in really any race. Or barely any. Sort of impossible for me to have any idea what they mean. Almost entirely looking at completed or near completed counties. And to be clear, look at what analysts I trust say about them. That’s when you can make comparisons.

10:08 PM: Possibly some encouraging signs for Fetterman in Pennsylvania. Will need to see more. Very early.

10:00 PM: Looking increasingly likely that Rep. Abigail Spanberger will hold on for reelection in Virginia 7. Not a shocker by any means but definitely a bellwether.

9:46 PM: Veteran Dem Rep Marcy Kaptur was supposed to be in some danger. Not likely to lose but in danger in a red wave. Ryan Brune now thinks she’ll win by double digits. It’s not a huge thing that Kaptur is reelected. But if she’s reelected comfortably that’s another result that puts some real bounds on the scale of Republican victories.

9:37 PM: Given all uncertainty created by different methods of voting and when those different buckets of votes are counted, it’s become a much bigger deal not just to look at key counties and precincts but when entire counties are counted. Once a county is fully counted, you can set aside election day voting, early voting and just look at earlier elections. So for instance, if a Democrat is exceeding Biden’s margins in a state Biden won that’s a clear signal, regardless of what the statewide totals are. Right now that kind of analysis is making New Hampshire look like a hold for Democrats and North Carolina a hold for Republicans. Again, broadly in line with the polls.

9:19 PM: Here’s one metric to see the night so far. Election watchers were looking at three House races in Virginia: Incumbent Dems Reps. Luria, Spanberger and Wexton. Luria was the most endangered. If she held on, significant GOP underperformance and conceivably even a Dem House hold. In a wave, Spanberger would lose too. In a total blowout even Wexton might go down. At the moment, Spanberger is looking like she likely, but by no means definitely, will hold on. And Luria is still in it. With those two races close, I assume Wexton is in good shape. Big picture that looks like GOP House pick up by probably not a big majority. Again, very early. Virginia’s one state. But this starts to give us a general sense of where we might be headed.

9:14 PM: You wouldn’t expect there to be much concern about a congressional race in Rhode Island. But there was a lot of Democratic fretting about Seth Magaziner in RI-2. The local affiliate has just called that race for Magaziner. So that’s a D hold.

9:05 PM: We’re still very early and at a point where it’s difficult to get an outline of big pictures. Having said that, what seems clear so far is that the polls were broadly accurate. By which I mean, the close races are super close. Maybe that’s not the hugest headline. But it’s a data point that’s still pretty key. On something like the Senate I think we’re still very much in jump ball territory. The races that were supposed to be super close appear to be super close. And we don’t know who is winning them.

8:58 PM: Fairly encouraging numbers for Democrats in New Hampshire. They’re running ahead of Biden’s 2020 numbers, at least in some key areas.

8:15 PM: We have some possibly encouraging reporting for Abigail Spanberger in VA-7. I want to be clear what my metric is here. To me, if a Dem who was supposed to go down in a GOP wave is holding there own, that’s encouraging. Doesn’t mean they’re going to win. But a good sign for them and gives us some hints about the broader story. More on this in a moment.

8:00 PM: So here’s something interesting. In the Georgia Senate race, a lot of the Walker-leaning polls showed him getting like 25% of the African-American vote. That would be super high for a Republican. And most political observers were skeptical of that. Early results throw a lot of cold water on that prediction. So that puts a dent in the credibility of a lot of the polls that showed Walker strength. Again, doesn’t mean Warnock wins. But it’s an encouraging sign for him.

7:52 PM: We’ve talked a lot about “red mirages”, same day votes showing big GOP wins then slowing giving way to early Dem-leaning votes. That pattern, which we saw in spades in 2020, is a big deal tonight too. But there’s more than that. We knew in 2020 that early versus same day was hugely, hugely polarized by party. It still is to a great degree. But we don’t know just how much. That adds another wrinkle of uncertainty. So just keep that in mind. That’s one reason I suspect we’re not seeing a lot of early calls. It doesn’t mean races are necessarily close. It just really complicates the way race-callers look at results and infer where things are going.

7:36 PM: Here are the five states where abortion is literally on the ballot. TPM Alum Alice Ollstein (we have so many amazing alums) has this deep dive on the referendum in Kentucky, which seems to be really struggling (i.e., the anti-abortion referendum is struggling.)

7:32 PM: In Georgia the Senate race looks about as tight as the polls have been suggesting for some time. Meanwhile, there’s a big divergence between Kemp and Walker, which is again just what the polls have shown for a long time. So the early look at Georgia says it’s pretty much what the polls have been predicting. Worth noting that at this point in the evening “close” may not be that close in the final result.

7:26 PM: The only signal I’ve been able to pick out from the noise so far is that Florida looks quite rough for Democrats tonight. But the shape of the electorate in Florida does not look like the shape in other states. That does not mean that Dems are having a great night elsewhere. Just that Florida looks like it’s own thing and not necessarily representative of the rest of the country. Again, very, very preliminary reads.

So here we are. The first polls are closing. Remember: whatever happens, history doesn’t end. There’s an up and a down side to that and it’s good not to get too bound up in either.

Twice in recent history the President’s party has gained seats in a midterm election. The last time was 20 years ago. For years, I was unreasonably pleased with myself that I was one of the few people I knew who predicted Democrats would gain House seats in the 1998 midterm — the Lewinsky scandal/impeachment midterm. It wasn’t any great insight. That was what the polls actually said. They said it very clearly. But the political class had convinced themselves that the polls were not accounting for the intensity of evangelical Republicans who would swamp the Democrats at the polls. That didn’t happen. The polls had said as much.

I learned a lesson from this that, for a political observer, polls are something like a pilot’s control panel when flying in stormy weather. What pilots are taught is in stormy weather you watch the instrument panel and absolutely disregard everything else. Your sense of direction — your feel for up and down — will fool you, and then you crash. The instrument panel is right.

There’s a real oddity for us, or at least to me, about this moment in our history. “Vote fraud” propaganda and voter suppression have been central issues in TPM’s coverage literally from day one, going back 22 years now. So all the different misleading and disingenuous games, the ways to fool people about alleged fraud, the ways to use the numbers to bamboozle people — we’ve been writing about all of these for years. But now what was a matter of relatively marginal mainstream media interest is the very center of the story. In stories like the one below and this one yesterday we see the same old — really, ancient story — “vote fraud” propaganda and suppression tactics targeting largely African-American cities — in this case Detroit and Philadelphia. So, from an editorial point of view, we find ourselves explaining to longtime readers points we were explaining and reporting on back in 2002 and 2006 and 2016.

This morning brings fresh news that in response to a GOP lawsuit, vote counting in Philadelphia will slow down dramatically. Said the city’s sole Republican elections commissioner: “I want to be very clear that when there are conversations that occur later this evening about whether or not Philadelphia has counted all of their ballots that the reason that some ballots would not be counted is that Republicans targeted Philadelphia — and only Philadelphia — to force us to conduct a procedure that no other county does.”

I’ve had numerous people ask me over the last 48 hours what I expect in the 2022 election. I’ve told them that I am generally pessimistic but also highly uncertain. Indeed, I’ve worked this uncanny combination through in my head so many times I’m not even sure what the combination means anymore. Big picture it seems like Republicans have a good night. But there continue to be a lot of discordant pieces of data that don’t quite fit. If Democrats were to have a good or better than expected result we’d look back at those discordant data points and think, “Okay, here were the signs people were ignoring.”

With the election one day away I wanted to take a look at the latest polls. They still don’t tell a totally clear story. Big picture what we see is still much better for Republicans than what we saw in late summer or even as recently as October. All the key Senate races are more or less tied. That means anything from a one or perhaps two seat pick up for the Dems to a four seat pick up for the GOP is entirely plausible. But with all this sobering news we’re not seeing the kind of late polling breakout I might have expected. The generic ballot averages have actually ticked slightly back in Democrats’ direction over the last couple days, though this could well be statistical noise.

There are going to be big political winners and losers on Tuesday. But there are also going to be big polling winners and losers. Through this cycle different classes of pollsters have been seeing a very different race. Wednesday morning we’re going to know who was right and wrong. But here’s an aspect of polling that doesn’t get talked about enough. It’s not just accuracy. There’s another part of this.

We know that polling has gotten both harder to do and more expensive to do as fewer people respond to polling phone calls. If one out of twenty calls gets answered, that’s a much more expensive proposition than if one out of three does. So non-response has been driving up the costs of polling, and that’s overwhelmingly hitting the pollsters who use live callers. Live calls are generally considered the most accurate, though it’s far from certain whether that’s still the case. Non-response also puts accuracy under growing strain because pollsters need to make sense of which political groupings are more or less likely to respond. If non-response is identical across all political affiliations it’s not a problem from an accuracy perspective. But that’s almost certainly not the case.