Terrance Woodbury | Can we trust the midterm polls? Three experts weigh in. | On with Kara Swisher

We've had midterms that have been about unpopular wars like '06, unpopular legislation like 2010 when the ACA cost Democrats 63 House seats, and unpopular prices and inflation when Democrats lost the House in '22. But at least in my lifetime, we've never had an election that's been about all three of those things at once.

Thank you for having us. Thanks, Kara. Happy to be here.

So my answer is the shift that didn't happen back toward the Republicans. It's more that, you know, if you look at the polls over the last two midterm elections, it's kind of normal for them to be pretty bullish on Democrats during the summer and then to converge more toward reality as election day approaches, where it's a little bit less of a big wave or Republicans sort of close the gap a little. That narrowing hasn't happened. I have seen in some of my polling Republicans expressing a little more enthusiasm to vote than they were over the summer. You know, Democrats have been fired up for over a year. They have been ready for this midterm, ready to go, where Republicans have been a little more checked out. I don't want to talk about politics. I don't want to think about it. The expectation is that as election day approached and as hundreds of millions of dollars got spent on air, they would tune back in. But that's only happening somewhat, and the polls have not moved in a big Republican direction since we got past Labor Day. So to me, the biggest surprise shift is actually the shift that didn't happen, hasn't happened yet at least. Dave.

Yeah, Kara, for the last decade or so, we've been locked into this mentality that we're down to a very narrow trading range of districts and states that decide our elections and control of the House and Senate. What has surprised me in the last month or so, since we saw even the Republican numbers ebb downwards a bit at the beginning of September as we saw diesel spike, as gas prices spiked, is that we're seeing shocking polling numbers for Democratic opportunities in districts that didn't just vote for Trump by five or 10 points in 2024, but in some cases 15 or even 20 points. The fact that we're talking about Kentucky's sixth district, which is a seat that I thought Republicans had kind of locked the door and thrown away the key 14 years ago. Now we're seeing polls showing a Democrat who's close with Governor Andy Beshear running competitively against a Republican who was former Governor Matt Bevin's running mate many years ago. Bevin is still unpopular in the state, or Elise Stefanik's district in the North Country of New York, where we're seeing a very MAGA Republican who's alienated some of the mainline Republicans in the district tied or even trailing in some polls against a Democratic dairy farmer who really wasn't on the party's radar a year ago. And the fact we're even talking about Kansas or Nebraska as competitive Senate races tells you everything you need to know about how bad things have gotten for Republicans in rural parts of the country. Terrence.

Yeah. So, you know, a lot of pollsters look at voter enthusiasm, right, as a proxy for who's going to actually turn out. And we found that with young voters and with voters of color, that they don't always vote enthusiastically. And so those numbers often trail their actual turnout. Another metric we've been watching is perception of power. We asked the question, regardless of how often you vote, how much power do you believe your vote has to make a difference? And all through 2025, we saw, especially amongst young voters, that those perceptions of power had plummeted well below 50%. And that as voting has started, as we've gotten into voting season and past the campaign season, that those perceptions of power have increased amongst Black voters, amongst young voters, and we've seen that same increase in turnout amongst young voters and voters of color in early voting in places like Georgia and North Carolina. That's very promising.

Specifically amongst younger voters of color, especially in Georgia and North Carolina.

That's exactly right. You know, the higher you perceive the power of your vote, the less likely you are to waste it. It's kind of intuitive, but right.

So we've seen young voters' percentage of power increasing to almost 60%. Young voters of color to almost 60%.

And hopefully that we continue to see their participation.

It's the polling that we're seeing from these individual races, and it's coming in in many cases, you know, five or 10 points more favorable to Democrats than our prior expectations. We're in particular noticing some of these open seats that Republicans are defending slide away from them. And the reason is that Republicans who have established brands independent from Trump, whether it's Susan Collins or Ashley Hinson or Brian Fitzpatrick in Pennsylvania, they're still in the ball game and in some cases ahead, but it's Republicans who are less well-known who are getting dragged down by Trump's low approval. And so in the mountains of western North Carolina, a House seat that Trump carried by about 10 points in 2024, we're seeing the Democrat, a farmer named Jamie Ager, ahead in that district, or in Nevada's second district, or I mentioned Kentucky 6, New York 21. These are all places where Republicans have unexpected migraines. And Republicans had tried to bolster their majority by instructing every red state to eliminate as many Democratic seats as possible through redistricting in the middle of the decade. And yet that has only yielded perhaps three or four additional Republican seats. Democrats only needed to flip three Republican seats to win the House majority. And now we've got more than 20 Republican seats that are toss-up or worse. And so from today's standpoint, I think the median outcome is a Democratic gain somewhere in the 15 to 20 seat range.

The Democrats have to lock down their own seats to win the majority. Michigan is the most vulnerable of them because Abdul El-Sayed has had to expand his base and convince voters that he's an acceptable successor in this open seat, given that Democrats were quite fractured in the primary. But then Democrats have more routes to the Senate majority than they thought they would at the outset of the cycle. North Carolina is the best pickup opportunity for them. They need to win three more Republican seats at least if they hold all their own. And although Maine started out looking like the best next shot, today it's looking like Ohio and Alaska are Democrats' next best opportunities, followed by kind of a tie between Texas and Maine. And then Iowa and Kansas, and Nebraska is the ultimate long shot. But Democrats on a really good night could get up to as many as 54 Senate seats, or they could fall just short by winning 48 or 49.

Yeah, why don't you defend our entire industry? Why don't you start us off? So, first I would say, you know, you mentioned a couple of examples, things like the Des Moines Register poll in 2024. I think for a lot of folks in the field, you could look at that poll and see, you didn't have to look too far under the hood to see, you know, I really don't think that white senior citizen men in Iowa are breaking for Kamala Harris. Like, something is not right here. Where right now what I think is challenging for the industry is the polls have been off in years when Donald Trump is on the ballot. He brings out a type of voter who is very hard to poll. In general, polls have a bias toward overcounting people who are really politically engaged and excited to share their opinion. The good news for political pollsters is that's a helpful bias in a midterm year. It's helping you screen down more closely to those types of folks who actually turn out to vote. And so the midterm polls in 2018 and 2022, while not perfect, were actually pretty good. And nobody should have been surprised by the results on either of those two nights. So the absence of Donald Trump on the ballot, pollsters are hoping means that certain type of voter who just turns out for him, you know, he keeps going around the country in his stump speech saying, 'Pretend like I'm on the ballot.' And pollsters are like, 'No, please don't.' Because when he's on the ballot, those are the years that it's just more and more challenging. So I think this year when I look under the hood, both certainly at my own polls, but at other people's polls, there's nothing obvious that's wrong. If we assume that Democrats are a little bit more fired up than Republicans, but otherwise, you know, have the right balance of folks with college degrees and without college degrees, Republicans and Democrats.

Well, there will always be outlier polls. But one of the things that I am apprehensive about this time around is not that you've got this huge secret Trump vote or whatever that's staying home, but rather when we look under the hood at our data, in many cases, the Democrats are the most eager to take the polls. Like, they're the first ones showing up in the data file. And so you as a thoughtful analyst and researcher and pollster can do things to correct for that. But what worries me, I'm not saying I think there's going to be a massive polling error, but the thing that does keep me up at night is at what point does the type of polling error that affects Trump elections begin to bleed into a midterm like this? And is there, in a state like Iowa, does it just, you know, ultimately there are folks who are never going to vote for a Democrat, but they're also never going to talk to a pollster, and we can do all kinds of things to try to correct for it, but that's the one thing that makes me a little bit nervous.

Yeah, I mean, I'll double down on that. You know, the Trump effect I think has been the largest impact on the kind of variance amongst polling, and I think that Trump effect comes in two ways. One is in surging more rural white voters than we included in our sample. But the other is in the swing. It's not just in who votes, but who they're voting for. That young people, and young people of color particularly, and men of color, are significantly more likely to vote for Donald Trump than they are to vote for other Republican candidates. In fact, from 2020 to 2024, we saw a 16-point swing amongst men of color where even Hispanic men under the age of 50, a majority supported Donald Trump for the first time, I think, of any Republican candidate in my lifetime. But that is a uniquely Trump effect that other Republican candidates are not winning 50% plus of Hispanic men, especially Hispanic men under the age of 50. And so there's one more effect here, and that is that Trump's approval rating amongst these same voters has plummeted, right? I call them Trump-curious voters. They didn't like either candidate, but they were willing to give Trump a shot. Well, now they not only dislike him, but they disapprove of the job that he's done. And so it's kind of returning to this medium. And I think that is hopefully contributing to more... They're not attracted to, like, ah, John Kennedy, what a charming Foghorn Leghorn he is. Ken Paxton, right? Ken Paxton doesn't have the same appeal to young Black men or young Hispanic men that Donald Trump did.

Yes, Kara. If I could also just jump in, though, and say, in addition to, you know, I just made the case for why the polls could be off in a direction that is underestimating Republicans, and that in some of these races where it's looking like Democrats are ahead, maybe they revert back to some kind of norm. There's also an alternate, you know, polling error doesn't always break the same way. In Virginia in 2025, I am proud that Echelon's polls were among the closest when it came to Abigail Spanberger winning by double digits, but even we on the down-ballot races didn't necessarily get it right. We had modeled the electorate correctly, but we underestimated Jay Jones versus Jason Miyares because we had a bunch of undecideds in our poll. And in the end, those undecideds all broke for the Democrats. So smart consumers of polls look at what percentage of a sample is undecided, because even if a pollster has modeled the electorate correctly, you don't know what those undecideds are going to do. And that could have huge implications for the margin in a race. And to Terrence's point, there were an awful lot of voters, including young men, young voters of color, who rehired Trump to the White House in 2024 because they were nostalgic for what things cost during his first term or they believed his promises to lower prices. And now we're seeing massive reversion to Democrats. And that's what contributes to this back-to-the-future dynamic where some of the places where we've seen the most Democratic upside are the places that had swung towards Trump the most over the last 10 years. But this is still fluid as well. And Terrence alluded to this. We're seeing in our surveys of battleground districts, there are around eight or nine percent of voters who dislike the job that the president's doing, but are not yet in Democrats' column and are still somewhat up in the air on their congressional vote choice. And these voters are twice as likely to be under the age of 40. They're a little less white than the rest of the electorate. And although they have very negative views of the left of the Democratic Party, AOC and Democratic Socialists, they also don't have overwhelmingly positive views of capitalism. They favor capitalism 58 to 38. And the one critical pain point that threatens to cleave these voters from Republicans ultimately is high gas prices.

I think we're still early in elections with betting markets. I think they remain a very unregulated industry where folks with inside information are still trading and can put a finger on the scale. And so I am highly suspicious of the betting markets. I think it's yet to be determined what impact they will have on these elections. But I think after this election, we'll see how accurate they are and how much people are putting stock into them as prediction markets and not just betting markets.

So I don't view prediction markets as competition for what a pollster does. You know, candidates or clients are not hiring pollsters to tell them who is going to win an election. They're hiring them to tell them how to shape an outcome. What can I say or do strategically to make a difference? They don't want to predict the future. They want to write the future. And so to me, I view prediction markets in some ways as another use case for polls. There was that famous case in 2024 of a guy who bet on Trump winning on, I think it was Polymarket, and you know, he had commissioned his own polls to figure out how to make that bet on Trump against the conventional wisdom. So I don't view them as competitive. I view them as a different way that someone who really wants to look at what the future is going to hold to engage with that and bet money on it if they want to, or invest, or whatever terminology we're going to use. But it's very different than what a pollster is doing for a candidate or a client.

Yeah, we're in a unique vantage point as political analysts and handicappers because for a long time there's been kind of a two-tiered polling knowledge structure where you've got the public polls, right, that are increasingly rare in House races because media doesn't have the budget to do what they used to. But then we also are looking at a lot of data that the parties and super PACs are compiling, right? Because they're polling these races constantly to try and make resource allocation decisions. And we build those source relationships. But to an armchair political better, they might not be privy to that kind of data, or, you know, we don't know how many people who are privy to that data are making big bets on these races, and there's nothing that anyone can really do to police it at this juncture. For us, yes, we see an increase in the number of outlets that are citing prediction markets instead of traditional political handicappers, but at the same time, we are getting subscription interest from people who are actively wagering on these races, and so they need to know it kind of evens out.

I think the way he ended the question is actually the answer to the question, right? Is that the voters that we need to mobilize the most, you know, these kind of less likely midterm voters, voters that participate in every presidential but don't always participate in midterms, those are the voters that will determine the outcome of this thing. Which set of those voters show up. Those voters don't trust politicians, they don't trust celebrities, they don't trust surrogates, they don't trust the pastors, they trust each other, right? I mean, we ask this question all the time: who, on this list of people, do you trust the most? And the one that's always at the top are people like me or my friends and family and peers, which is why we have to continue to arm and mobilize young people to be their own messengers, but also where these kind of influencer programs come in. That if you're able to arm the influencers that they have selected to follow and selected to get information from, that those, exactly right, by following, selecting, liking, and resharing, they are demonstrating their trust of that information. And so those are frankly just the most trusted messages of the people, like friends and family.

Yeah, I agree that there's not one single answer to this question. I think one thing that I love about my job is that I get to talk to folks who are inside the political bubble all day and remind them of what people outside the political bubble really think, feel, how they're really getting information. And the answer is, I mean, whether it's, you know, I don't want to just say influencers because that feels very silly, right? Like, oh, it's just somebody who pops up a live stream and says whatever they want and they're moving the needle. And for a lot of voters, older voters are still watching a lot of local TV news. I mean, I don't want to be totally dismissive of traditional media, but I do think that there's not one single answer anymore. And that folks, when they hear something that sort of confirms their priors, it's not as though, ah, I trust this person therefore I believe what they're saying, but it's that you hear a little bit of like, oh, I heard somebody say this thing that confirms what I already kind of believe, and therefore they must be a trustworthy person.

I think this year demonstrates the advantage of being the out party because Democrats have the ability to run candidates in every state and district who may vary in their viewpoints, but kind of have a better idea of what resonates and what media breaks through in a rural district or a more college town type of district. Whereas the message from the top on the Republican side is that the economy is great, it's never been better, and it's setting up this cognitive dissonance with voters who can't unsee the prices they're paying at the grocery store or the pump. And so that distrust is real.

The Democratic brand's not good. It's, you know, roughly 30% favorability, and the irony is, you know, not even Democrats like Democrats at the moment. But there's an enthusiasm gap, and 70% of Democrats in our most recent survey of the 37 battleground districts rate their election interest as a 10 out of 10 compared to 58% of Republicans. And for Democrats, this brand issue is really a 2028 problem. They've got to hash it out in the primaries and determine who the face of their party is going to be. But Donald Trump continues to insert himself into this election in ways that are taking the spotlight away from Democrats and on to himself. And keep in mind that he's underwater in our tracker by 24 points. 37 to 61, approve, disapprove.

And the generic congressional ballot, you know, take Nate Silver's average, any other average out there, has Democrats ahead 50 to 41. Would you, if you're Republicans, would you rather be fighting out this election at a nine-point deficit or a 24-point deficit? Because every time Donald Trump hits a midterm convention or promises $5,000 checks or, you know, whatever he's up to that day, it's shifting it back towards that negative 24 dimension.

Yes. So there's two different types of questions that I've asked to try to gauge this. One shows the enthusiasm gap has closed a little bit, and that's simply where we ask people, how motivated are you to vote? The percentage of voters who say they are extremely or very motivated to vote used to favor Democrats earlier this year by a 10 to 15 point margin. That's now shrunk somewhat to where Republicans are now saying they're more motivated to vote than they were a few months ago. However, we also ask a question about what do you think the stakes of this midterm are? And so for those in our poll who said they're planning to, they lean more toward voting for Democrats, we said, you know, how important is it that Democrats win this election and, you know, stop Donald Trump from implementing his agenda? And Democrats are like through the roof on enthusiasm on that question. It is extremely important. When we ask Republicans an inversion of that question, so it's not the same, but we say, 'How important is it to you that Republicans maintain control of Congress and continue to help Donald Trump implement his agenda?' It's like 16 points, I think, lower among Republicans saying that they think that's an extremely important thing to do. So that doesn't mean that they won't vote. Someone who feels meh about things can still cast a vote that counts the same as someone who's really fired up, but ultimately, you'd rather have it be easier to get your folks out to the polls. You'd rather them want to go than feel that they're being dragged.

That is not what you want from your base.

Yeah. I mean, as many of those Trump-curious voters have become quite Trump-furious. I mean, I think many of the Democratic candidates I'm working for are welcoming Air Force One into places like Georgia and Texas. Please come more often. We would love to have you here every day because it is... That's exactly right. But you know, we asked a very similar question to Kristen that, you know, only about 40% of the Black voters in a poll we got out last week, 40% of Black voters are certain that if Democrats win Congress that their lives will get better, but almost 80% are certain that if Republicans win Congress, their lives will get worse. Right? So they're being mobilized by the other side. They're being mobilized to stop the pain, right? To stop the hurting. This is where this red undertow, I think is how you described it, that, you know, that red undertow could result in places like Texas or Georgia where Ted Cruz becomes a senator with less votes than Beto, or Keisha Lance Bottoms becomes governor with less votes than Stacey Abrams ever got, because of that red undertow that's dragging so many Republican voters back to the couch, right? Possibly disgusted.

Well, keep in mind Democrats are starting out from a higher floor this time of 215 seats. So a 20-seat gain would get Democrats to roughly the same place they ended up in 2006 or 2018. If it goes higher than that, then that's even more impressive because there are fewer competitive seats today than there were in either of those years. We've had midterms that have been about unpopular wars like '06, unpopular legislation like 2010 when the ACA cost Democrats 63 House seats, and unpopular prices and inflation when Democrats lost the House in '22. But at least in my lifetime, we've never had an election that's been about all three of those things at once. And I'd add in a fourth midterm as well. That's 1986, 40 years ago. Now, I don't remember it too well because I was two years old at the time, but back then, there was a 25% foreclosure rate on family farms. There was... And Democrats picked up eight Senate seats. There was a young guy elected to the Senate named Tom Daschle that year. And today, we have the worst anxiety in farm country since any point since 1986. And it's not just that diesel spike, it's also high interest rates, record input costs for fertilizer, and then a trade war and the White House importing beef, which has created this blowback amongst cattle ranchers in Iowa and Kansas and Texas. And it's not that those voters are going to defect en masse to Democrats because they're just culturally incompatible with today's Democratic Party, but to Terrence's point, it's back to the couch. And if those voters don't vote, their votes come directly out of the Republican column vis-a-vis 2024.

I think there is an enormous diminishing returns problem, and there are some of these races. So let me give you an...