What A Day - Should You Trust Midterm Polling?

Episode Date: September 24, 2026

With the midterms just around the corner, new polls are hitting the headlines every day. But why are there so many polls, and what can they actually tell us? On today's show, Jane talks with Galen Druke, co-host of Crooked's new podcast, Still Counting, about what he's seeing in the latest polling and what it could mean for November.Show Notes: Check out Still Counting – youtube.com/@StillCountingPodcast Call Congress – 202-224-3121 Subscribe to the What A Day Newsletter – https://tinyurl.com/y4y2e9jy What A Day – YouTube – https://www.youtube.com/@whatadaypodcast Follow us on Instagram – https://www.instagram.com/crookedmedia/ For a transcript of this episode, please email [e-mail]

There will be almost certainly error in the polls that we are seeing right now. That comes with the game. We should expect the average error in statewide races like Senate governor in a general election is four or five points. So we should expect the polls to be off. I can't tell you which direction they're going to be off in though. It's Thursday, September 24th. I'm Jane Koston. And this is what a day.

President Donald Trump Welcome to Chinese President Xi Jinping to the White House today The two leaders gave speeches during the arrival ceremony We can't be able to, We should cooperate with sincerity The interests of China and the United States

are deeply intertwined and there is plenty of room for us to work together China-U.S. cooperation may not solve every problem in the world. But without our cooperation, it would be hard to solve many of the world's problems. I love a sentence so vague that you could read it in any direction. Anyway, Xi's visit comes as a new poll from the Washington Post shows that Americans' views towards China have improved since 2021,

from 11% positive to 19% positive this year. Sure, 54% of Americans don't approve of China, but more due than before, and of course Trump loves China. But we don't live in China. Politico, MSN, and CNN returned to the White House grounds today, after a judge ruled last night that the Trump administration had to restore their access after Trump banned the outlets last week. Initially, the Secret Service still wouldn't let them in, but eventually the White House relented. And speaking of not controlling the media, it turns out that running a bad poll is not fraud.

A judge on Wednesday dismissed Trump's lawsuit, accusing the Des Moines Register and its former pollster of misrepresenting the state in the 2024 presidential race. The judge rejected the president's claim that the newspaper's poll coverage violated Iowa's consumer fraud law, because obviously it didn't. Calling the case extraordinary, the Iowa district judge said that allowing the lawsuit and its faulty claims to proceed could have a chilling effect that, quote, can discourage others from reporting on and debating public affairs, which sounds like what Trump had in mind the whole time. But let's talk some more about polls. Like, why are there so many of them? And what do they actually mean?

Joining me is Galen Droke, co-hosts of Crooked Media's newest podcast still counting to talk about what he's seeing in the polls for this fall's midterms. Galen, welcome to Wadaday. Thanks so much for having me. Galen, I remember back in, in, you know, thousands of years ago, say like 2014 or even 2018, I don't remember there being this many polls for midterm races. Even just yesterday, I have seen, would feel like dozens of polls,

of races where I can't even begin to wonder how you're actually polling races that are this localized. So am I wrong? Are there actually not more polls or are there more polls? And if so, why? It's so funny that you asked me that because I actually just got finished recording another podcast, which I also host called GD Politics, and we had a listener question that it got exactly at this point. I didn't know the answer, so we had to go look it up. And we are, in fact, seeing more polling than we have in previous midterm cycles, what have you. What's the reason why?

I think there's just a lot more people out there iterating. We used to live in a world where in order to conduct a poll, you had to pay a bunch of real people to get on the phone, open a phone book, and start calling real life voters. Now we live in a world where you can send people push polls. You can set, I mean, those are straight up fake polls where you like go on, I don't know, Fox News.com and they're like, who just won the debate? That's a fake poll. That's not sort of anything that we would take seriously. But there's a lot of email polls, text message polling. there are a lot of different ways to reach people to try to gauge public opinion.

To be clear, some of those methods are more accurate and more reliable than others. But the barrier to entry has lowered a lot. And so there are a lot more players in the game. And it's made the whole data ecosystem probably a lot more confusing for people. It has. It also, Galen, is confusing because we just from actually earlier this year, we had a host to primary races, where we had polls that said that, for instance, Abdul-El-Said was going to win his primary race by like 20 points, and that's not, he won by like a much smaller margin.

We had a host of polling misses in primaries, and I get primaries are weird. But we've had repeated either polling misses or confusion about polling in the most recent presidential election, in elections that have happened over the last couple of years. and yet we're getting more polls. Even as I think anecdotally, I hear from people saying like, eh, I don't really trust polls anymore. I mean, you saw what happened with Trump or something like that. So what's going on?

I respect that impulse a whole lot. And I would say, first of all, that horse race coverage, polling coverage is not the only thing you should pay attention to. I hope people are also paying attention to policy debates, conversations about AI. I, oil, the war, all the things. You should be paying attention to policy debates. But like, you know, there are lots of people who are like they want to get on the Newartimes.com and they want the needle.

Yes, 100%. And so to the extent to which that market takes up the news space, I think that we just have to do it responsibly because the desire to know the unknown is not going to go away. And the best tool we have for understanding it is the scientific method. Now, polling has become more complicated, but it's the closest thing we have to taking a sample of the public, engaging what they think about the candidates, about policy issues, what have you. And in a democracy, I would hope we all agree that what the public thinks matters. And trying to get an accurate sense of what they believe, whether they agree with you or not,

is an important endeavor to undertake. And so it is flawed. and we have gotten high-profile polling misses over the past decade, but I don't think that's reason to throw the baby out with the bathwater, to try to just make a little bit of sense of what we've seen. The polling error in 2016 was an average polling error. It was about three points off the national popular vote, where the polls suggested it was going to be.

It was a big surprise because people couldn't believe that Donald Trump would win, but three points is not that big of a polling error. In 2020, we saw a larger than average polling error, but Joe Biden still won because it was leading so much in the polls. I should add that in 2018 and 2022, the last midterm cycles that we have to go off of, the polls were very accurate. Like, the polls don't always fuck up to be clear. Am I allowed to swear? You are allowed to swear. I don't want to offend anybody.

But I think it's important to be careful about which polls you're looking at. More often than not, you want to look at the average. We can also delineate between high quality and low quality poles. I want to get into that because I think that, again, we're in a polling, boom. We're in a polling glut. So you're someone who looks at this. What are you looking for to choose which poles you look at? Because you could have polls that are put out where it's like, that's, you know, a number sounds objectively insane, but who knows?

So what are you looking at when you think, this pollster, I trust, this pollster, eh. And when you look at a poll, what are you looking for? Let's say we've got a trusted poll, which you will explain how you decide that. What are you looking for? Yeah. So first of all, pollsters putting out numbers that are outside the norm, outside of expectation, what we call an outlier in the industry, can be an indication of a high-quality pollster, right? They trust their methods.

They're going with their gut. They're not hurting just to put their poll in line with every. everybody else's poll. And we have seen in past cycles, for example, a high-quality pollster like the New York Times in 2022 publish outlier polls in House races where Democrats ended up doing quite well, despite the fact that, you know, the expectation is that we would have a sort of red wave in reaction to Joe Biden. Now, when you want, when you're looking for a high-quality pollster, you want a track record, you want them to be transparent about what they're doing, the methods by which they conduct their poll. And those methods are really important because it used to be the case that

high-quality polls were live caller polls. We're calling up people. We're asking them their opinion. And all of these other polls are text messaging, you know, email, whatever, these online panels, less reliable. Over the past decade, we have come to realize that some of those other techniques for polling have become more reliable. And today, sometimes, if you, only do live call or polling, you're going to miss a bunch of less engaged people who might actually respond to a text message or might open up an email. The New York Times has even tried mailing people, like in snail mail and giving them money if they respond. And what they found was a lot of these less engaged people who don't respond to the text messages and the calls have different

concerns than people who are like eager to talk the one, actually literally, the one percent of Americans who are willing to give a pollster their time. So now what we think of as high quality polling is mixed method. You want to try to reach people where they are, whichever way that you can. And it's an iterative process. We don't have this simple, straightforward open of a phone book and start calling anymore. And so you want, just to reiterate, the pollster to be transparent, have some kind of, ideally some kind of track record. They've done this before we know who they are. And they're using the most rigorous methods we know to reach people. So what kinds of polls do campaigns have access to about their own races?

And what can those polls tell the campaigns that we're not getting? I think we've seen a host of stories over the last couple of days in which a campaign will start, say, talking about how we have to end the war in Iran, a Republican campaign. We'll have to end the war in Iran and do something about oil prices. And you'll get a response online from political journalists basically saying, oh, their numbers must be terrible. must be worse than whatever they're telling everybody else. Are they getting different numbers than what we could see if we look this up on, say, real clear politics or something like that? Like, are they getting different data that's better?

What's going on there? What we're talking about here are what we call internal polls. And I want to be clear that you don't have to have, if you're a Republican running today in America, you don't have to have access to internal polls to feel like shit is bad, right? So this may just be a reaction to publicly available information. And we have seen, I mean, brutally bad polls, some of the polls for Republicans in places like Texas, Alaska, Kansas even, over the past week or so post-Labor Day, folks have been tuning in. We are in full-swing campaign mode. Folks have left Washington can now sort of take their month and a half to try to create some distance with Trump.

So that's where we are in the electoral cycle, the political cycle, which is why you're seeing some of this. But yes, campaigns do have access to internal polls. I would say the place where it's perhaps the most useful is for House candidates. Because not a lot of pollsters are going to, you know, a congressional district in Iowa or a congressional district in Texas. I mean, some are, but and getting granular information about a 700,000 person congressional district. Most pollsters, if they are doing localized polling, it's just at the statewide level. And so for House candidates, that can be really useful because there can be shifts in the electorate that are not nationalized, especially in a midterm election where candidate quality matters more, the facts on the ground in your state matter more. There's no presidential campaign at the top of the ticket to flood the zone and kind of, like during a presidential campaign, there's one contest that's being played out in 50 different states.

in a midterm, you're seeing hundreds of different content. The focus is on hundreds of different contests playing out across the country. And so there can be more variable factors. So like, for example, if you're seeing a massive swing towards Democrats in South Texas or South Florida, those candidates may be reacting to things they're seeing in the internal polling that we just don't have a lot of access to publicly. We'll get back to my conversation with Galen Drook in just a moment. What a Day is brought to you by Onskine. Founded by four PhD longevity scientists, one skin studied skin at the cellular level and found

that visible skin aging starts deeper, when certain cells stop working the way they should and begin affecting the healthy cells around them. So they engineered OS1, their proprietary peptide to work with skin's biology, targeting those dysfunctional cells instead of just sitting on the surface. It's a foundation of every product they make, and what you see over time with their products is firmer, smoother skin, more even tone and radiance, and visibly fewer fine lines. for skin that both looks and acts younger. I love that One Skin was made to work with your skin, not just on it. It's why I've gone back to the OS1 moisturizer over and over again.

And this isn't just my experience. One Skin's results are backed by four peer-reviewed clinical studies, over 10,000 five-star reviews, and they've been recognized by Bloomberg and Fast Company as a leader in skin longevity. Born from over 10 years of longevity research, One Skin is helping you unlock your healthiest skin now and as you age. For a limited time, try OneSkin with 15% off using code Wad at OneSkin.co slash Wad. That's 15% off at OneSkin.com with code Wad.

After you purchase, they'll ask you where you heard about them. Please support our show and tell them we sent you. This episode is sponsored by BetterHelp. I wish I were better at running, lifting weights, keeping plants alive. Okay, there's a lot I'd like to be better at. Because Better is a feeling defined by you. And for Better Help, it's a Better Help therapist who really gets you. It's a focus on global emotional well-being with thousands of therapists and millions of members in over 100 countries.

And it's getting better at stress management and handling what makes me anxious. Sign up and get 10% off at betterhelp.com slash wad. That's better, help.com slash wad. Let's get back to my conversation with Galen Drook. You mentioned a little bit about the past two midterms. Is there, based on what you're seeing right now, a comparison between, say, this year in 2018, or, as I've seen, are, is this looking closer to 2006, which was a massive blue wave election that essentially swept Democrats into power in Congress and essentially

ended the George W. Bush's administration's ability to do most things. I wish I knew. I left my crystal ball somewhere else. I am at home. I was going to say I left it at home, but, you know, that wouldn't be honest. It's in my closet. I can't find it. It's buried somewhere. It's tough. It's so hard. And here's a thing. There will be almost certainly error in the polls that we are seeing right now. That comes with the game. We should expect the average error in statewide races like Senate governor in a general election is four or five points. So we should expect the polls to be off. I can't tell you which direction they're going to be off in though. If the polls are underestimating Democrats, the polls that,

already look pretty good for them. They're underestimating Democrats. Yeah, this is 2006. I mean, 2006 might even be putting it lightly. I mean, we're seeing not just Democrats winning Kansas, we're seeing them compete in South Carolina, maybe Mississippi, and picking up Nebraska, right, in the Senate at least. Now, if the polls are overestimating Democrats by four or five points, the Senate is off the table. Republicans are keeping the Senate. And the House is probably still in play, but we might have to wait until, you know, California finishes counting all of its votes in order to be certain of who controls the house. And, you know, that's not even quite like 2018, that maybe even a little bit closer to like 2022, although, you know, the reverse in terms of

who was the out party. It's always a fool's errand to try to guess which direction the polls are off in. You know, it doesn't get as much attention, but the polls were pretty wrong in 2025. But, but they underestimated Democrats' performance. So Mikey Cheryl, Abigail Spanberger, governors of Virginia and New Jersey, the polls suggested that they were going to win. But the polls did not suggest that it was going to be a blowout. And ultimately, Republicans stayed home. There were flips around the margin, especially amongst Latino voters. And that did it. And so when you have, especially when you have differential turnout, this is what Republicans are really worried about right now, that their voters stay home.

Then you can see really big swings. This I would relate to 2010, right? When after Obama has this big victory in 2008, Democratic voters, his coalition stays home. And it's not necessarily that all those voters flipped. It's just that they weren't enthusiastic to show up the way they did in the presidential election before. If something like that happens this time around, you could see an underestimation of Democrats. Now, all this money that we're seeing spent, like, you know, the $100 million sort of bomb being dropped on Tala Rico, trying to brand him as Tala Frico, is trying to remind Republicans, hey, these guys are weird, like you better turn out or else they're going to be in control. And will that work?

Will that out sort of play gas prices, the Iran war, inflation across the board, worries about corruption, things like that? It's not going to outplay them. But will it inch Republicans closer? that's not an impossible scenario. So we have pollsters that seem generally pretty confident about Democrats at least taking the House and being competitive in the Senate. But you have expert ratings like Cook Political Report and they're being way more cautious. There are some battleground races where they're like, oh, it's very likely a Republican's

going to hold on to this. And you have pollsters saying, I don't know what you're talking about. Why are they so different? There are a bunch of different reasons. One is that, you know, it seems that the expert raiders sort of every election bring the sort of national environment back down to a baseline of zero, right? And from there, shift things in the direction towards, you know, Republicans or Democrats accordingly. We already have data suggesting, according to the polls, that we're not in a neutral environment. that Democrats could be leading in terms of Americans' preference for which party controls Congress

after the next election by like eight percentage points. Again, there could be error there. But even if there's error there, that's not a neutral environment. And so they take generally this neutral environment. And then as they see the race evolve, you know, the primary process play out, the candidates get picked. You know, you see maybe some more polling in the actual district. Then they start pushing things in the direction of, in this case, Democrats. It wouldn't surprise me if by election day, the expert raiders lined up more closely with the pollsters and the modelers, they will have just taken a lot longer to get there because in their conception of the world, every election begins at a baseline of the national environment being zero. I know this is a little bit wonky.

That's what we're here for. And so, you know, that's one thing. The other thing is that data can do weird things. I mean, for example, if you look at the Alaska governor's race, there's three Republicans running, one Democrat running. There's a lot of consternation about, you know, that two of the Republicans should have dropped out. So it's just a Democrat versus a Republican. There's ranked choice voting. It's, you know, it's not really clear how it's hard to model that, I will say. And there's not a lot of polling either. So the polling-based forecasts can get weird results. The expert readers go in and they say, I know this candidate. I know their campaign. I've talked to them. I know their opponent. I've talked to their campaign too. I know the weird things

about Alaska. I know the issues that are at play. I know that this is a stronger candidate or this is a weaker candidate. And regional factors and candidate matter more in a midterm environment than a presidential environment. And therefore, my intuition, again, these are mostly sort of like intuitions, is that this candidate is going to outperform or underperform, what have you. That's, that's valuable, especially at a very sort of district or race-specific level where the modelers just don't have as much information. There aren't polls in all of these congressional districts. And so I think that's probably where some of the disagreement comes from. We also have prediction markets. And prediction markets, yeah. So we have polls, which in my

understanding are basically snapshots in time. How do this group of people who were polled this three-day period by this entity feel. The prediction market is like, aha, based on this, we will say that there is a 62% chance that this person will win, and you can vet the over or the under on that. How are prediction markets skewing, or maybe not skewing, are understanding of what the midterms are going to look like? Well, I think the good news is, and I also believe this about the polls more broadly and forecast models more broadly, that not enough Americans are sickos like us for all of these things to really shape people's understanding of the race all that much. You know, most people, a lot of people are like, when is the midterm?

Yeah. You know, like who exactly is my congressional candidate? These are the people what time is the Super Bowl. Yeah, exactly. And we, on this call, might even be some of them. Especially when it pertains to the Super Bowl. But, you know, so I think are the prediction markets making people, feel, like are people refreshing the prediction markets and thinking like, oh my goodness, should I

change the way that I behave? Should I vote or not based on what the prediction markets say? Should I switch which candidate I might be supportive? I don't think that is a big factor. I think the way that it might be a factor is that like it filters out through the media ecosystem. So like journalists pay attention to them. And then that might shade how they cover an election or a race or whatever. But we saw, for example, there was a lot of confidence that, you know, Abdul al-Said was going to win in Michigan and then things started flipping back and forth on election night when it seemed like

actually Haley Stevens was doing better and clearly the prediction markets were way more confident in Abdul al-Said's victory than they should have been given how close that race was. Similar thing happened with Francesca Hong in Wisconsin. And so, but people still voted however they were going to vote. It wasn't like people looked at the prediction market and said, it's not worth voting. Francesca Hong's definitely going to win or what have you.

So I don't think it plays out in quite that way. I think, though, that you now see Polymarket, Kalshi, all these different places, embedded in like the Wall Street Journal, the different news sites. So people are paying attention to them. I don't think it means a lot to a lot of people, right? I think we are in this era where people kind of feel like maybe you're getting at this. Data is bullshit. Like what do we even know?

Yeah. What's to be believed? What's to be believed? And yeah, I don't think. I don't think your rank and file Americans take it all that seriously. Wow, that's something. Galen, thank you so much for joining me. Thank you for having me.

That was my conversation with Galen Droke, co-host of Still Counting. Before we go, if you're starting to feel anxious or confused about the midterms, fear or not. Still counting has the data and info you need to understand what's happening right now. Every week, data journalist Nate Silver, Galen Drewk and Claire Malone, break down polling data and public opinion. They're not focused on how things should be. They're giving us the reality checks we need to move forward with clear heads.

On yesterday's episode, they're talking about all of the most contentious midterm races. Tune in to Still Counting wherever you get your podcasts or watch on YouTube. That's all for today. If you like the show, make sure to subscribe, leave a review. Let's listen to more wisdom from First Lady Melania Trump and tell your friends to listen. And if you're into reading, and not just about Melania Trump speaking to Fox business yesterday. What are pieces of advice for moms who feel like they're so stretched thin, trying to do it all? But at the end of the day, just want maybe a dinner at night with their family.

It's good to schedule, right? It's always good because everybody is so busy. You know, my husband is working. Sometimes he's in the overall office still 9.30 at night, right? So do you wait that you will have dinner? You don't because he's busy. I'm busy too, yes. But we try to find time.

It's less and less, you know, the family time. It's suffering because of it. But that's okay. We are here to work, to do a good job, and that's what matters. Like me, what a day is also a nightly newsletter. Check it out and subscribe at crooked.com slash subscribe. I'm Jane Koston. And first, is she an inspiration to moms?

Which moms? An inspiration to do what? And secondly, if doing a good job matters, I have really bad news for you, Melania Trump. Why Today is a production of Crooked Media. Our show is produced by Emily 4, Erica Morrison, and Adrian Hill. Our team includes Haley Jones, Greg Walters, Metberg, Adam Lippurg, Johanna Case, and Desmond Taylor. Our music is by Kyle Murdoch and Jordan Cantor. We had helped today from the Associated Press.

Our production staff is proudly unionized for the Writers Guild of America East. I'm Nate Silver. I'm Gaylon Druk. And I'm Claire Malone. We're back. This time with our friends at Crooked Media for a new show called Still Counting. Yeah. Numbers are hard.

It takes her easy. And frankly, numbers have only gotten harder. What does the data actually say? How do we reconcile that with the other things that we're seeing in American politics and around the world? And how do we use that data to understand or maybe even dare to predict the future of American politics? New episodes every Wednesday starting September 9th. Subscribe to Still Counting on YouTube. slash at Still Counting Podcast or wherever you get your podcasts.