Modern scientific polling began with George Gallup and the 1936 U.S. presidential election. Ninety years later, the polling “industry” has been, at different times and by different people, cheered, vilified, politicized, and challenged. In recent years, political polls have been under increased scrutiny as political parties have used results to manipulate voter turnout or influence outcomes. The recent Wedgewood Poll scandal is the latest example. But one thing is certain: polls aren’t going away. So, can we still trust them? The Daily Wire’s Ben Domenech discusses with Brent Buchanan, the founder and CEO of Cygnal, a Washington, D.C.- based international polling, public opinion, and predictive analytics firm, the consequences of Wedgewood, what makes for a good (or bad) poll, and the one thing voters can trust, in this weekend’s Punch interview.
***
Ben Domenech: Let’s start with the fact that you have a story breaking in the late stages of a very critical and extremely expensive midterm election surrounding yet another one of these fake pollsters that is conducting so-called research. Their activity has cut to the core of a lot of people’s frustration and paranoia about the polls they read. And I would like you to explain what’s going on there and what the attitude should be toward entities like this that claim to be performing an experiment. But to me, in this case in particular, they seemed to come forward with that language only after they were discovered.
Brent Buchanan: It’s an embarrassment to the media — that’s the real story on this fake polling because this is not the first time it’s happened this cycle. This is just the most expansive. But the media’s job is to gather the information and vet the sources and present the story. Polling should not be exempt from that. If anything, it should be more scrutinized than even a source in a media interview. And when I saw the Wedgewood post two days ago, where they apologized and explained it away before shutting everything down, it brought me back to incredible anger at the laziness and incredibly obvious progressive bent of our media. It’s like, “Look! these polls look horrible for Republicans. Publish!” Never pulling the thing up and saying, “I would like to see the cross tabs. I would like to look at the demographic composition of this poll. Does it properly account for what the last election looked for and what we call a retro vote — who did you vote for in the last presidential election? Does the partisan composition make sense? Do the age breaks make sense?” There are just some very basic things that the media should do in determining: is this poll worth publishing — should I publish this?
Now, that being said, there are plenty of crappy pollsters out there. It’s never been cheaper or easier to do crap polling — not making it up, just doing it really badly and putting it out there. But all of that should be scrutinized.
BD: I want to ask you a couple of things related to this. First, as a member of the media, what astounds me when it comes to offering commentary where people will cite polls frequently, because they’re using it, by the way, as an appeal to authority, which I’m extremely skeptical of for the most part. They oftentimes will also betray a total lack of interest in the quality or the background record of the pollster in question, meaning that they will cite anything that goes to their argument regardless of the history of the polling company or their partisan bent or how they’ve been shown to be off in significant ways over the course of years. And it really frustrates me because when I cite a poll on air, on Fox or for the Daily Wire, I’m always someone who wants to look at the actual data.
I want to look at the cross tabs. I want to make sure — now I guess that it’s a real poll — but in the old days it was just, “Is this a good poll? Is this a poll that I can cite with any confidence?” And especially when it’s a pollster who I’m less familiar with, where we can all cite the misses that have happened, or big misses over the years with various places that have a lot of name I.D. And so that level of being able to check your citations, that’s not even happening. It’s not happening at all. And I can tell you that from experience in green rooms going back years, people do not check the sources on this. What has to happen to get them to do that? Even that most basic thing of saying, “What is this pollster rated? Have they had a good trend line? Can I cite them with authority?” It seems like people cite these things with less effort than they put into picking a mechanic.
BB: Because it fits their ideology. So why would you find out that the poll sucks or is fake and then be real sad that you don’t get to publish it to make Republicans look bad?
BD: Is it just that? You just assign it as fitting the narrative, and the established narrative of the legacy media is that it makes Republicans look bad. And that means that people are willing to run with idiocy like Wedgewood saying that Mississippi is a four-point game?
BB: It’s that or complete laziness and incompetence. So those are your two choices in the media.
BD: So, a little bit of column A, a little bit of column B.
When it comes to fighting back against this, what advice would you give? Particularly, I’m thinking of people who comment on polls just to their friends, people who are intelligent consumers of political media; how should they decide which polls to trust and which to throw in the scrap heap?
BB: Go to the bottom of the survey first. One, read it yourself. Don’t just look at the media article or citation on X or on a television newscast. Go look at the survey yourself. If you can’t find it, don’t just completely throw it in the bin. But if you do find it, scroll to the bottom and look at the composition of the survey. What percent Democrat, what percent Republican, what percent Independent? Does that track with how that state somewhat looked? Now you’re not going to have it identical, because electorates do change a little bit between elections. Did they ask a question of who you voted for in the last presidential election and does that look somewhat like the actual results within that district or state? If you have 27% of the sample is under the age of 30, there’s nowhere in America where 27% of the turnout is going to be under the age of 30.
So there are a lot of just very basic things you can look at on the survey before you look at the actual ballot results.
BD: One of the things that I think is critical for people to understand when they’re reading polls is some of the kind of baked-in assumptions that pollsters have based on the historical experience when it comes to evaluating heavily red or heavily blue states in terms of either undercounting or overcounting segments of the electorate. What should people know about that?
BB: There is a science of statistics underpinning polling, but there’s an art to it also, and you’re making projections of what the electorate is going to look like. And that’s why looking at those demographics that I just mentioned allows you to see into their mind of how they think that electorate is going to look.
BD: I think that is something often left to the side when people are looking at these things. But when it comes to the way that people are feeling right now about the polls that they’re seeing out of a heavily red state, I’ll just pick Texas in terms of its performance as an example. What assumptions should they have about the polling that they’re seeing in the Senate and the gubernatorial race there based on past experience in Texas?
BB: You have to look at what’s happening in the country as a whole also. And if you aggregate all the trustworthy polls, people who have a pretty good record, you can’t expect any pollster to be right 100% of the time. If they’re right 90% of the time, they’re probably a pretty good pollster. And you’ve got to take into account margin of error. Polls are not precise. The science of statistics is not precise behind it. And that’s why there’s a margin of error. That’s why there’s a confidence interval. And so if you look at the country as a whole, it has shifted Left drastically from 2024. And if you apply that to a place like Texas, you can say, “Okay, this is not the Texas of 2024 or 2022.” So I think that’s the most important thing if you’re sitting in Texas saying, “Why is Paxton not winning by five right now in a state Trump won by 14?” It’s because that state that Trump won by 14 is not who’s voting this go-round. And if they are, there’s a number of people who have switched. And I think this is a completely misread part by political analysts, but also by regular folks, is that the Trump coalition does not look like all the people who show up to his rallies.
A lot of people who voted for Donald Trump really didn’t like him, but they liked what the economy was pre-COVID the last time he ran. And there may be things they don’t like about the guy, but they were willing to put up with that in order to get the things he promised. No wars, prices will come down — or at least stop rising. And they haven’t gotten any of those things. And so they’re punishing Republicans. And then there are Republicans who are somewhat Republican just because they’re not Democrats, who are sitting home.
BD: How difficult is it to poll now in certain areas of the country based on the concentration of people? Is it more difficult to get an accurate number from rural areas versus suburban versus urban? What are you seeing in that regard?
BB: It’s still an educational attainment challenge, meaning that the higher up you are on educational attainment, the more likely you are to respond to a survey, but also the more likely you are to vote. I’ve seen some Republicans put out there — like people who vote in four, or four of the last elections, the last general elections, drastically over-respond if the Democrats do, where the Republicans are under-responding. And we see some of that. Our firm is going to do 100,000 interviews this month, just this month. You think we would see something if it was that big in the fact that we’re probably doing more interviews than any other polling firm in the country this month. So I don’t know if they’re not good at polling or what, but people who vote consistently and people who are higher educated always have responded to polls more.
It’s harder to get folks who are lower down the educational attainment scale, didn’t go to college and finish college, and maybe didn’t even finish high school. But it’s also harder to get younger voters. It’s harder to get minorities. And so in my mind, it balances itself out because non-college voters are the base of the Republican Party, so they’re not responding at the same rates as PhDs are. It gets balanced out by the fact that younger voters and minorities lean more heavily Democratic.
I don’t think anybody’s making that argument right now. The reality is that there are things in polling that are also, if corrected, would be good for Democrats.
BD: When it comes to the way that these aggregators have changed the way that the public views polling, what’s been good about it and what’s been bad from your perspective?
BB: It helps add some smoothness to what you’re seeing. So that way, if a poll comes out that says your guy’s down by 11, but all the other recent polls in the last few weeks show he’s only down by one, that’s probably an outlier. So it helps you more easily see those things. I also believe that when put together, as long as the inputs are good, if you’re throwing Wedgewood polls in there, that’s going to screw with your average. But if you’re looking at RealClearPolitics, which I think does a pretty good job of aggregating decent to good polling numbers, it helps you see trend lines. And that is the real value in polling: it’s not a single survey. It’s multiple surveys over time that show where things are headed based on where they’ve come from.
BD: One of the things that we’ve learned in the past couple of years is that now the public pays a lot more attention to these things that used to be paid attention to just by political insiders and activists. Talk to me about what the real fix could be for public polling, because it seems to me that some of these attempts — either the grades that were offered to pollsters or the attempts to track them in terms of their performance that have been out there — it seems to me that those haven’t really succeeded. Is there something that would help when it comes to this proliferation of public polling that people really rely on and that we have a lot of evidence changes behavior in terms of the way that decisions are made and the way that people think about races?
BB: Back in the day, newspapers did their own polling. They paid for their own polling. TV stations paid for their own polling. The demise of the news media writ large is, I think, a contributing factor to where we find ourselves today. And it goes with what I mentioned earlier: it’s never been easier or cheaper to do crap polls. Especially with AI, you can figure out how to run a survey; that doesn’t mean you did a good job. It’s really challenging to do a survey properly these days. And that’s why we have over 40 people at our firm, because we don’t outsource the functions that we believe are most important to get a good sample. So if we had media outlets, and I know we do some with The New York Times and the Post and Fox News and others, but there’s nobody locally; none of the local TV stations or local newspapers have the resources anymore to do that.
And I think that’s important because it puts their name on the line. If it’s wrong, it’s their reputation. And so they would have a strong motivation to ensure that the polling that they’re putting out is well conducted, believable, bulletproof, and holds up under scrutiny, because their name is on the survey.
BD: One of the things that I’ve heard a couple of people say this cycle in the analyst and pundit community is that Republicans dumped an enormous amount of money into the advertising side of things over the past several weeks, a complete attempt to change things, change the direction, salvage some stuff from this election, and that it didn’t move anything, that nothing really responded to it. Is that what you are seeing in terms of a broad truth, or are there signs that that’s not the case?
BB: It ignores the fact that what would’ve happened if they hadn’t spent that much money? Where would the races be without the additional hundreds of millions of dollars? The other is when you’re the party in power, and people are pissed off — they’re not looking to hear more from you. They’ve already seen and experienced the last two years and made a judgment on their own for that. So when you’re coming and saying, “We’re going to do this, we’re going to do that,” they’re looking at you going, “Well, why haven’t you already done this and that?” And if they’re mad enough, they’ll just look at the other side and go, “I don’t really trust them, but it can’t be that bad. It’s only two years.” And if that’s the place people are, how do you credibly advertise?
Unless you take the Domino’s pizza strategy, which is, “Hey guys, we know our pizza tastes like the cardboard that it’s delivered in. You’d be better off eating the cardboard. We screwed up. Here’s what we’re doing to change it.” Unless you come out and you’re that contrite and authentic and honest with folks, they’re not in a place where they want to listen to you. So the Democrats have it so much easier to drive a message because they’re not the ones in charge of the mess that voters believe the country’s in. It’s the same thing. We probably, as Republicans, had a lot of leeway to do and say things in 2024 because we weren’t the party in power.
BD: You see these statements about the difficulty of polling in midterm elections, and I wonder what your perspective is on that, having gone through multiple cycles. What makes it difficult to poll, particularly given the dynamics of the new coalition that has proven to be the basis of the Republican electorate in the past several presidential elections?
BB: You have higher turnout in presidential elections, so you have a better idea of what the electorate’s going to look like. In a midterm, the drop-off in who votes can be different because we’re in a historically high-turnout period in our country. If you look back over multiple generations, you’d have to go back into the 1800s to really find a period that’s at the high turnout level that we’re seeing right now. And that just complicates the ability to project what a likely electorate looks like.
BD: I’m curious as to how you think people should be paying attention to these polls that are released in the final week before the election. There are always numbers that come in late that tend to surprise people one way or the other. Some of them turn out to be true, some of them turn out to be complete crap. What should people take as a grain of salt when they see a poll being boosted in the coming week?
BB: Is there an agenda behind this and does it match the trajectory of the rest of the polls that are coming out? That’s the big question to ask. The other would be, does this match what I think the turnout’s going to look like? Does this trend with what other people are seeing and what the turnout is going to appear? It’s kind of like early vote numbers. You can only draw so much from an early vote statistic because the electorate changes. It’s one of the reasons that we run mortality calculations when we build our likely electorate, because there are a noticeable number of people who voted in 2024 that are not alive for this election, and they may have been perfect voters, and you’ve got to take account for those things. But let’s close with this. If you feel strongly about a candidate, go vote for them.
Who cares what the polls show?
