Polls Show a Blue Wave Coming. Why Are Democrats Still Nervous?

Airdate: Monday, November 2, 2026 at 9 am
Democrats are poised to take the House of Representatives and maybe even the Senate on Tuesday – at least that’s what recent polls tell us. But as we learned in the 2016 election, polls can be wrong. We’ll talk to pollsters about how they gather their data, how their methods have changed over the years, and the key races they’re looking at this midterm election.
Guests:
- Michael Dimock, president, Pew Research Center
- Charles Franklin, director of the Marquette Law School Poll, Marquette Law School
- Mike Madrid, political strategist; author, “The Latino Century: How America’s Largest Minority is Transforming Democracy”; co-founder, The Lincoln Project
Episode Transcript
This is a computer-generated transcript. While our team has reviewed it, there may be errors.
Alexis Madrigal: Welcome to Forum. I’m Alexis Madrigal. Hey, it’s almost over, election season. I mean, the constant texts, the ads, the anxieties over what the results might say about the direction of the country. Tomorrow, the votes will be tallied, and we’ll know some things that we are only guessing at today.
But sometimes it’s important to preregister what the prevailing sentiment about an election is before the actual voting. We’re all interested in having the best possible information about how people in this country are feeling, what they care about, and how they might vote. And pollsters are charged with figuring out how to provide that information in a wildly dynamic technological, social, and political world.
So here we are. We’ve got a panel of experts on polling and the electorate, and we’re hoping to help you understand what the polling says, what it does not say, and what it cannot say.
We’re joined by Charles Franklin, director of the Marquette Law School Poll in Wisconsin. Welcome, Charles.
Charles Franklin: Good morning. Good to be here.
Alexis Madrigal: We’re joined by Michael Dimock, who is president of the Pew Research Center. Welcome.
Michael Dimock: Thank you for having me.
Alexis Madrigal: And we’ve got Mike Madrid, political strategist and author of The Latino Century: How America’s Largest Minority Is Transforming Democracy, also co-founder of The Lincoln Project. Welcome.
Mike Madrid: Thanks so much for having me. Looking forward to this.
Alexis Madrigal: Yeah. So, Michael, why don’t we start with you? Set us up, broadly speaking. People have probably heard so far what it looks like in this midterm election, but how does the landscape look to you?
Mike Madrid: Well, I think the fundamentals of this race put the incumbent party in a rather precarious position. It’s historically accurate to say that the party that is in power does very poorly during midterms, but this is exceptionally bad for the Republicans at this moment in time. I would say generationally bad.
Everything from economic indicators to the generic ballot to early vote returns, two years of polling — the behaviors of both sides, both parties — are all indicating that something very significant is happening that is likely to see a very big Democratic year.
But, of course, we won’t know that until ballots are actually counted. The framework, the mood of the electorate, it’s extremely sour. It’s likely there’s going to be significant Republican losses, but we’ll have to wait and see.
Alexis Madrigal: Yeah. So given that is sort of the backdrop, you’re there in Wisconsin in the Upper Midwest. What do you think about — there’s big breaking news over the weekend out of Michigan and the Senate race. Tell us what happened and how much you think it could actually affect things in that race, given the kind of setup that you’ve described.
Charles Franklin: Sorry, are you talking to me, Charles?
Alexis Madrigal: Charles, yes, Charles.
Charles Franklin: Yeah, sorry. And I am in Wisconsin. We do touch and border with Michigan, but I’m not an expert there.
Look, the Democratic candidate who lost the primary endorsed the Republican Senate candidate three days before the election. That was already strained relationships, but has now become especially strained. And the thought among Democrats is that this raises the question about whether Democratic unity will break down in the voting tomorrow.
It’s late in the campaign, and voters have had plenty of time to make up their minds. Every late development makes all candidates and all analysts go, “Oh, look at that. That’s happening at the last minute.” So we’ll see, but it is pretty late, and people have had a lot of time to think about the race.
Alexis Madrigal: Yeah. You know, Mike Madrid, there have been lots of changes. There was a whole gerrymandering fight across the South, and California responds, and there’s just been changes to the map. So even given what this polling looks like, how are you looking at this race, given that Republicans and Democrats have both been trying to kind of reshape the electorate that’s voting in these various House races?
Mike Madrid: Well, the truth is we’ve been reshaping the electorate for the better part of a couple of decades now. So even though we do see very significant anti-incumbent and change directions from the polling and all of the signals politically, the limit of what can actually happen is pretty significant as well.
And what I mean by that is a 15- to 17-seat majority would be pretty extraordinary for the Democrats if they won that many seats. Compare that to a big, massive change year like 1994, when there were 80 seats changing hands. Those days seem to be gone.
So we are, in many ways, getting better — demographically better, as a relative term — at kind of carving up our country through representation means, locking them down from a partisan perspective.
But having said that, I also do believe we’re going through extraordinary transformation, redefining what being a minority actually means with the rise of the Latino vote. Why are young voters — why were they so Republican just two years ago and moving back now? What are non-college-educated voters doing?
A lot of what we have historically understood voters and vote propensity to be is being redefined. They’re simply changing. And both parties, I think, are struggling to figure out what exactly that means and how durable their coalitions actually are.
Alexis Madrigal: Tell us a little bit more about what you mean by voting propensities changing among these particular groups that you mentioned.
Mike Madrid: You’re seeing, for example, something that I’ve never seen in 35 years, which is propensity — the likelihood to vote. You’re seeing Latinos right now outperforming in early votes over Black and white voters. I’ve never seen that before because it’s never happened before.
And even more so, it’s not just with Hispanic voters. It’s with young Hispanic voters under 30 with little or no propensity, meaning a lot of these voters have never voted before or voted only sparingly.
So all of those are very significant indicators of people who are not civically engaged. They receive no investment from either party. And frankly, they’re very rarely polled because why would you poll them as a pollster? They’re not in any sort of likely voter universe model.
And as a result, I think that propensity, that likelihood to show up, they’re demonstrating that they are going to have significant impacts on what the races actually look like when the votes are counted, compared to how we were looking at them a few days before in polling.
Alexis Madrigal: Michael Dimock, what Mike Madrid is saying is sort of like what people said after the 2016 election. There was a pool of voters out there. In this case, they turned out to be Trump voters and the beginnings of the sort of MAGA coalition who weren’t being polled or weren’t being seen.
When there is a possibility of this happening, how do you try and get ahead of that, especially since 2016?
Michael Dimock: Yeah, it’s a huge challenge for any pollster. I mean, you want to make sure you’ve included everybody in the sample you’ve drawn, everybody who’s a potential voter. Some of those folks are decidedly harder to reach, and the amount of effort you have to put in to reach those harder-to-reach people costs you at the end of the day. You have to work really at it.
So if you have a limited budget, you’re trying to move really quickly, as a lot of election polls are, they often will, to Mike’s point, not worry so much about reaching those low-propensity groups. And then when those groups really are fired up, when they are kind of animated about the election, you can miss that energy and miss a lot of those voices.
You saw that in 2016. There were a lot of ways in which pre-election polls clearly were not reaching everybody. And even, we were underestimating that animation, that excitement from some of the Trump supporters with the measures that they were using.
But you’ve also seen it go the other way, where it’s actually the support among Democrats that’s underrepresented and under-supported. So we’ve had some cycles where the bias or the miss has been missing Trump voters, but that’s not always been the case, even in recent elections.
Alexis Madrigal: Hmm. Charles Franklin, what about polling midterms versus the presidential years? How different is that, and how different is this underlying methodology that goes into these kinds of midterm polls?
Charles Franklin: At least my methodology is largely the same, but if you look at the last three presidential races, we’ve tended to see larger misses in them than we’ve seen in the last two midterms during the Trump era.
I think a lot of us believe that that’s significantly because President Trump mobilizes a set of low-propensity voters, as Mike was saying earlier. People who are hard to contact. They’re quite supportive of President Trump, but they’re also very negative toward government and media and, sadly, to pollsters, and so they’re hard to get in the samples.
But those folks also drop out in the midterm electorate. And so, if we can’t poll them as well in the midterm electorate, it matters less because they’re less likely to vote. We’ll test that tomorrow and see if we’ve missed.
And I’d also point out that what Michael was just saying is that we’ve had misses that are too pro-Democratic, but maybe we’re forgetting that we have had misses that are too Republican. So it’s hard to predict ahead of time which direction a miss will be.
We do all we can to reach those hard-to-reach voters and to try to figure out who’s actually going to vote, but if we knew better how to do it, we wouldn’t miss as often.
Alexis Madrigal: I mean, Michael, I wonder, how do you try and find — when you talk about that voter enthusiasm, there are sort of measurables there, right? So do you end up shifting the way that you’re weighting your polling based on that, or do you, once you establish a methodology at the beginning of the election season, try to carry that through?
Michael Dimock: Yeah, I mean, ideally, you’ve started the season with a model and a plan, and you stick to it. But, you know, conditions change. This has been a particularly unusual year. You’ve had all the redistricting that Mike mentioned earlier. You’ve had sort of ebbs and flows in the national versus localization of these elections.
I would say that, you’re right, you use the term weighting, and we don’t want to get too technical here. But at the end of the day, there’s a universe of Americans, and then there’s the universe of Americans who are actually going to show up and cast a ballot, whether by mail, before the election, or on Election Day.
So that even itself creates a complexity, right? Mechanisms by which people can vote. And there are really two big ways to try to guess at who those likely voters are.
One is their past history of voting. So in a perfect world, you might be doing your interviews with the knowledge of their past electoral behavior, because maybe you’ve looked — most states make available their voter files. So you can actually see that record.
The second is to talk to them and ask how enthusiastic they are. But they’re kind of self-reporting, and the passion they express may not be the same.
Alexis Madrigal: We’re talking about the latest polling ahead of tomorrow’s midterm election. We’ve got Michael Dimock, president of the Pew Research Center. We’ve got Charles Franklin, director of the Marquette Law School Poll in Wisconsin. And we’ve got Mike Madrid, a political strategist and author of The Latino Century: How America’s Largest Minority Is Transforming Democracy.
We want to hear from you. Simple question: Do you trust the polls? Why or why not? 866-733-6786, [email protected].
I’m Alexis Madrigal. Stay tuned.
Welcome back to Forum. I’m Alexis Madrigal. We’re talking about the latest polling ahead of tomorrow’s midterm election. Pretty much every poll shows that Democrats are doing about as well as they possibly could have imagined, and yet among a lot of Democratic voters, there’s a lot of anxiety about whether or not that polling is going to be reflected in the actual vote counts tomorrow.
So we’re talking polling this morning. We’ve got Mike Madrid, political strategist. We’ve got Charles Franklin, director of the Marquette Law School Poll in Wisconsin. We’ve got Michael Dimock, president of the Pew Research Center.
Michael, I just wanted to kind of finish the thought on the difficulty of trying to get accurate polling. You said there’s the universe of Americans, there’s a universe of people who are going to vote, actually, and then there’s how they’re going to do it. But then there is also the universe of people who you’re able to get to answer polling questions, right? And so it’s kind of trying to map that last category onto the other ones, right? That’s kind of where a lot of the difficulty lies these days.
Michael Dimock: Yeah, no, it’s really true. I mean, I think there are different approaches to trying to solve that problem.
Some polls really establish long-range recruitment plans into their surveys. Pew Research Center is probably on this end of the spectrum. That makes our surveys pretty solid when it comes to talking about all Americans, but actually, our polling methodology, as rich and expensive, honestly, as it is, makes it not well suited to pre-election polling because we don’t have the ability to talk quickly with people, right, in the weeks before the election. Our surveys are slower-moving.
And we don’t have that immediacy of the voter file records to sort of map in and really understand who those most likely voters are going to be.
So I guess the point I would make is even some of your quote-unquote blue-chip polls aren’t necessarily well suited to pre-election polling.
And then there’s this other pool of more opt-in polls, where people just sort of join the polls through their own volition. And there you’ve got an enormous range of methodologies that makes it very hard to know what their past behavior is. And you’re really just trusting what they tell you in the survey as an indicator of their likelihood of voting.
In some years, that emotional charge is a great predictor. In other years, it has been a weak predictor. And that’s where we’ve seen some of this variation across polls, because they’re using so many different methods these days to reach people.
Alexis Madrigal: Charles, let’s bring in Sandy in Santa Rosa, and we can send her over to you. Welcome, Sandy.
Sandy: Yeah, hi. I’d like to know how the people are selected to be in the polls. How are they contacted? If it’s by phone, how do they get the phone number? What’s their perspective on the demographics of the people that are polled?
Alexis Madrigal: Yeah, that’s a good one. Charles, do you want to talk about it?
Charles Franklin: Yeah, that’s a great question. I’ll tell you two methods. We use both of these, but they are pretty common in the polling world now.
For our state of Wisconsin polls, we sample from the registered voter list. So that’s the complete list of everyone who’s registered to vote. Those are public documents, and they also include whether you voted in past elections, but not, of course, who you voted for.
So we sample from that, and then we send an email and a text message to the sample, asking them to click on a link to do the interview. If we don’t hear back, we send them follow-ups, and if after those four attempts we haven’t reached them, then we still phone them.
As a result, we end up with about 5% to 10% who are interviewed by phone, though most people respond online.
But the key point is everyone was sampled from the voter file and is given an equal chance to be selected into the poll, and you can’t just volunteer to do the poll.
For our national surveys, we sample from a panel which is selected through postal addresses around the country. Again, random sampling gives every address an equal, or at least a known, probability of being selected. And then those folks are asked to take part in multiple surveys over the coming years.
The company that we hire does that for us, and those folks know that they’re going to be sampled, but they were selected randomly to begin with. And again, the key point is the random selection, not volunteers.
Alexis Madrigal: Yeah. So this is a question that I have purely for myself. I do not know the answer to it, Charles Franklin.
So as I understand it, right, you’re doing this random sampling. And then, you know, I was reading The New York Times. Nate Cohn was kind of breaking down how they do their polling, and there’s this really interesting table where he says, like, okay, because we have a more difficult time reaching Republicans, we essentially need to — every Republican we’re able to reach, even though we tried to do this random sample, every Republican we’re able to reach, we’re actually going to put a little more weight on what they told us.
And he actually says, you know, if we used our older weighting, Michigan would be D-plus-16, Alaska D-plus-6, Iowa D-plus-4. Instead, they published D-plus-11, D-plus-1, D-plus-1. So it’s sort of just taking a few points off the top for Democrats.
This is such an interesting thing because how that number feels not arbitrary, but it does feel like it’s something that you have to select. It’s a subjective judgment.
Charles Franklin: Yeah, and it’s complicated. Nate is doing a great job of opening the doors and showing you what goes into these things, but it may also be a little confusing to folks.
One of the things we do in Wisconsin is divide the state into over a hundred regions that go from the most pro-Trump regions to the most anti-Trump regions. And then we try very hard to get the right proportion of interviews from each of those regions. That’s a way of trying to make sure that we’re not underrepresenting any group, including those that may be less mobilized.
But in the end, of course, we can’t make anybody take the survey. So a lot of this does come down to the motivation of people to participate in the survey.
And as for giving more weight to Republicans, it’s common, universal to give more weight to groups that are harder to contact. So, for example, in Wisconsin, about 15% of voters are under age 30, but in any given sample, that’s more likely to be only about 10% of the sample. And so we give more weight to the 10% of younger voters that we can contact so that in the weighted sample, the whole sample, they make up the 15% that we know from voter files they really do represent.
And that’s done across a variety of demographics, precisely because different groups do respond at different rates. And that includes partisanship, but includes a host of other demographic and geographic variables as well.
Alexis Madrigal: Yeah. Thank you for that explanation. Yeah, I didn’t want anyone to think it was something nefarious. It’s an attempt to try and capture a realistic example of the voting universe. So thank you for that.
Mike Madrid, one of the things in your work that you’ve drawn attention to is that Latinos are acting, at least in this last couple of election cycles, in quite unusual ways. Like, they’re actually acting like the swing voter of myths. Someone who’s actually changing parties or seemingly going to do so.
Do you want to talk to us a little bit about what you’re seeing there? And, of course, not just maybe in the traditional Latino places we think of, like California, but across the country?
Mike Madrid: Yeah. And I love the way you set this up because I want people to really reflect on this.
Latino voters are rejecting the politicians of both parties that they believe have failed them. And the way the question was characterized, understandably, was that this is sort of a mythological, peculiar voter.
This is actually the way that the system was designed to work. It’s not that Latinos are voting peculiarly; it’s that everybody else is.
The fact that we have become such a hyperpartisan nation and we have now accepted that as normal — despite pandemic and recession, depression, war, people will not leave their partisan foxholes — is remarkable.
Latinos are rapidly emerging as a group that is the least partisan, the most independent, and the least racially polarized. Actually, they may be the seeds of what can help this American project get back on its feet and start being a functioning democracy again, as opposed to kind of walking down.
The reasons why that’s happening is largely because Latinos constitute the fastest-growing segment of the blue-collar working class, overwhelmingly focused on the economy. And those economic issues are not being addressed by either party, either in campaign tone or certainly in practice and results.
And so what we are seeing is this rejection. There’s truly a dealignment. A lot of discussion about realignment in our two-party system. Are they becoming more Republican? Are they actually becoming more DSA? Is MAGA appealing to them?
The truth is it’s neither of those, and it’s none of that. There’s a true emergent Latino populism that is redefining our political system, and I think it has the promise, or at least the potential, of getting us out of this partisan gridlock that has choked us for the past 30 years.
Alexis Madrigal: Hmm. I mean, one question I would have about this is you describe this as being about the economy and that both parties are failing on the economy, but might it also not be about immigration and the way that the Trump administration has pursued its goal of deporting many, many people in this country?
Mike Madrid: That’s a great follow-up question because, again, what I try to do is continually break down the mythology and the characterization of who we are as a community.
It’s great to be on the phone with two pollsters because I will tell you there has never been a credible poll in 30 years that has suggested that it is anything other than the economy that is the driving political decision-making process for Latino voters.
That does not mean that the ICE raids and the terror that our community is experiencing isn’t serious. There’s a need to find some sort of racial characterization for non-white people. I’ll be very direct. And it’s just not backed up by quantifiable evidence.
UnidosUS, one of the largest civil rights organizations, conducted a poll earlier this year that showed that it was fifth. It was fifth on the list of five issues: immigration. And fully one-third of Latinos wanted more immigration raids, more immigration crackdowns. So if you take them out, the number falls even lower, which is standard for what I’ve been seeing for decades.
So, yes, it is important. Yes, it is a motivator. It’s certainly becoming an economic factor at this point. That’s how bad it’s gotten.
But overwhelmingly, this is by a three-to-one margin, the economy, affordability, cost of living, are the issues driving this electorate. And I just feel the need to kind of really clarify that for people to better understand what’s happening in their country.
Alexis Madrigal: Michael Dimock, I mean, especially Pew does all kinds of issue-based polling as well, in addition to the sort of more electoral stuff. As you reflect on what Mike is saying about what’s driving Latino voters, what’s driving the electorate overall, what would you say?
Michael Dimock: Yeah. Oh, I think Mike nailed it. I mean, not only is this an election that’s predominantly around the economy and affordability, but that is often the case.
You know, if you had only one meta-indicator to try to predict an election outcome, it would be how people feel about the economy, and they’re going to hold the party in power accountable for that, particularly when that party holds the presidency and both chambers of Congress.
So that is the case. I think Mike is dead-on about the characterization of Hispanic voters.
And I would say I want to double-click on two points Mike made. One is the sort of myth that this is a population driven by immigration issues. Not irrelevant, but it’s the core issues that really matter: economy, education, the outlook for the nation, the outlook for jobs, just like it is for all Americans.
And secondly, the point about swing voters. It’s an old-fashioned term we used to use for people who actually enter an election cycle and are going to look at the choices in front of them and make up their mind, right? They don’t come in with their mind already made up.
And one of the biggest trends in American politics over the last 30 years, no surprise to anyone listening, is that there are fewer and fewer swing voters in most American elections. People are coming in on their team, locked in, and it’s not because they love their team. It’s because they’ve grown to be afraid of the other team, right?
And so it’s this negative partisanship that’s got a lot of the country really locked in. But you are seeing, you know, there are still some persuadable swing voters who are coming into elections looking at the choices in front of them. And I think Mike’s right, that’s a higher share among Latino voters.
Alexis Madrigal: Charles, I want to bring this one first to you. One listener out there has this criticism: “Why are we doing this show? The polls are always wrong, and even worse, all this kind of reporting keeps people complacent and home from voting. Polling is just conjecture. We all learned the hard way in 2016, and this kind of conjecture is just frustrating. The whole show is saying we really don’t know.”
How would you defend the honor of the pollsters, Charles? I’d love to hear it.
Charles Franklin: Yeah. Well, I think, first of all, it’s understandable to be skeptical of polls.
In 2016, Wisconsin, Michigan, and Pennsylvania were pivotal to Trump’s victory. And of 130-some-odd polls in these three states, only one ever showed Trump leading. So the skepticism is clearly there, and it really ballooned after 2016.
I can’t tell you how many times I’ve answered this exact question over the last 10 years. There’s a deep, deep skepticism about polls, and journalists constantly repeat the question. And again, rightly so. There are errors that have happened, and it’s hard to…
Well, we do quantify what the size of those errors are, and we issue big reports after each election about how well or how poorly the polls did. But those are very technical things. The average reader or listener of the news is unlikely to be motivated to dig down and say, “Oh, it was pretty good this year. It was pretty bad this year.”
So these stories come up all the time.
I think the broad picture that I would push a little differently is that we focus so obsessively on the vote margin. Did you get the vote margin right in this case? The winners and the losers, but also how close you came to the actual result.
That’s a valuable thing, but it comes out of this intense need to have a crystal ball that’s going to tell us what’s going to happen tomorrow and do it infallibly. And that’s asking too much.
But you opened the show by saying, what do our voters feel? What do they care about? And how are they going to vote?
And I would say those first two questions are the biggest contributions polling makes. How do we feel about the economy, about President Trump, about the candidates in our race? What do we care about?
The economy by a mile in our national polling, as well as in our state polling. It’s in the high 40% range who say inflation and the cost of living is the most important issue. The next most important is down at 10%.
How can we not pay as much attention to those measures of what people are thinking about and care about?
Instead, we pay so much attention to the vote margin and whether the polls are off by 2%, by 3% on average, by about 5%, plus or minus.
Those are important errors, and I don’t want to diminish that. But I think there’s a host of information we get that seems very reliable. I don’t think anybody would argue with a straight face that Americans are not concerned about the economy this year, or not concerned about gas prices or the war with Iran.
All of those things are important, and that’s the big contribution polling makes: understanding why we’re voting the way we do, not just what the exact margin in a race would be.
Alexis Madrigal: We are talking about polling, how it works, what it says, what it means ahead of tomorrow’s midterm election.
We’ve got Charles Franklin, director of the Marquette Law School Poll in Wisconsin. We’ve got Mike Madrid, a political strategist, author of The Latino Century: How America’s Largest Minority Is Transforming Democracy, also co-founder of The Lincoln Project. And we’ve got Michael Dimock, who’s president of the Pew Research Center.
We want to hear from you as well, whether you trust the polls, whether you have a question about the methodology behind polling. Maybe there’s a demographic that you’re worried isn’t getting represented in these polls.
Give us a call, 866-733-6786. That’s 866-733-6786. We’re going to get to a bunch of your questions in the next segment of the show. So email those in if you want: [email protected]. Social media, Bluesky, Instagram, Discord, or KQED Forum in all of those places as well.
I’m Alexis Madrigal. Stay tuned for more right after the break.
Welcome back to Forum. Alexis Madrigal here. Maybe you’ve seen the recent polling that shows Democrats looking about as good as they possibly could going into the election tomorrow. But, of course, a lot of people don’t necessarily believe the polls.
So here we are talking about polling this morning with the president of the Pew Research Center, Michael Dimock. We’ve got the director of the Marquette Law School Poll in Wisconsin, Charles Franklin. And we’ve got political strategist Mike Madrid, taking your calls.
Here’s a little packet of comments that have come in from listeners, and Michael, I think we’ll take it to you first.
One listener writes, “I think the failure to address spam and scam calling and texting has damaged trust in the most reliable avenue pollsters have to make contact. What sort of guarantee do I have that someone isn’t just claiming to be Pew Research and sending me a malicious link, claiming it’s a survey?”
Sharon writes, “I would like to take the surveys. However, each time I receive an email or text, I delete it, as there’s so much spam and malware out there. I’m afraid to click on it.”
Brenno writes, “I am a high-propensity voter. I don’t answer calls or click on links from unknown numbers. The last poll I did in the pre-caller-ID age was a push poll, which kind of soured me on the whole field.”
And Vivian writes, “I wonder if the huge amount of election marketing damages poll quality at all. I’d be open to answering a poll, but at this point, it’ll never get there because I ignore all calls from unknown numbers.”
What do you think, Michael?
Michael Dimock: Oh, you’ve got some savvy listeners. I sympathize. I’m the same. I don’t answer my phone most of the time.
I would say that this has been a challenge of polling all along. I mean, early on, a lot of polling was done by mail or by knocking on doors. And then people’s mailboxes got full, and too many door-to-door salesmen were coming around.
Then we all moved to using telephones in polling. And for a while, that was great, until our phones got bombarded with junk calls and, you know, scams and so forth.
And then a lot of polling moved to kind of online, and now online is polluted with all the scams.
So it’s interesting. Charles mentioned some of this earlier in talking about how polls are done today. One of the biggest and most interesting things I didn’t think I would see in my career is this sort of back to the future of the U.S. Postal Service.
A lot of the biggest national polls, including the Pew Research Center and many state polls, have gone back to doing that initial outreach and recruitment by mail.
And honestly, our mailboxes are less junk-filled than they used to be, because a lot of that cheap junk has moved online. It’s a cheaper way to spam people.
And you can actually communicate the credibility of your outreach much more easily through the weight of the mailing that you do. You can really show up in people’s mailboxes, and they’ll take you seriously with a really solid envelope and just clear branding and clear distinction from that junk marketplace.
So, again, that’s actually a lot of the big, serious polls that you might be interested in participating in. They’re actually going to prove it to you in a way that you should be able to take them seriously.
And I think your skepticism about the, let’s call it, over-the-transom outreach — you know, in a random email or text to your phone — I’m not saying that some of those might be legitimate. Those are actually good sampling methods in some cases, but you’re absolutely right to have the instinct that those are also spaces with a lot of spam.
Alexis Madrigal: What about — Casey writes in and says, “I think there’s a real need for paying participants in polling. It’s the norm in social science research, so why not in polling? Doesn’t have to be a huge amount, but it seems fair if polls are something a firm is profiting off of, and I’m expected to give my time to participate.”
Michael Dimock: Yeah, yeah, it’s totally legit. Neither Charles nor Pew Research Center are profiting off of our polls, but we do compensate.
So when we’ve recruited somebody in to participate, we’re not going to ask them to participate in maybe 10 or 12 surveys over the course of a year. And that’s a meaningful piece of their time.
So we do want to compensate people for the time and effort they’re putting into it. We don’t want paid employees, right? At a certain point, you know, we do still want the people represented in the survey to be that random cross-section, not professional poll-takers.
But, you know, a light compensation, a nod of respect for the effort that people are putting in, that is built into a lot of serious polling these days.
Alexis Madrigal: Let’s bring in George in San Jose. Welcome, George.
George: I hope you know that there’s more than one way to affect or predict the future. And another way that is often used is these gambling websites. You know, Americans love to gamble and play all sorts of things just based on their own personal observation and opinion, rather than measuring the population.
So what is that? What are your folks’ opinions about using the gambling websites as a predictor of election results?
Alexis Madrigal: Yeah, George, I appreciate that. I’m sure our pollsters have feelings. Charles Franklin?
Charles Franklin: Yeah, this will be a good year to test that because the gambling sites, as they’re called, prediction markets, are much more prevalent this year. And we’ll see in the next couple of days how good their track record is.
The main point I would make is, first of all, we have 90-plus years of experience with polling and when it’s been accurate and when it’s been wrong. We don’t have anything close to that experience with these prediction markets. So we’re just learning whether they’re good or not so good as a mechanism.
But the other point would be, even if they’re as good as polls or conceivably even better than polls, what they won’t do is tell you any of the why people voted the way they did.
All they would do is give us the estimate of how the vote’s going to turn out, but none of the motivation behind that vote, none of the considerations voters bring to it.
In the end, you can’t get that other than by talking to people, asking them, whether it’s online or on the telephone or some other mechanism. But it’s those motivational things that polling uniquely provides that the prediction markets completely cannot provide.
Alexis Madrigal: Mike Madrid, which races do you think will show us — like, that’ll kind of be indicator-species races — whether the polling was on target this time around, or whether there was a missed group of Democratic voters, a missed group of Republican voters? Do you have your eye on a few different House or Senate races?
Mike Madrid: For a whole host of reasons, I’m watching the Texas Senate race very closely because I think there’s these competing theories of whether this insurgent Latino, younger Latino, low-propensity voter is going to be found in the polls, but also this nonresponse bias consideration that Republicans have been relying on.
There’s this older white Republican group that’s less confident or trustful of institutions like polling and the media, that don’t want to respond to calls. And both of those theories will be pushed to the test.
So Texas, I think. But I also think, in many ways, we could be sitting on something of such significance, a wave-type election, that tends to paper over a lot of these issues. They’re not as apparent as they otherwise would be.
I’ll defer to the pollsters on this. My strong sense is a large victory has a way of kind of overshadowing a lot of the underlying rationale and reasons for why they actually came about.
But Texas, that’s Texas. Texas is the number one place I’m looking at.
Alexis Madrigal: You mean, like, if Texas is more or less a runaway, people won’t be looking necessarily, say, along the Rio Grande and saying, “Oh, it was these places that swung so hard.” Is that kind of what you’re imagining?
Mike Madrid: Yeah, I’m sorry. Yes, that’s exactly right. That’s what I would be looking for.
I mean, look, there are — there’s a whole race. When there are so many Senate races under contention, too, and especially in so many Republican areas, you’ve got to also look at notoriously difficult states to poll, like Alaska, for example. That’s going to be fascinating, and I think Michigan will be very interesting.
So there’s going to be a lot of things to be looking at, a lot of different variations. But, yes, inevitably, I think sometimes there’s a lot of explanations as to why things happened. And in a year that isn’t necessarily as much of a nail-biter — perhaps it will be, but if it’s not — I think there’s a lot of missed opportunities to look at what was driving the electorate in different quarters, in different places, at different times.
Alexis Madrigal: We’ve got Tom in Berkeley with a question along these lines. Welcome, Tom.
Tom: Hi. So my question is, you know, in recent years there’s been a lot of talk about underrepresentation of Republicans in elections because we see this discrepancy between polling and the vote count. You know, the shy Republican support, but all of that is simply conjecture. There’s no actual real evidence that that’s true.
I know that in other countries we use the discrepancy between polling and the vote counts as evidence of vote fraud or vote manipulation, and yet when it comes to our own country, if there’s a discrepancy, we just say, “Oh, the polling is wrong.”
Alexis Madrigal: That’s interesting, Tom. Yeah. Yeah. Go ahead.
Michael Dimock: I mean, I would hold firm that we should trust elections, not polls, when it comes to really judging the outcome.
In the United States, we have a very solid electoral system. It’s very distributed, of course, right? It’s very localized and at county and state levels. But it works. And there is a verifiable trail of evidence when it comes to the election outcomes.
I would say, even as a pollster, and I want to stand up for polling, polling is designed, as Charles has been saying, to understand the why, right? What are people asking of their elected officials?
They’re not actually well suited to predict behaviors. How likely is someone actually to cast that ballot?
And I think we would make a big mistake in this country to take polls as an indicator of whether the elections were actually accurate.
I think that dynamic can be different in different countries, where the veracity of the elections could be drawn into question. But I would not recommend that in the United States.
Alexis Madrigal: Here’s a question for you, Charles Franklin, and maybe we’ll go around the horn a little bit on this one.
The question is, Trump clearly had some effects within both polling and the actual elections, like the combination of those two things, that are really quite unusual, at least in the history of American politics.
Do you think that after this election, and after Trump is off the political stage, whenever that day actually and truly comes, that there will be a kind of reversion to the way things used to work? Or was there an underlying kind of structural, technological, or social change that was sort of enabling that, and Trump was somehow calling upon this thing and making our polls seem less accurate when his name was on the actual ballot line?
Charles Franklin: Yeah, well, I think there’s one simple answer in that: the midterm elections have been more predictable. The polling has been a bit more accurate than in the presidential races. So that’s a clear indication that his presence on the ballot has had an effect.
But I also think we imagine that voting behavior changes are more permanent than they maybe really are.
The example is, I don’t think there was any question that President Trump has mobilized a modest group of voters, but a critical group of voters, maybe 4% to 6%, who previously didn’t participate but were brought in by him. The low-propensity Trump voters, with him gone or just off the ballot in a midterm, we find that those folks participate less.
Mike Madrid has been talking about Hispanic voters being unusually mobilized. That’s a great example of another low-propensity group that can be brought into the electorate by circumstances and the issues that are driving the election.
And so maybe what we need to do is appreciate that these things aren’t permanently baked in, but can change and can change from election to election and can certainly change after President Trump is no longer in office.
Alexis Madrigal: Yeah. Mike Madrid, let’s come to you. I mean, I imagine, just based on your previous answers on this show, that you think there is a bit of a sea change because of the growth of the sort of Latino — I don’t want to call it a voting bloc, but just the rising importance of the Latino vote.
Mike Madrid: Yeah. Let me give you a specific example of what Charles was saying.
There are now more polls that are looking at Latino voters. Pew, incidentally, Mike’s job is doing really the gold-standard work on this stuff nationally.
But one of the innovations they introduced was polling Latinos by generational differences, not just country-of-origin changes. They recognized very accurately that first-, second-, and third-generation Latinos have very different political perspectives.
So even though they all may have the last name Hernandez, it’s not necessarily a representative sample until you stratify that for generational change.
That was one of the problems that really afflicted the Hillary Clinton campaign and Joe Biden’s campaign because they had dramatic oversamples of first-generation, recently naturalized Latino voters. And that showed a very, very significant emphasis on immigration and migratory issues when we knew quantifiably that the economy, by a three-to-one margin, was what was driving the electorate, as I mentioned earlier.
And that changes the focus of campaigns. And if you’re not getting that polling accurate, if you don’t understand that nuance, especially as it’s growing as a share of the electorate, you’re going to come up with some pretty significant misses.
Alexis Madrigal: So interesting. One last question here for you, Michael Dimock.
One listener writes, “Is there any understanding of polling impacting turnout? I’m thinking of the narrative, ‘We’re already winning, so my vote isn’t critical.’ I can actually imagine it going the other way, too. ‘Oh, well, we are already going to lose, so I’m not going to turn out.’ Do we have any sense that polling has an effect on turnout, positive or negative, depending on how it seems like one’s preferred candidates or parties are doing?”
Michael Dimock: Yeah, it’s hard to really pin that down.
There’s probably a good sense that a close election, right, when the polls are predicting something to be close, whether it’s the individual race in your district or the aggregate, like we’re talking about now, what is the seat share going to be? That grabs people’s attention, right? It makes your vote feel more consequential, and turnout is likely to go up in those circumstances.
But the idea that, you know, you’re right, you can argue it both ways. It’s the people who feel like they’re safely winning. This was talked about a lot in 2016 with Clinton voters. Maybe they didn’t feel like they had to vote, but you can say the same for the folks who are slightly behind.
I guess one thing I would leave on, though, is we have been going through a series of incredibly high-turnout elections, historically, in America, right? The voter engagement in elections in these cycles is super, super high, and it’s spreading across segments of the population who used to be low-propensity voters, less engaged, less likely to participate.
So the composition of the electorate is broader. I think it’s harder to predict for pre-election pollsters who’s actually going to show up and not. And that’s a part of why a lot of people have felt uneasy about pre-election polls’ ability to actually nail what truly is a moving target in terms of who the electorate’s going to be.
Alexis Madrigal: Yeah. Charles Franklin, how are you going to watch the election results come in? How do you do it?
Charles Franklin: I’ll tell you. I worked at ABC News on the decision desk, calling elections for 20 years. I’ve retired from that, so I sit at home with a big bowl of popcorn and refresh every website I can find. I refresh every five seconds.
Alexis Madrigal: Mike Madrid, how about you? How do you do it?
Mike Madrid: Election night? Yeah. Usually it’s just a mess of different media appearances and jumping from radio to TV to podcasts. But I’m not — I was just getting some notifications on some changes here last minute. So I never know. I don’t have a tradition. I just kind of go where I’m called to go.
Alexis Madrigal: Yeah. And Michael Dimock, do you have a preferred data source? Like, are you someone who’s glued to X or Bluesky, or are you more like, okay, I’ve got a data dashboard. I look here, and that’s how I stay calm and centered?
Michael Dimock: I am a really strange pollster and data geek, which is I’m not actually that addicted to the election-night bit. I’m actually pretty comfortable going to bed early and waking up in the morning and seeing what happened.
Alexis Madrigal: Yeah, I don’t think I’ll be doing that, but I can appreciate the sentiment. That is very reasonable. Very, very good for the body and mind, I’m sure.
We have been talking about the latest polling ahead of tomorrow’s midterm election. You can get those ballots in, of course.
We’ve been joined by Michael Dimock, president of the Pew Research Center. Thank you very much.
Michael Dimock: Thank you.
Alexis Madrigal: We’ve been joined by Charles Franklin, director of the Marquette Law School Poll in Wisconsin. Thank you so much.
Charles Franklin: Thanks. I enjoyed it very much.
Alexis Madrigal: And we’ve been joined by Mike Madrid, political strategist and author of The Latino Century: How America’s Largest Minority Is Transforming Democracy. Thanks for joining us this morning, Mike.
Mike Madrid: It was a real pleasure. Thanks.
Alexis Madrigal: And thank you so much to all of our callers, people who commented or left us questions, called in. Really appreciate it.
I’m Alexis Madrigal. Stay tuned for another hour of Forum Ahead with Mina Kim.
Announcer: Funds for the production of Forum are provided by the Jenerosity Family Fund and the members of KQED.