Program transcript:
Grant Reeher: Welcome to the Campbell Conversations, I'm Grant Reeher. My guest today is the pollster, John Zogby. Mr. Zogby is the founder of the Zogby International Poll and with his sons, Jeremy and Benjamin, he serves as senior partner at John Zogby Strategies. He's also the author of a recent book, it's titled, “Beyond the Horse Race: How to Read Polls and Why We Should”. Mr. Zogby, welcome to the program.
John Zogby: Hey, Grant. Good to talk to you again.
GR: Good to talk to you, always good to see you. So, my questions are going to be driven by your book and also your background. They'll also be driven by my own background, as you know, I’m a political scientist. So, I'm going to draw on some of the discussions I hear from my colleagues when we're talking about polls and surveys, get your take on those. And then I also wanted to ask you some questions about some recent survey and poll data that's been out there and has been in the news. So let me just start with a basic question here, one question that always comes up a lot regarding polling in recent decades is how all the big changes in communication technology have complicated it. I mean, let me just throw out a couple things. You know, landlines are basically gone, people don't answer the phones if they don't know who's calling them when they see the number coming in or the person. It may be harder to find certain demographic groups to tap for their opinions and so on. So, I just wanted to get your sense as a pollster, what in your view are the biggest challenges for polling that have arisen since those golden days when we all used to have landlines and answered the phone when it rang?
JZ: Well, of course it's technology and non-response, or low response rates. When I started in business in 1984, response rates averaged 65%. Two out of three that you reach, actually reached by phone we're willing to conduct a survey. However, change has been part of this business from the beginning and the pace of change has increased greatly. So, you know, imagine in the 30s, 40s and 50s, folks like Gallup and Roper and Harris would develop sampling points throughout the United States, identify people in each of those sampling points to travel to those sampling points, mail envelopes to them with survey questionnaires, instructions on First and Elm Street, fifth door on the left-hand side, and conduct the survey over 3 to 6 weeks, mail the surveys back. And that's, you know, providing everything that went out came back, then you'd get results. Well, our world got faster, things move more quickly. We moved to the telephone. As you pointed out, that was a perfect world. 95% of people had a landline telephone and socially, they welcomed a long-distance call, making them feel important. And then that all dissipated. The answering machine, *69 when they would call you right back and find out who you were, call waiting, that sort of thing, and we had to adjust. You know, by the mid to late 90s, response rates had plummeted to 15%, in major metropolitan areas, 7%, 6%. So, we have always had to adjust. The internet was not welcomed just as, incidentally, the telephone wasn't welcomed in the beginning. How are you going to know who you're talking to? Same thing with the internet. And now, of course, you know, technology has changed rapidly and people who conduct polls can't wait 3 to 6 weeks. Frankly, Grant, they can't wait 3 to 6 days. They want their data and analysis right away. We've had to make those adjustments. So, for every advance, and it has been an advance, there have been detractors.
GR: So let me ask you a question about the phones specifically and how things work today. Has there been discussion, among you and your polling colleagues about simply just moving away from phones entirely to, say, the internet? I'll just tell you, political science research in recent years has essentially done that. That all political science survey is done by the internet, but that has introduced a whole new set of problems. But how are you folks dealing with it? What do you do with that?
JZ: We have well, we had a very large 124 phones call center in Utica. We don't have that anymore, we farm it out, you know, to call centers. I mean, that's obviously one change, but we encourage as much to be done by internet, by email as possible. We also use text to web, so we'll send out emails, we'll also send out texts, but ultimately inviting people to a secure web site to conduct the poll. To answer your question directly, there are still some applications where you really can only do it by phone. If it has to be spontaneous, if it has to be ultra confidential, not to say that the internet can't be confidential, but there are applications, very small areas or very targeted audiences, you know, may require the phone. But I would say, you know, of our survey work, 85%, 90% is online.
GR: I'm Grant Reeher, you're listening to the Campbell Conversations on WRVO Public Media, and my guest is the pollster, John Zogby. He's the author of a recent book titled, “Beyond the Horse Race: How to Read Polls and Why We Should”. So, polling regarding the elections always involves this subset of people that are called likely voters, whether that's likely voters in a primary or likely voters in a general election. How do pollsters identify those people, and does that introduce some kind of inevitable error or challenges? I'm remembering, I think I'm remembering this right, but you correct me, I believe that was one of the problems in the early polling that thought Dewey was going to beat Truman, right? Was the guess about who was going to be voting in that election. So how do you folks deal with that?
JZ: Well, first of all, that's 1948, a long time ago. And, the last polling was done two weeks to ten days before the election. And so, those who have read about it or may even remember it, that's when Harry Truman was on his famous whistle stop, and he had built momentum at the top, you know, two weeks before the election. He felt he was going to lose the election. That's why he manically, you know, traveled throughout the country. ‘Give ‘em Hell, Harry!’. And so, there was the issue of timing as well. And to a great degree, there still is the issue of timing. So, first of all, likely voters we ask, we have 3 - 4 question screen that we use to determine if they’re likely voters. Now, of course they could be lying, but it has served us well. You know, what does anybody get out of lying to a pollster? And then how do they create a conspiracy of a thousand people who are also lying? That's, it doesn't happen. You know, I think folks like to beat up the polls and say, oh, you were five points wrong. But when you're talking to 1500 people, five points wrong ain’t so bad. You know, when you think about it, you at least get a direction. So yeah, we poll likely voters when it comes to politics and policy because they're the ones that matter.
GR: And you mentioned some of the pollsters getting beaten up recently. And I wanted to ask you kind of a more specific version of that. There has been a lot made of the supposed inaccuracies of polling in very recent years, and particularly regarding Donald Trump and elections that either he's in or he's been heavily involved in. And there's been a lot of different kind of effects that have been hypothesized to be going on that might be driving some of that, maybe people, you know, a bias and who's responding and not responding. And like you said before, maybe not necessarily a conspiracy, but a general sense of, you know, I'm going to tell the pollster some garbage because they don't like Donald Trump or whatever. What do you think's going on there? Is there anything real in that? Is there a Donald Trump effect of some sort?
JZ: No, I don't think so. I think if we close our eyes and picture, you know, the red, ‘Make America Great’ cap again, Donald Trump has a lot of people who are ready to jump and scream and write and attend rallies and be proud that they're for Donald Trump. That secret Trump effect, maybe you could have said at one point on college campuses. A professor may not have seen it in his or her best interests to be wearing a MAGA cap, or to sit at lunch, in a...
GR: (laughter) That's putting it lightly. I can tell you from my experience at the university. Not that I would put on a MAGA cap, but I could just tell you if someone did, their life might be in danger. But anyway, go ahead.
JZ: But, you know, there hasn't really been anything measurable. Donald Trump finishes strong. And the thing is, many of the pollsters conduct their last poll Friday or Saturday before the election. I try to keep the lights on longer, so I'll go till Monday night. Sometimes I've even gone into Tuesday morning, meaning, of course, Election Day to capture that last minute spark or whatever happens, or the last-minute lack of spark where you begin to see. Let me give you a prime example, you know, in January 2008, the focus in New Hampshire was Obama having won the Iowa caucuses against Hillary. And usually a candidate gets a big, huge bounce after Iowa, and then there's two weeks into New Hampshire. This one year, Iowa was on Thursday and New Hampshire was on Tuesday. And so, there was momentum up to Obama, you know, by Sunday night leading by 12 points. And then, Monday night, Obama is still leading by 12 points, we get to 5:00 and our last 125 calls had Hillary leading by six. So now what do I do? You know, do I report that, or do I do what the responsible thing is? It's a running average. I've got to average the previous 24, 36 hours in with that and I still came out with an Obama double digit lead. Hillary won by six. And so, we all got it wrong, we all got it wrong. And part of the problem is, there are some folks who love to bash the pollsters, you're only as good as that last number you put out. No nuances. Those are seen as excuses.
GR: Well, so quick follow up on that and then we'll have our break in the middle. But, so just to clarify, you do not think though, as I have often heard and read in the media, that there is a systematic underreporting of Donald Trump's support in polls, not approval rating, but you know, trying to predict what the voting outcome is going to be.
JZ: I do not. But there is a noticeable difference between our polling and a few others, and other polling, mainly like the networks and so on, over-representing Democrats under-representing Republicans.
GR: Interesting, okay. You're listening to the Campbell Conversations on WRVO Public Media. I'm Grant Reeher, and I'm talking with John Zogby. The pollster is the founder of the Zogby International poll, and with his sons, Jeremy and Benjamin, he serves as a senior partner at John Zogby Strategies. He's also the author of a recent book, it's titled, “Beyond the Horse Race: How to Read Polls and Why We Should”. I want to, John, to pick up on the point that you made, which was that the mistake that pollsters tend to make these days is not the one that I had identified in terms of some specific Trump effect, but rather they oversampled Democrats as opposed to Republicans. Why is that happening? What's going on there?
JZ: Initially, it was a demographic thing. Back in the 80s, you are more typically a Democrat as a blue collar person and as somebody who, not afraid to share opinions and wanted to feel important that somebody was asking you. Republicans more than likely to be, you know, suburban, professional, and not wanting to be bothered on the telephone. The interesting thing is, though, that I determined that an adjustment needed to be made for a political party. That we were getting, maybe 1 or 2 points more Democrats, maybe 2 or 3 points fewer Republicans, that's a 5 or 6 point swing. And given the fact that 90% of self-identified Dems are going to vote Democrat, then same thing with Republican, that could impact the election result. What I was finding in my colleagues and I mean, the top of the line colleagues, was that they were coming in with 15, 16, 17 point swings, like 42, 44% Democrat when it should have been like 36, and 28% Republican, 24% Republican. And you know, that bothered me, because that was really skewing the results. There was a seeming real Democratic bias in it. And there were two implications. One was folks at the New York Times told me they used out of work actors and artists on the telephone to make their calls, and immediately, boy, you know, some middle class, some blue collar woman would say, “ Good afternoon, I'm doing...”
GR: Right, right.
JZ: The other thing is that, personally, I'm a progressive Democrat. My firm has always, I hope, been independent. And I became, like, the darling of Republicans.
GR: For pointing this out.
JZ: For pointing that out. And Rush Limbaugh and the New York Post and all of that, you know, adopted me. It was said at one point that Republicans are going to be naming their firstborn, Zogby at some point. And it stuck in my craw a little bit, but that was the honest thing, that it was more of a demographic and partisan thing, and not based on whether the candidate was black or, like the Bradley effect that we talked about or Obama effect or the Trump effect that they talk about now.
GR: That's fascinating. And, well, you also have to remember, John, the enemy of my enemy is my friend, and so that's why that happened. But it's fascinating that political polarization amplifies this problem. I hadn't really, I hadn't made that connection before. Political polarization is just involved in so many different things, it's fascinating. Let me move to a different kind of polling, which is approval ratings and President Trump's recent approval ratings, in particular. They've been very low, dipping below 40% in some recent surveys. Is there a particular number regarding those that suggest either a tipping point or an inflection point? Is there like a panic level for Republicans in terms of Trump's approval rating and have they hit that?
JZ: Well, generally, if you hit, you know, in the high and mid-30s, that's the real trouble. That's on the surface. When you look at those numbers, though, and you see that Trump was elected with over 50% of the vote, that's the second real trouble. That means he's dropped 17 points. When you realize that 42% is the MAGA base, and there are polls that have him well under that MAGA base and getting trounced among independents. You have to ask, where is he going? Where's he going with numbers like this? So, it's more a dynamism than it is a hard and fast rule.
GR: Right, placing it in context like that, yeah, that's a very good point. You know, I'm sure you saw this, Gallup recently announced that it's no longer going to do its regular approval ratings of the president. It’s been used for so many years, 88 years, I think, and they're going to stop that for other officials as well. What's going on there? What do you think is happening?
JZ: Well, you know, several years ago, after the aggregators, you know, the Nate Silver’s, the Real Clear Politics of the world where they aggregated averages of polling and waited for certain track records and so on. The focus became too much on, did you get the election right to the percentile or did you get it right within a point and a half? A whole point and a half. And Gallup said that they weren't going to do the presidential horse race anymore because they were tired of being embarrassed.
GR: Okay.
JZ: Pew followed with the same thing. Now you won't see any daily tracking from any of them. Now, you're not going to see anything from Gallup, I think Pew will follow shortly. I stopped, doing, you know, the daily tracking or the frequent polls. Number one, we didn't need it anymore and number two, like Gallup, I didn't need anybody telling me, oh, you got the last election within 1.8, that placed you 23rd among pollsters for having a 1.8 differential between your polling and the final result. So, the focus became too much on, I think the Icarus syndrome. The aggregators had a few elections where they flew too close to the sun and, you know, this kind of science I don't think was ever designed to tell you exactly how much the election, down to the 10th of a percent. I was fortunate to get a lot of those right, but I don't consider myself wrong if I'm off two points.
GR: Gotcha, yeah. Well, yeah, the point prediction problem. I assume, though, that there will be organizations that will tell us how many people, or what percentage of people are approving of Donald Trump, right? That information will not be lost to us completely, correct?
JZ: No, it won't, but there is a further problem, and I promise I'll take 30 seconds on it, and that is the standard deviation. So, we're getting at a 42% average when you'll have five polls that have 47%, that neighborhood, and five polls that have 36-37%. That's not the way you should be getting a mean, you know, of 42 or 43%. You should have them mainly clustered around that average number, then here and there, an outlier. So, I believe what's happening is you've got polls with a distinct Democratic bias, polls with a distinct Republican bias, and what they're doing is they're coming up with, I think, what is an artificial average.
GR: Very interesting. If you've just joined us, you're listening to the Campbell Conversations on WRVO Public Media. I'm Grant Reeher, and my guest is the pollster John Zogby. Let me ask you one more question in this vein, then we'll turn to something else. But the generic ballots, these are, you know, if the election were today and all you knew was one candidate was a Democrat and the other one was a Republican, who would you vote for? And they're heavily used in the midterms, and they've, been a lot of discussion about these in recent weeks that the Democrats have opened up, you know, a pretty good lead, historically, over the Republicans. I think it's ranging from, you can correct me if I'm wrong, like 3 to 6 points right now, depending on who's doing it. Does that spell real trouble for the Republicans in the 2026 midterms?
JZ: Yes, today it does. So, let's look at that generic ballot. It's not the greatest measurement, but it is very useful. And overall, because of gerrymandering and because of Republican domination of state legislatures who do the reapportionment every ten years, the Democrats need to lead by four and a half to 5% to be nationally, to be in the territory where they're gaining seats. So right now the average is at about 5% lead, which means they're not in comfortable territory, but they're in, you know, they're heading in that direction. When you add to that, though, the special elections that they have won, or at the very least overperformed, say in very red districts, Trump may have won the district by 20 points, Democrat lost in a special election for whatever it is, state legislature by four points. Then you're seeing that, and I think that's mainly on the basis of independent voters swinging way over to the Democrats now. But there's a long way to go, a long way to go.
GR: Interesting, yeah. I'm thinking of the special election in Tennessee where Matt Van Epps won, but it was much closer than expected. So, we've got about, oh, really just a couple of minutes left, unfortunately. I could talk to you all day about this stuff, but, in a couple of minutes left, I want to turn to a couple of things that come more specifically out of your book, and I'll start with the title. If you could briefly just tell us, other than the fascinating conversation we've had so far, why should citizens pay attention to all these polls?
JZ: Well, I think we want to feel connected. We want to feel that, hey, this is my opinion, but is the majority with me? Is the minority with me? Am I part of a very vocal, intense minority? But I want to feel connected. I want to know where my opinions stand as far as the Main Street is concerned. And, you know, secondly, I do want to have some kind of an idea of who's ahead. I don't need to know who's winning, or going to win, not this early, but, you know, have some sort of an idea where voters nationwide stand. That is supposed to prevent the notion, oh, he'll never win, nobody at the barbershop supports him.
GR: I see.
JZ: Which usually is translated into, nobody in America supports him. Why? Because I was at the barbershop.
GR: Right. Or my grandmother or whatever.
JZ: My grandmother, who was usually, you know.
GR: Right. So, we've only really got a minute left. I mean, these are going to have to be sort of our lightning round. What was your worst mistake as a pollster? I know there's a longer story there, but what was your worst mistake? You're very open about your mistakes.
JZ: My 2004 election had the numbers absolutely correct, Bush over Kerry. I thought Kerry was going to win and said so.
GR: Interesting, okay. Finally, things pollsters hate, you've already indicated why, but I do want you to make some just one word point predictions for 2026 midterms. In the House, who comes away with the majority? How many seats?
JZ: As of today, Democrats 20-plus seats.
GR: Wow. That's a big shift. And then what are the chances, percentage-wise, in your view, that Democrats could retake the Senate?
JZ: Less so than the House, but I think it's, I think they're both in play. In fact, there’s some, you know, some new races that weren't expected.
GR: Interesting, interesting. Okay, I’ll have to leave it there. That was John Zogby, and again, his recent book, it's titled, “Beyond the Horse Race: How to Read Polls and Why We Should”. John, thanks again for talking with me, this is fascinating. I had fun and I also got better informed, thank you.
JZ: Thank you. Grant.
GR: You've been listening to the Campbell Conversations on WRVO Public Media, conversations in the public interest.
