Podcast

Borg: Measuring and analyzing public opinion is a science far beyond pondering comments around morning coffee at one Iowa's restaurants. Iowan George Gallup, born in Jefferson, University of Iowa journalism grad, is credited with developing public polling. But pollsters such as Ann Selzer, aided by sophisticated technology, are enhancing the credibility, the potency and the effects of those polling results on political campaigns. Ann Selzer, welcome back to Iowa Press. And I will say that at one time we had the late George Gallup here as a guest on Iowa Press. In fact, he rode over from Iowa City in my vehicle and that was a real pleasure talking with him. Would he, his methods survive in today's polling?

Selzer: Well, yes, to a certain extent. Obviously things have changed. He was doing polling door-to-door, which we don't do anymore. That wasn't until 1978 that the Des Moines Register started using the telephone as its way of gathering data. So that part, but I think he probably made that transition okay. What he said is, as long as you as the investigator are choosing the sample, giving everyone who is in your meaningful universe an equal chance to participate in the poll, then this is science-based. And to the extent possible, I'm a science driven pollster, and that is always in my mind. It's like George Gallup sits on my shoulder. There are other pollsters who are taking other approaches. That's certainly true. But you can still do it by science.

Borg: We'll get into that in our conversation. I want to introduce to our viewers the journalists across the table, James Lynch who writes for the Gazette published in Cedar Rapids and Radio Iowa's News Director Kay Henderson.

Henderson: Well maybe we'll set aside science for a little bit and get the magic wand out. If you were able to determine the criteria by which candidates would participate in these televised debates, what criteria would you use?

Selzer: Well I was astonished when I saw the methodology that first Fox was going to use and CNN with its own little spin just struck me as they didn't have a pollster at the table when they were deciding to use national polls to do it. I care about a meaningful universe. That is whose opinion matters in how we're going to select the president? So having a national poll, we don't nominate presidents through national voting in any way, so right from the get-go I'm going, well that's kind of crazy. And then it sort of worked out. It turned out to be a robust way of measuring popularity. If it were up to me, I would have given part of the equation to Iowa and who was polling well in Iowa, part of the equation to New Hampshire and who was polling well there and then I might take a look, broaden it, and look at the Super Tuesday states. Obviously South Carolina and Nevada are in there as well. But those are the states that are most likely going to be the ones deciding who the nominee is going to be and I would have figured out a way for them to have more voice.

Henderson: You, at the Des Moines Register, are going to be hosting a debate. Do you have a role in determining the criteria for participation in that forum?

Selzer: I do not. The Des Moines Register is my client. If they ask my opinion I will of course give it to them.

Lynch: The approach you're talking about though would ignore some of those big states like Texas, New York, California, where a big part of the electorate lives.

Selzer: Yes, but here's the problem with including Texas and California and New York and Florida, if you put all of those together in a general election they're about a quarter of the vote. So a candidate wanting to do well in national polls could concentrate their efforts in those big states and end up with a better national poll standing than they could ever get in Iowa or New Hampshire. Now, that would be an expensive way to go, don't get me wrong, but you could buy your way onto that debate stage and have it be a little bit less linked to the kind of public opinion that actually matters.

Lynch: And those states don't really matter in the nominating process too much do they?

Selzer: Well, we respect them and we like them and the math of it just turns out that they're not necessary to the process.

Henderson: Before we go any further, you just in a passing way mentioned it, but you're not technically employed by the Register. Explain the relationship.

Selzer: No. I'm the President of Selzer and Company and when I first moved back to Iowa I was on the staff of the Des Moines Register. And so we have a very close relationship. When they decided to outsource the poll my company won a competitive bid. So from 1997 until now we have been directing the Iowa Poll.

Henderson: And you're polling for Bloomberg and other media organizations as well.

Selzer: Separately. I've had a contract with Bloomberg, it is now Bloomberg Politics, since 2009. And so there was a happy moment when both sides said, let's do the caucus polling together. And caucus polling is the most expensive you can do. And so having multiple players who can bring some cash to the table to make it happen benefited everybody. We could do more polls.

Borg: Let me get back to the conversation that you and I opened the program with, but getting into the integrity of the poll. In the age of cell phones and even maybe polling fatigue on a part of people who continually get phone calls, it is phone calls that you depend on? How do you do that and ensure a broad spectrum in the era of cell phones that George Gallup didn't have to contend with?

Selzer: Well, cell phones were a problem initially because relatively few people had them and the work that the Pew Research Center has done, and they're a national treasure, said, if you leave out cell phones you're still going to be okay as long as you get age correct. And so we could ignore them. And then once the recession happened and people started dumping their land lines as they were cutting household expenses you could no longer ignore them. But we can purchase cell phone samples the same way that we purchase land line samples and we can random digit dial into them and in many ways they help us with the younger voter who were harder to reach just looking at land lines.

Borg: So you ensure that you get a demographic cross-section?

Selzer: What we do is to take our best shot. So that is my goal, which is there's a group of people who are going to show up on caucus night and everything that I do is designed so that every one of those people has an equal chance of being contacted by my poll. And we screen to be sure that we think they're a likely caucus voter. So, Dean, anything can happen. So I don't want to ever be too arrogant about it. But what I don't want is when things are all said and done to know in the back of my mind there was something I could have done that would make my poll better methodologically and that I didn't do it. So we take our best shot and the chips fall where they may.

Borg: And you're pretty happy with the results.

Selzer: So far things have worked out.

Lynch: Ann, you talked a little bit about the expense of caucus polling, but one of the things that strikes us as reporters is every day in our inbox there seems to be at least one more new poll, sometimes more than one a day. Are there too many polls? And does that sort of spoil the impact and the integrity of the polls?

Selzer: Well thank you for bringing that back up because that was part of Dean's question too about polling fatigue. I worry about this more than anything this year because you do have this proliferation of polls and people are doing robo-polls, things that are very easy to do and inexpensive to do and easy to do a whole bunch of them. Not all these polls are operating the same way that we operate. So I do worry about poll fatigue and every time I give a talk somebody raises their hand and says, I don't participate in any polls and yet I go to caucus so your poll can't possibly be accurate. And I say, well, thank goodness that there's a doppelganger who has been voting just like you because if there were some bias in that I wouldn't be able to be as accurate as I have been. But I do worry about it. The one thing we have going for it is when we call you up we say, we're calling from the Iowa Poll and it's a name that people know and our cooperation rates so far have been good.

Lynch: But response rates overall have been going down. I saw some numbers that back in the mid-90s it was like a third or better would respond to pollsters. Not it is down in single digits. How does that affect your results, again, going back sort of to the integrity of the poll results, if you have to call ten people to get one answer?

Selzer: Right. And you would think logically that it isn't possible. And all I can do is look to the data. And so far it has not gotten in our way of getting accurate results. That could all change the next time we have a caucus, the next time we have an election. It's something to worry about. There's not a lot I can do about it except worry. So I'm doing plenty of that.

Henderson: Do you worry about how the polls influence the campaign itself?

Selzer: I don't. That is one thing I don't worry about because there are lots of things that influence the campaign The debates just influence the campaign, the advertising is influencing the campaign, what your next door neighbor is saying is influencing the campaign, the polls maybe they're a piece of it. I've seen no definitive data that says they help an underdog or they help the leader in a bandwagon effect. To the extent that they influence, I don't know it's one way or the other. But it's just one thing out of a whole lot of things that are going on.

Henderson: What do you make of these aggregators, the poll of polls?

Borg: What are you asking there, Kay?

Henderson: Well, she'll explain it.

Selzer: Well, a poll of polls is a bit of a misnomer because they’re not polling a poll, they're just aggregating. That is, they’re taking a number of polls and then they are adding them together, some of them with some algorithms in a more sophisticated way. But what they are basically saying is there's sort of the law of big numbers, which they'll eventually sort of hover around something that is true. I don't think it is any more true than anything else. So I think in some ways it is a little misleading that people look at the average and say, well your poll is 5 points off the polling average, you're an outlier, it can't possibly be right. And I've had a number of times where in fact the polling average was wrong by several points and my poll was pretty doggone close. So I think they mislead a little bit in making people think well there, that's the number I should think.

Henderson: Speaking of misleading, there is something called a margin of error. And a lot of times people will say oh, somebody has surged to the lead, when they're still within the margin of error with that person who is in the second slot. What are your thoughts on that?

Selzer: My thoughts on that is that the most, the superior way for us to talk about the race are exactly what the numbers are. So if so-and-so has a 2 point lead, the best way for us to characterize that race is that it is a 2 point lead. And my little bit of a soapbox with reporters is they want to say, well that's within the margin of error meaning it could be tied. It is equally likely that it is farther apart, that that leader has an even bigger lead. And the point I make to journalists is to say, if you're only playing one side of that margin of error that is bias. That's a little bit of a wish that this is a closer race. If you want to say could be tied, could be farther apart, well that's equal and balanced. But the best way to characterize the race is to just say what it is. And if someone has moved from having 10 points to having 18 points that is a surge, even if it is within 2 points of somebody else.

Lynch: Without getting too far into the weeds here I want to ask you --

Selzer: Too late.

Lynch: Probably. You've talked about averaging polls and one of the other things that some pollsters do is weighting the results so that it reflects the known demographic. And I've heard you say that little old ladies are more likely to answer pollster calls than younger people, for example. So how does that work? If you're calling people and you're not getting people that match the demographic, how do you make it reflect all of Iowa?

Selzer: Okay. And I'm going to try to do this without the weed part so stop me if it gets too much. But what we do is to weight to known demographics. That is, when we call people and they don't qualify as a likely caucus-goer, that is they don't say they'll definitely or probably go, we collect their age and sex information so that when we're done, for everybody we talk to, and that will be thousands of people, we have a group we can weight to because we know what that population actually looks like in reality. Then from that big pool we just pull out the likely voters. And that helps us make adjustments for the little old ladies, bless their hearts, and get things in the right proportion for the general population. But that's, there's still going to be differences between what our sample group looks like and the general population, but we have taken care of that.

Borg: In my way of explaining that, and not your technical way, would it be that you collect a lot more information than you're actually going to use because you might go back and pick from the pool some opinion that was given that you maybe hadn't intended to use in the first place?

Selzer: No. Because if you're not a likely caucus-goer we're going to hang up on you, politely of course. But if we know that you are something in terms of your age, I won't guess, and male, then you're out here and we have a big population and now we have that properly weighted to look like the population we know. And then when we pull out the likely voters then whatever is interesting about how likely voters differ from the general population we haven't messed up.

Borg: Do you take their word for it that they're likely caucus-goers? Or do you discern this person may be saying they're going to the caucus but I know they're not?

Selzer: You see. When I listen to you say that I think, well how arrogant would that be? How do I know? And we all at this table remember in 2008 that the democratic caucus had 60% of the people who had never caucused before. So anything that a poll would have done that they said this would be their first caucus, to say well you're not likely to come, it would have been a problem. So our polls scared people, shocked people, including myself when we said there was going to be about 60% first-time and I think the entrance polls showed 59% and I might have those two numbers reversed. But we would have been wrong to insert our assumption about how many first-time people show up on caucus night. We let our data tell us.

Borg: If I say I'm going to the caucus you believe me.

Selzer: What else do I do?

Lynch: You use terms like likely caucus-goer, registered voters, likely voters. What do those mean? And how does that affect the outcome and the accuracy of your poll?

Selzer: Well we do have a lot of people who are polling different universes. And, again, I keep using that term, that what is the population of interest. And it is easier, less expensive, to talk to registered republicans, for example. Well, you can register on caucus night and you can change your registration on caucus night. So to just talk to people who are currently registered republican you're talking to a bigger group than will actually go. So for us we have to, you have to say you will definitely or probably go to caucus. And some will just say I'm somewhat likely to go or I might or might not, again, because it gets more people in the pool, they're hanging up on fewer people. We have been doing this a long time and our methodology has worked for us and so we want a smaller rather than a larger more amorphous group.

Henderson: Speaking about a larger group, the discussion under current days has been is Joe Biden going to make the jump in the pool? What does your polling show about his favorability rating in Iowa? And given your history polling here, is his best day the day before he jumps in the race?

Selzer: Well that I don't know. What I do know, we included him in the last horse race question we did here in Iowa. I think he got 14%, somewhere less than 20%. He has run in Iowa twice before and that number that he just got in our most recent poll was better than his two best findings put together from the other two times that he ran. So people said, well that's not a very good number. I said, well it depends on what context you're looking for. He's not an announced candidate. You still have people who are excited about Bernie Sanders who probably would think twice about making a shift. You do have people who are longtime supporters of Hillary Clinton. They're going to think twice about that. So I don't know that I have anything to say other than it's kind of a neutral finding, which is he has very good favorability numbers. We do not know how that would translate into votes.

Borg: Speaking about favorability, what are you finding about the favorability? What do people tell you if you can interpret or interpolate about Hillary Clinton's falling numbers?

Selzer: You know, it's fascinating isn't it? I think it is just more as a campaign observer I don't think she has been in control of this campaign at all. So her ability to tell a positive story and to create positive events which get people thinking about the issues that they care about and getting them excited that here's somebody who shares their values and are going to make the things happen that they want to see happen, what I'm describing there is what Bernie Sanders is doing. He's holding big rallies, he's making people feel good about being a democrat. And so she is not spending a lot of time doing these things. I think just as an observer it's not surprising that there is this little bit of a fall off there.

Lynch: Sticking with the democratic race, despite the time he spent here Martin O'Malley doesn't seem to really be catching on with Iowa voters. Do you get a sense of whether that is about him or is it just that there's not room in this race for a second Clinton challenger and Bernie Sanders is assuming that role?

Selzer: I don't know that I have an answer on Martin O'Malley. We know that Jim Webb isn't getting anything. We know that Lincoln Chafee isn't getting anything. And their numbers don't seem to be moving. But what I caution myself and people I talk to is John Kerry's numbers weren't moving, Rick Santorum's numbers weren't moving. So things can and may still change.

Lynch: It's early.

Selzer: It's early. The one other point I'll just make while we're on the democrats about Bernie Sanders is I don't know this to be true but I'm going to be looking harder at the polls. If he is concentrating what shows up in the poll as a pretty high number and approaching Hillary Clinton's numbers, if those are concentrated in select number of communities it looks good in a poll that looks statewide, it's going to look problematic in terms of translating into delegates on caucus night because Johnson County only has a certain number of delegates and you can win --

Henderson: We need only look back to the Bill Bradley/Al Gore race.

Selzer: Exactly. And it's something that Barack Obama knew and the caucuses were held on January 3rd and he had all of those university students go home to caucus. And he had the advantage of the calendar, which thank goodness we don't have that advantage this time around.

Henderson: Speaking of the calendar, we have had on the republican side what even Donald Trump has called the summer of Trump. Obviously angst among republican voters, anger some are describing it. Can you go back to 2010 and 2014 and in hindsight see that this electorate was ripe to be motivated by someone from the outside?

Selzer: Well I think that's exactly what happened in 2010 and 2014. You have the rise of the Tea Party, you had people starting to talk then about taking their country back. Donald Trump has sort of turned that into make America great again, which maybe has, it's maybe a little more politically correct, I don't know. But I think if you look at his popularity and you look at Ben Carson doing well, Carly Fiorina we do not know yet whether her good debate performance will turn into votes, but we'll see in the polls. It is as though they have gone along with the establishment, experienced politician election after election after election and they're not getting the change they have been talking about cycle after cycle after cycle. So is it a risk to have an outsider be in the White House? Yes. Are they ready for that risk because they have done it the other way and nothing has changed. Now that's one way of looking at it. It could be that Trump explodes and Ben Carson implodes and Carly Fiorina really never takes off. Who knows. We don't know just yet. But that's a bit of the mood.

Borg: I'm sorry to interrupt. Are you discerning any difference in this election cycle given that it is more of a national campaign with national debates rather than concentrated in Iowa as it has been in past cycles? What effect is that having on Iowa voter opinion, if anything at all?

Selzer: I don't know that I know the answer to that. I think as we look at Iowa and we look at the national thing we don't really see a huge difference in how people are lining up. Now in New Hampshire John Kasich is doing quite well, better than he is doing here, and he has basically said he is all in, in New Hampshire, and maybe not going to do that much here. We know that Scott Walker was doing very well here and he has now said Iowa is everything to him. We'll see if he can sort of walk his way back up the ladder, he has fallen a bit. But I think there is always a little bit of that provincial, meaning localized, ways in which the campaigns on the ground are doing things. But I'm not an expert in the ground game and maybe things are going on that are just invisible. But the poll numbers don't surprise me nationally versus what is happening in Iowa.

Lynch: Early in this race Jeb Bush was sort of a presumed front-runner and as you mentioned Scott Walker early on had the poll numbers that he was the front-runner, now Donald Trump seems to be the front-runner all over. How many more front-runners do you think we'll see before February 1st? How fluid is this race?

Selzer: I'm not going to predict that. We have seen races where every month we had a new front-runner. But we have also seen races where someone took hold and just sort of went all the way through. So you had George W. Bush in 2000. We've seen it happen so many different ways that I just don't make predictions like that. Who knows.

Lynch: One of the things we're seeing right now is, for example, Bobby Jindal is advertising heavily, Scott Walker is starting to advertise heavily. Do those ads affect what you're finding when you poll? Can you sort of see a correlation between those ads and poll results?

Selzer: Well let me answer it in a slightly different way. I was asked whether my algorithms and my protocols take into account the campaigns and how much advertising they're doing and the yard signs in particular and the bumper strips and all of that. And I said, yes, to the extent that those things are influencing the people I talk to in terms of telling me they're going to go to caucus and telling me who they support, it is factored in. But I don't take, it's not something that I'm paying attention to in terms of trying to have that be a part of our procedures for how we vote. To the extent that it works, that is really all I need to know.

Borg: Another question, Jim? 30 seconds.

Lynch: 30 seconds. Well, let's real quickly, the future of polling. Right now there's Internet polling and robo-polling and stuff. Where do you see this going? What will you be doing in ten years?

Selzer: Worrying probably still. There are so many possibilities for the way this is going to go. I'm going to be as science driven as I can for as long as I can.

Borg: The last time you were on here you said, what keeps me awake at night and you described something else. Now you say worrying. You're really intent on what you do. Thanks for being our guest today.

Selzer: My pleasure.

Borg: Before we close, I'd like to note the passing of an Iowa journalists this past week, Forest City's Ben Carter. Ben's byline wasn't on the news pages of the daily newspapers, but he is a pillar among the state's community journalists. For 37 years, Ben Carter edited and published the Forest City Summit. Four times the Iowa Press Association judged it the state's best weekly newspaper. And the Press Association gave Ben its highest honor, that is the Master Editor-Publisher Award and also elected him its president. Ben also appeared on Iowa Press as one our journalist panelists. He was a World War II Navy bomber pilot and he devoted that same precision to community journalism. His legacy is community involvement and helping build Iowa's excellence in small town newspapers.

Borg: Next week on Iowa Press, Texas Senator and republican presidential candidate Ted Cruz. He is carving a conservative segment of voters and hoping that carries him to the republican nomination. Senator Ted Cruz next Friday night at 7:30 and noon Sunday. I'm Dean Borg. Thanks for joining us today.

More from this show

Iowa Bankers Association
Associated General Contractors of Iowa