NEAL CONAN, host:
This is Talk of the Nation. I am Neal Conan in Washington. With the election just a couple of hours away now, national polls agree that Barack Obama leads John McCain but they don't agree by how much. A Fox News polls shows the Democrat up three points. The New York Times-CBS survey says 11 and even on the radio, I can see a lot you rolling your eyes. Remember New Hampshire, you say, when all the polls showed Senator Obama leading big over Senator Hillary Clinton until he lost. Pollsters don't survey people who only use cell phones, you say and a lot of people won't talk to them at all. What about the Bradley effect and how can the McCain camp argue that it's much closer in places like Pennsylvania on the basis of their internal numbers. Today we want to hear from those of you who have been polled. Did you tell the truth? Did you turn them down? Tell us your story.
Our phone number is 800-989-8255. Email us, email@example.com. You can also join the conversation on our blog at npr.org/blogofthenation. Later in the program, your chance to campaign for the best movie president. Murray Horwitz joins us and you can cast an early ballot now by email, go to firstname.lastname@example.org. But first, can we trust the polls? Scott Keeter is director of Survey Research at the Pew Research Center for the People and the Press, with us here today in Studio 3A today. Nice to have you on the program.
Mr. SCOTT KEETER (Director, Survey Research): Thank you, Neal, nice to be with you.
CONAN: And all of those polls show Barack Obama in the lead, so, this thing over?
Mr. KEETER: It sure looks like it's over but as you said in your opening remarks, let's not forget New Hampshire and that was a very useful corrective to hubris for pollsters. All of the polls there said Barack Obama was going to win the New Hampshire primary and of course, he didn't. So we have to be careful about saying what we see now is what's going to happen. But this is a very consistent picture of a lead for Obama. There is a lot inside the polls that suggest that he's going to do very well. He has more enthusiasm among his backers than McCain does. And I think the question maybe is just how big the margin is going to be but as I say, we're - pollsters are a superstitious bunch and we never say, never.
CONAN: Is this a science or is this an art?
Mr. KEETER: There is a bit of both. The best pollsters I think come at it scientifically and when judgments have to be applied are able to draw on the experience of many years of practicing the craft. There are judgments that you have to make about how tight to make a likely voter screen or what the assumption is going to be about how many people are going to vote. And there some judgment has to come into it but this is largely a science.
CONAN: And so you feel pretty confident this at this point.
Mr. KEETER: If we were seeing polls on both sides of the zero point, if there were polls showing McCain leading right now nationally, if there weren't such a consistent pattern of Obama leads in the states, where lots and lots of polling is going on, then I would say we need to be more cautious but right now the candidates themselves obviously believe this because you don't see Barack Obama in Democratic states trying to defend them, and you see McCain in places that he didn't think he was ever going to have to go to, barnstorming through them today.
CONAN: Yet, you also see McCain working very heavily in the state of Pennsylvania where, according to all the independent opinion polls, the ones you see in the newspapers, well, he's up, Barack Obama is up 8-11 in around there somewhere and the McCain people said, no, no, no our internal numbers show that it's closer. Are they fibbing or is this possible?
Mr. KEETER: No, there are some public polls that are showing it a little closer and I think that the McCain campaign may believe that if there is a chance, if they think that the polling is not fully picking up on some skepticism about Obama's experience and so forth, that it would be in place like Pennsylvania that he struggled in, in the primaries against Hillary Clinton and so there thinking that if they have a chance to steal a Democratic state away from Obama that that would be the place they could do it, and that would be a big prize. So I think that's the logic, he needs not just to hold everything that Bush held in 2004. I mean, if he did that he would win.
CONAN: If would be president, yes.
Mr. KEETER: But you know, he has enough - there are enough states where is trailing by significant margins that Bush won that he thinks he needs to steal a Democratic state and Pennsylvania is the place he thinks he has the best chance to do that.
CONAN: We want to hear from those of you today who have participated in opinion polls. Give us a call. They us what it was like. How did you respond to it? 800-989-8255. Email us, email@example.com. Let's begin with Darren. And Darren calling us from American Falls in Idaho.
DARREN (Caller): Yes, about two weeks ago, I had a poll. It was very good. It took about 20 minutes. I was honest but I was surprised being from Idaho. It must have been a national poll. I am a political scientist and so I was very interested in it and it was a very well conducted poll.
CONAN: You're a political scientist. So you felt no compunction about answering the questions?
DARREN: Oh, yeah. And so I was listening for push questions or and I actually asked who had paid for the poll and they said it was one of the two parties but they wouldn't reveal which one. And it was a very, very thorough poll, it took about 20 minutes.
CONAN: And what were some of the more interesting questions?
DARREN: Some that I felt were more interesting is they ask about previous voting, like in other presidential elections, how many elections I have voted in and favorable or unfavorable with candidates like, you know, like in a scale of one to five, if I say that the candidate George W. Bush or John Kerry or Ronald Reagan and that so. I thought it was a very good poll.
CONAN: So let me ask you, Scott, if the question about have you voted in the past. They are trying to determine whether he is a likely voter, isn't that the category?
Mr. KEETER: That's right. That's a standard question in the so-called likely voter scales that most pollsters construct. Because we know that not everybody who says they're going to vote really is and so, over the years, pollsters have developed a methodology for trying to identify them and one of the questions is, have you voted in the past? How often do you vote? Do you know where to vote in your neighborhood? And one of the controversies right now in the polling business and one of the questions about the polls this year has to do with whether past behavior is a good predictor of the future.
CONAN: Predictor of the future.
Mr. KEETER: Because it may be that this is an election that has energized a lot of (unintelligible) never voted people.
CONAN: And is Darren right that they would also not tell him which of the parties had paid for the poll?
Mr. KEETER: Yes, normally, they certainly wouldn't tell you that anywhere upfront. Normally, you know, people might tell you at the end. At a minimum, they might tell who's doing the poll that is, what's field house or organization that's doing the survey. And usually, if you call them back, you can find out but most people don't go to that much trouble.
CONAN: And one more question, we'll get back to you, Darren in just a second, stay with us. But one more question, when he talked about push polling. That's when they might ask if somebody answers the phone like Darren, how would you respond if you learned that John McCain was in fact a card-carrying member of the communist party?
Mr. KEETER: Right, and there is a legitimate polling technique that is - under the rubric of message testing that involves offering people information about the candidates and saying, how would you feel if you knew that - fill in the blank. That's done very routinely by campaign, not so much by public pollsters but by campaigns to test the effectiveness of different appeals. As long as what they're saying is not outright falsehoods or very misleading then that is a legitimate technique to use to test your messages. Push polling is a different thing, it's not a poll at all. It's campaigning under the guise of a poll using an invitation to participate in a poll to get people to cooperate and then slamming the respondent with lots of negative information about the candidate that, you know, is not the one paying for the poll.
CONAN: And Darren, that didn't seem that way to you. It seemed pretty straight.
DARREN: Oh no, it's very straight. And in fact, I was very impressed by the amount of demographic information, income, education, race, religion - and I was sure at the (unintelligible) when I said that I was, considered myself very liberal and from Idaho and a Mormon. I think I'm the only one who ever responded that way to one of the polls.
CONAN: And that makes him the outlier. OK. Darren thanks. Are you going to vote tomorrow?
DARREN: I actually voted early last week.
CONAN: Congratulations. Thanks very much.
Mr. KEETER: Thank you for taking part in that poll too, Darren.
CONAN: Let's see if we can get another caller on the line. This is Emma, and Emma's with us from Fort Lauderdale, in Florida.
EMMA (Caller): Hi.
CONAN: Hi Emma.
EMMA: I was - I've been called so many times I can't -I've lost count of how many times I have been robo-called. And was, called yesterday again by someone who wanted to know, if I was voting and who I voted for, and then actually had someone come to my door, and knocked on my door yesterday. He's from the Obama campaign to see if I was going to vote or had voted already. And I found it annoying.
CONAN: I can understand.
EMMA: Yeah, it's been like - they're like hounding me. I feel like. So I kind of don't appreciate it. I guess I've been hesitant to participate.
CONAN: Now Emma, hold with this just a second and let me get back to Scott Keeter. Those don't sound like polling calls, the robo-calls. Those are automated calls to say, hi I'm, you know, so and so and I'm urging you to vote for or against this person.
Mr. KEETER: Right. That's definitely what it sounds like now. They may be saying,I'm conducting a poll as a way to get you to cooperate. If so, that's really not a good practice. You maybe targeted because of your age or your cellphone-only status or something like that, as a potential Obama voter and it may be that that's why you're getting these calls. If you are cellphone-only - and maybe I should ask him, are you only - do you only have a cellphone?
EMMA: I do only have a cellphone. And as I said, I live in Florida, which I know is a very hotly-contested state. And I'm also registered independent.
Mr. KEETER: Well, you would be one of those voters that the campaigns would be after. But the chances that you would be called by a survey organization, given that you're cellphone-only, are pretty low. It really is only in this election and may be a little bit in 2006 that pollsters were including cellphones in their samples. And so, it's still is an uncommon practice and almost unheard of it at statewide level.
EMMA: I didn't know that.
Mr. KEETER: Yeah.
CONAN: And the knock on the door was, somehow, they had targeted you as a potential Barack Obama voter, and we were trying to urge you to go to the polls. But Emma, there's early voting in Florida, in fact I was down there over the weekend. I saw incredibly long lines.
EMMA: Yes. I voted. I forgot what day I voted but I voted this week, and it took me three hours.
Mr. KEETER: Wow.
EMMA: I brought my knitting.
CONAN: Pardon me?
EMMA: I brought my knitting with me.
CONAN: You brought your knitting with you. Well, congratulations. And thanks very much for the call.
EMMA: Thank you.
CONAN: Bye-bye. We're talking with Scott Keeter from Pew Research Center for the People and the Press about whether or not we can trust the opinion polls about the methodology that goes into these polls and what is and what isn't legitimate. We're taking your calls at 800-989-9255. Email us firstname.lastname@example.org. Have you ever been op-polled? Tell us what it was like, what your experience was like. I'm Neal Conan, stay with is. It's the Talk of the Nation from NPR News.
(Soundbite of music)
CONAN: This is Talk of the Nation, I'm Neal Conan in Washington. A little bit later this hour, our movie pal Murray Horwitz joins us, we'll talk politics sort of - who's your favorite movie president? You can vote now by email email@example.com. Right now, we're passing the opinion polls, why should we believe them. What do the numbers really tell us? Scott Keeter is our guest. He's director of Survey Research at the Pew Research Center for the People and the Press. We want to hear from those of you who have been called. Did you tell the truth? Did you turn them down? Tell us your story, 800-989-8255, email again firstname.lastname@example.org.
And we were talking earlier with Emma in Florida, cellphone-only, and one of the real questions that people have about this year is whether opinion pollsters are getting an actual sample. Are the people who are not called, those who only have cellphones, are they going to be the same as the people who have landlines?
Mr. KEETER: It's amazing. We got this question a lot even four years ago when the number of cell-only voters was much smaller. The exit polls indicated that seven percent of voters four years ago were cell-only. We think that number is going to be double that this year - 14, maybe 15 percent of voters. Most major national pollsters are now including cellphones in their samples, and our work since the middle of the summer has included cellphones in every sample, and we are seeing that it's making a difference. That is, when we blend in the cellphone respondents, we are finding the number for Obama being at least a point or two higher, the number for McCain being possibly a point or two lower. So it's making a significant difference on the margin when you look at all of the polls collectively.
CONAN: And how are you doing that? Calling people on cellphones or you ask people, do you have only a cellphone and the quizzing them?
Mr. KEETER: Exactly. We're calling people on their cellphones, we're offering them a reimbursement for the minutes that we might be using up on their plans. And then we conduct the interview like normal, and we're getting the same response right on cellphones that we do on landlines.
CONAN: Let's get Luke on the line. Luke is calling us from Tucson, Arizona.
LUKE (Caller): Hi there, Neal. I was contacted by the Gallup daily tracking polls the night before last. And I was - about 10 minutes into the poll, I realized that I was running out of time and I tried to hurry to vote (unintelligible), but I ended up having to get off the phone because I just didn't have enough time to complete the poll. And I was wondering if anybody has studied this problem as a form of bias, and I'll take my comments off the air.
CONAN: It sounds like you've got a young voter there you need to take care.
CONAN: All right, bye-bye.
Mr. KEETER: Well, yes. We have studied that. I think that it's almost certainly the case that his responses to the questions that he was asked would not have been included in their poll. They would only include him if they could make it to the demographic section where they could properly wait(ph) him into the sample. We find that the people who break off during the interview tend to be less interested in politics, less engaged, they tend to be a little less educated. And that does create a potential for a small bias, but the number that we lose that way is very small - typically fewer than one in 10 interviews that we start end up having to be terminated.
And you know, he sounded like somebody who would've been happy to finish if he just had and hadn't something else going on. A lot of pollsters, including Gallup, will probably try to re-contact him and maybe that given the lateness of the call that he got that he wouldn't get re-contacted. But you know, we do make an effort. So by in large, this bias is very small. But it is - there is a potential there.
CONAN: He also used a term of (unintelligible) that a lot of people don't understand. Tracking poll - what's the difference between a regular opinion poll and a tracking poll?
Mr. KEETER: Yes. The tracking polls involve what are called rolling samples. The pollster will open up, which is to say start calling a new fresh piece of their sample, maybe a couple of thousand numbers on a given day. They'll interview people out of it. The next day, they'll open up another sample of equivalent size but they'll keep calling the people they've called the night before, and maybe they'll do that for three or four nights. They'll average the numbers together and then as they go forward, they'll drop out the sample from the day that they started with and they add the new day in. And so several of the polls, including ABC, Washington Post and the Gallup Poll that he mentioned, are doing that.
CONAN: Let's get June on the line, June with us from Nashville, in Tennessee.
JUNE (Caller): Yes. Hi.
CONAN: Hi June.
JUNE: Hi. Well, I have to admit it that I lied on the polls.
CONAN: What did you say?
JUNE: Well, it was automated. I would never have done this to a real person. It was during the primaries, and the first question was - Are you Republican or Democrat? And I'm a Democrat, and I don't know why. Oppositionality, I guess. I pressed Republican. And then I answered honestly from that point on. And I went through the whole list of all the primary candidates, and I answered that part honestly. And then after that, I felt really bad. I've kind of didn't know, you know, what kind of echoing effects this might have been in the universe. And then I started getting calls from - I got a call from Minnesota against Al Franken. Wanted me to contribute to...
CONAN: Norm Coleman.
JUNE: Yeah, Norm Coleman. And I was honest with that guy and I said, I think I just got the wrong list, you don't want to call me, take me off. And there were couple of those and then it kind of went away. But it made me feel really guilty. I would never do that. I think the part of the issue is I would never do that to a real person.
CONAN: We're glad you stepped into the radio confessional.
JUNE: Yeah, I feel better already.
CONAN: Good. And as for your penance in a few minutes but is this is an issue, Scott?
Mr. KEETER: Yes. Well, I think she already paid her penance by ending up on.
JUNE: Thank you. Thank you.
Mr. KEETER: On a partisan list, but certainly the confession takes care of. We don't think it's really a very serious problem. I mean I think people sometimes do and sometimes people just mistakenly answer. But if it were a really serious problem, polling would not have been as accurate as it has been in elections over the years. Four years ago in a very tough race between Kerry and Bush, national polling was very accurate, the average of the major national polls came within about a half a point of the actual margin.
There certainly are questions that people find uncomfortable to answer such as whether you know, whether they have cheated on their taxes, how frequently they go to religious services and the like. And there maybe some exaggeration to make a person look better, make them look less bad. But we don't think that in politics, there's any particular stigma into saying that you're voting for one person over another. And so we generally don't worry about lying.
CONAN: June, thanks very much and - go away and sin no more.
JUNE: I will sin no more, and I've obsessively watched the polls now I tell you. Thank you.
CONAN: Thanks very much. And here's an email we have from Karen in Elkhart, Indiana. I'm a black American and I find it strange that polls come out giving the black view but very few people of color are ever polled. This is a discussion I have had doing my own poll of people of color that polls come out, but we can't recall ever in the last four elections receiving a call. So it's hard for me to respect the polls.
Mr. KEETER: This is something that we hear from a lot of people and not just African-Americans. A lot of times we get comments from people in our website or over the phone that just say, I've never been polled, how can we believe the polls? The fact is, there has been an enormous amount of polling this year. But even at that, the chances that any given American voter would have been called are very small. There are hundreds of millions of Americans, 200 or something million adults. And the collective total of all the polling is going to fall probably well below 50 million - if anywhere near that, it's hard to estimate.
So there's a very good chance that as an individual, you wouldn't be polled. All of the polls that use random sampling give every person an equal chance, and so we know that when we get our sample back, we have reached roughly the proper proportion of African-Americans that there are in the population. If we fall short, we correct that with statistical waiting, giving the African-Americans we interview a little more of a wait in the poll. But generally speaking, we're pretty close on our demographics.
CONAN: Does that increase your margin of error when you have to wait it a little bit?
Mr. KEETER: It does, waiting - depending on how severe it is, can make your margin of error larger than it would be if you had a perfectly random sample.
CONAN: Let's get another caller on the line. And let's go to Jim, and Jim is calling us from Greenwich, Connecticut.
JIM (Caller): Hi, how are you.
CONAN: I'm well, thanks.
JIM: Interested in politics and polling. I'm guess whenever I have been called I'm just reluctant to answer pollsters because over the phone, I honestly don't know if somebody is who they say they are whether its a push poll or, you know, just legitimate big-name poller. So I just kind of answer a few questions and then often just say, well, thanks. I'm going to stop some answering. So my question is, is there a significant bias based on either where you live or educational or home or ethnic background to where people just drop out? Thanks. I'll listen on the air.
CONAN: OK. Thanks for the call Jim. And Scott Keeter, I've heard that since the no-call list also came out. Of course pollsters are allowed to call, but nevertheless more and more people feel empowered to say, you know what? I don't want to talk to you.
Mr. KEETER: We are experiencing the same thing that all pollsters are experiencing which is a decline in response rights. Some of it comes from people who are worried about the loss of privacy, they wonder if they're being scammed in some way. But it's also the case of people just to have very busy lives and they have a lot going on and they are less likely to talk to us simply because of that. The good thing is that we have not found in some very extensive research that none response is creating a serious bias in our polls, that is, the kinds of people that aren't talking with us do not appear to be significantly different from the people who do. It's as if none responses just random - people are busy one day they won't talk to us, but the next day they're at home and they don't have as much going on and they do the interview. And that seems to be helping us to continue to get good samples.
CONAN: Here's an email from Al in Milwaukee. Television networks stop predicting results based on Election Day exit polls before all the polls close because the fear that people hearing a candidate had already won would not vote. Shouldn't the same apply to pre-election polls? Would a poll saying one candidate is strongly in the lead and do some people think they don't have to vote or that it's futile to vote. Does your guest have a solution for this?
Mr. KEETER: We think the polling is just one more source of information to voters to help them to assess what's going on in a campaign. There wouldn't be any particular reason to throw polls overboard rather than measures of fund raising and other indications of candidates' strength. We think that polling actually is very helpful. I mean, the fact that you know that one candidate is ahead or not may not be particularly useful unless you are somebody who contributes to candidates very early in the primary season, and you don't want to give your money to somebody who has no chance of winning. But most polling...
CONAN: Like John McCain a year ago.
Mr. KEETER: Like John McCain a year ago, and quite a story of his ability to overcome that. But polling does a lot more than just tell you who's ahead. We try to explain what kinds of people are supporting each candidate, why they're supporting each candidate, what's the dynamic going on? And that it seems to me is a service to the voters. It helps people to understand whether the claims that a candidate is making about why they're ahead in terms of support for some of their policy proposals are really borne out by what people are telling us. It seems that's very valuable information for the public to have.
CONAN: We're talking with Scott Keeter of the Pew Research Center for People and the Press, he's with us here5 in studio 3A. If you'd like to join us, 800-989-8255, email us email@example.com. And you're listening to Talk of the Nation from NPR News. And one more term of art before we get back to the phones and that is exit polls. How are exit polls different from opinion polls?
Mr. KEETER: Well, an exit poll is an opinion poll. It's just that it's conducted in person at a polling place. They're actually are exit polls that are done not just for elections, but to judge how people reacted to the movie as they leave it or even to intercept people as they come out of a library to find out how they felt about it. But exit polls in the political context or interviews with voters as they leave the polling place and those are useful to us also in understanding why people voted the way they did, what kinds of people supported each of the candidates and ultimately that helps to service a basis of an explanation for the election.
CONAN: Let's see if we can now go to Miller, and Miller is on the line with us from - is this Mannheim in Germany?
MILLER (Caller): That's right.
CONAN: Go ahead, please.
MILLER: Yes. Well, I was wondering with polling, whether overseas voters are over taken into account.
CONAN: Americans abroad.
MILLER: And there is, of course, of quite a large number of overseas voters.
CONAN: And I think roughly about half of them actually register and vote absentee in - but they vote in their states where they're originally from. Do you that, Miller?
MILLER: Yes, that's what I did.
Mr. KEETER: The answer is as far as I know most pollsters do not try to capture overseas voters. I don't know exactly how we would do that unless we were sampling off of registered voter lists rather than using random digit dialing which is the methodology that most of the national pollster use. With the registered voter list I could get you in a sample and then if I could figure out how to get to you either by email or by a phone call I might include you. But polls don't include overseas voters.
CONAN: Do they include the military?
Mr. KEETER: The military is eligible. I mean, we don't ask people what they do for a living or if they are active duty when we call them, so if they're stateside and one of the 48.
CONAN: Ah, but if they are overseas they're in the same this category as voters.
Mr. KEETER: Yes, yes exactly.
CONAN: OK. By the way, Miller, you may want to join us on Thursday we're doing a show called Talk of the World. That day in this time slot Talk of the Nation becomes Talk of the World we're going to get calls from people all around the globe we hope to react to the American election and tell - have them tell us what they think Americans need to know about them and their world now so.
MILLER: Well, I'll be glad to do so.
CONAN: All right. Talk to you then. Bye-bye.
MILLER: OK. Bye.
CONAN: And let's see if we get one last caller in. And let's go to - this is Russell and Russell is with us from Leavenworth County in Kansas.
RUSSELL (Caller): Hello.
RUSSELL: I've been polled twice once by a private number, private name, which I don't know but it is still pushes polling and after that I got a robo-call to just seemed that voting for Obama will be tantamount to pulling my throat which I - you know, took it with a grain of salt because - well, Obama is my man. But with the other poll I was a legitimate poll I suspect they wanted a lot of information about who I was, what I did, why I shall registering - to vote with the good idea and so forth and so on, and then asked about my candidate choices. I answered honestly in all cases.
CONAN: And so you thought that was a legit poll?
RUSSELL: I did. The first one, they - it was strange. I think the questions were tailored based on my previous answers. You know, there was a delay while they thought up the next question or picked it out of a linked list. And it felt very sad and overly I can't really put my finger on it but since then the robo calls started I had my suspicions.
CONAN: All right, Scott Keeter, are there decision trees in some of these surveys that if they answered this then you go this way of the answers.
Mr. KEETER: Oh yes. Yes. I mean, and certainly there are questions tailored to different kinds of people. McCain's supporters might get one kind of question and Obama's supporters might get a different kind of question.
RUSSELL: Yeah. that's what I felt. You know, they had to ask me how I felt about - I'm a Palin and I said I'd sooner poke myself in the eye with a dull, salted pencil and vote for a ticket with her on it.
CONAN: Well, if you get salted pencils in the mail you may have asked yourself what a funny old list there.
RUSSELL: There you go. Thanks very much.
CONAN: Thanks very much for the call, Russell. And just for listeners information I think the last of few survey came out that today show that Senator Obama with a six point lead.
Mr. KEETER: Yes. That's right.
CONAN: So we'll see how that comes out tomorrow night. Scott Keeter, thanks very much for being with us today.
Mr. KEETER: My pleasure.
CONAN: Scott Keeter is the director of Survey Research at the Pew Research Center for People and the Press joins us here in studio 3A. Coming up, a special presidential edition of our movie festival from Raymond Massey's "Honest Day" to Anthony Hopkins "Tricky Dick."
(Soundbite of movie "Tricky Dick")
Mr. ANTHONY HOPKINS: (As Richard Nixon) My friends call me Mr. President.
CONAN: And the rest of our fictional commanders in chief that defended this country, Murray Horwitz has them all. Who is your favorite? Give us a call, 800-989-8255 or drop us an email, that's firstname.lastname@example.org. Stay with us I'm Neal Conan, it's the Talk of the Nation from NPR News.