ad info

Editions | myCNN | Video | Audio | Headline News Brief | Feedback  





Bush signs order opening 'faith-based' charity office for business

Rescues continue 4 days after devastating India earthquake

DaimlerChrysler employees join rapidly swelling ranks of laid-off U.S. workers

Disney's is a goner


4:30pm ET, 4/16









CNN Websites
Networks image

Burden of Proof

Do Political Polls Unfairly Shape Public Opinion?

Aired October 26, 2000 - 12:39 p.m. ET


ROGER COSSACK, CO-HOST: Who's winning? Who's losing? Will the polls predict the results of the November 7 election? Today on BURDEN OF PROOF, do polls unfairly shape public opinion? Are the numbers gathered fairly? And will the Internet change all the rules of the political polling process?

ANNOUNCER: This is BURDEN OF PROOF, with Greta Van Susteren and Roger Cossack.

COSSACK: Hello and welcome to BURDEN OF PROOF.

Several polls are calling the race for the White House a statistical dead heat. The polling numbers for George W. Bush and Al Gore are so close, they fall in many polls' margins of error.

GRETA VAN SUSTEREN, CO-HOST: But the media's polls aren't without their critics. Some argue that the sampling of surveyed voters is too small to represent the enormous nationwide electorate. Others say the polls can't possibly represent the diverse nature of the American voters by interviewing a select few.

COSSACK: And joining us today are constitutional law scholar Bruce Fein, CNN polling director Keating Holland, and professor of government Stephen Wayne.

VAN SUSTEREN: And in our back row, Allison Salyer (ph), Jess Mary (ph) and Heather Smith (ph).

Keating, first to you. How many people do you have to talk to before we can conclude it is scientific or reliable in our polling?

KEATING HOLLAND, CNN POLLING DIRECTOR: We're doing about 750 interviews over the three days that we release the tracking poll. We're going to be upping that shortly so that we'll eventually get 1,000.

VAN SUSTEREN: Who picked 750? Or how did that -- you know, people seem to talk about 750 or 1,000. I mean, why those numbers?

HOLLAND: It's basically based on the fact that the margin of sampling error is determine entirely by the number of interviews you do. At 750, we have a margin of sampling error of about 3.5. Some people round that up to four. When we get to 1,000, we'll get that down to plus or minus three. VAN SUSTEREN: How do you actually do this? What are the mechanics? Take me through it.

HOLLAND: Well, very simply, a computer will pick telephone numbers at random. It's a process called random digit dialing, spit them out to a bunch of interviewers sitting in a big room -- actually, several big rooms across the country. They'll call up, do the interview, computer crunches the numbers, we get it the next morning.

COSSACK: Stephen, is there any evidence that polls really do influence the way people vote or don't vote or continue to go or do not leave to go to the -- to go and vote because they think the race is over?

STEPHEN WAYNE, PROFESSOR OF GOVERNMENT: I think there's some evidence that polls influence you in the media to say someone's ahead, someone's not ahead, here's how the horse race is. We know that polls influence the donors, particularly at the beginning of the nomination process -- you don't want to spend a lot of money on a losing horse -- but I don't think polls -- I don't think people read the polls and then determine how they're going to vote.

VAN SUSTEREN: I notice you say -- you said influence. I've got to admit -- maybe I'm just the only one -- but I always sort of with -- for the underdog. I mean, like, now I'm rooting for the Mets. I might have started with the Yankees, but now I'm a little bit for the Mets. I mean, I see these polls are so close, I mean, are there other people out there likely that are sort of being -- so it balances in terms of the influence, some will go with the flow, go with the winning poll and others will sort of feel sorry for the underdog?

WAYNE: The studies that we've done, for example, when the polls close in the East and you get the report of the results, how does that affect California and Oregon? It depresses turnout, but it doesn't seem to impact on the vote for the winning candidate or the losing candidate.

COSSACK: I don't understand what that means, it doesn't impact, if it depresses turnout.

WAYNE: It depresses it equally for all the candidates. So the front-runner is not necessarily advantaged. If someone says that Al Gore has won in New York, and if it looks like Al Gore is going to win the election, the California vote will decrease, but not necessarily more for Gore than Bush.

COSSACK: What if someone says, candidate X -- and I don't want to use any of the ones that we're talking about today -- wins in a runaway. The East -- the whole East is going for candidate X, it's all over but the shouting. Now what happens in the West?

WAYNE: People go home. They don't vote.

VAN SUSTEREN: Keating, I read a number of statistics. I don't know if it's a valid one or not, but 1998, 94 percent of American households had telephones, so I presume those are the people that are likely to be polled. What, do we leave out that 6 percent in polling? And are we satisfied with leaving it out because those people may, perhaps, aren't likely, who don't have telephones, may not also vote? Is that why, or is that 6 percent just such a small number that we don't care about it?

HOLLAND: It's a relatively small number. To a certain extent there are some cost implications. But, yes, if you don't have a telephone it probably means that you're a student in a dorm, you're perhaps a homeless person, you're in prison, you're in a convent or a group housing situation of some sort. Those people don't strike us as likely voters.

COSSACK: Or you're Greta Van Susteren and you don't want to be bothered.


All right, Bruce, let me ask you this: Should we put some kind of -- and I don't know if it's possible given our First Amendment -- but should we be thinking about putting, trying to put some kind of regulation, government or otherwise, on the dissemination of polls, or at least some kind of standards to make sure that these polls really do reflect what they're supposed to?

BRUCE FEIN, CONSTITUTIONAL LAW SCHOLAR: Well, I think the disclosure standards might be permissible. I think the Supreme Court decisions are pretty clear that the government can't seek to regulate speech or dissemination of truthful information because they think mature adults will ask unwisely, either by not voting or changing their vote.

But with regard to disclosure, I think it would be proper, for instance, to require a poll to indicate how many times in the last years they forecast elections correctly or incorrectly, you know, and be more elaborate in explaining the statistical errors so that a scrupulous reader would understand how hazardous it would be to extrapolate from a poll a likely winner, just like Tom Dewey...

VAN SUSTEREN: That's like seeking the FTC, though, on Keating or something, by doing that, but saying, how good are you at your business? But...

COSSACK: What would be wrong with that? I mean, what would be wrong with a full disclosure on polling?

VAN SUSTEREN: Well, because I think the American...

WAYNE: First of all, it might be too technical. Secondly, you don't need it. We're getting hundreds of polls every day. And if one poll is way off, people are going to say, hey, that's crazy, everybody else is showing something else. So it's a kind of self-correcting mechanism.

What you do need is you need to know the size of the sample, you need to know that there's a random selection, and you need to see the questions. How the questions are worded affects the answer. For example, one of the issues today is, who are the likely voters? That's what they're polling. Now, there are different ways to draw the screen on likely voters. The tighter you draw it, the better the Republican candidate will do.

There are -- many of the polls simply ask today, if you're voting, who would you vote for, X, Y or Z? If you ask, if the election were held today, would you vote for X, Y and Z, or haven't you made up your mind yet? you're going to get a different response.

VAN SUSTEREN: Keating, I can see the questions like, who are you going to vote for, Gore, Bush or Nader? would probably get a pretty good answer on that. But I remember hearing Bill Schneider, who's a CNN political analyst here at CNN, talk about polling, that when you call up people that they think there's a right answer. For instance, are you likely to vote? I mean, who wants to say, no, I don't care, it's un-American, I don't feel like voting. Most people will say, yes, I'll vote. How do you sort out that from?

HOLLAND: That's one of the reasons why we ask multiple questions to try to find likely voters and not just rely on a single question. Some pollsters do. I think they're wrong in doing so. We ask a battery of questions to try to really boil it down. Turnout is about 50 percent. We try to throw out about half the people that we talk to in order to make turnout look -- our likely voter model look like turnout models.

COSSACK: Which half? Which half do you throw out?

HOLLAND: The half that admit that they haven't voted in previous elections, the half that say that they are unlikely to vote. You do get some people who are perfectly willing to admit that. And there are various other questions you can ask. For example, Gallup asks, do you know where people who live in your area go to vote? If you don't know where your polling place is, you're probably not going to vote.

VAN SUSTEREN: How do you know if people are answering honestly or if they think they're answering a quiz, like, and they want to make sure they answer the right question so they may not give one that's genuinely true to them?

HOLLAND: That's why we ask multiple questions, so that, eventually, we can boil it down.

VAN SUSTEREN: You can sort through it and figure it out.

HOLLAND: We can drill down until we find a voter's true feeling on one question.

FEIN: Now, Greta, I think your questions illustrate exactly why a compulsory disclosure of this kind of information that is not in the ordinary story you read out of the "New York Times" or "Washington Post" could be very useful for the reader.

VAN SUSTEREN: Well, it might be useful, but I don't think you can force these people. I mean, that's -- what's more, the government could end up... (CROSSTALK)

FEIN: Why can't you? You can enforce disclosure all the time in...

HOLLAND: There is an enforcement disclosure which CNN and the other major networks belong to, the National Council of Public Polls. We agree that we're going to release most if not all of the information that both you have been talking about. Most of that information winds up on our Internet or on Gallup's Internet site because, as you can tell right here, it would take 20 minutes to give that information out on television. But if you look in "USA Today," you'll get most of that information in the small print.

VAN SUSTEREN: We're going to take a quick break.

Up next, many countries have restrictions on when polls can be gathered and when can they be published. No such restrictions exist here in the United States. Should that change? Stay with us.


According to a new government report, crime rates in the nation's schools have dropped during the last decade. The findings, based on surveys of students, teachers and school administrators, showed that the number of victims in schools dropped from 48 to 43 crimes per 1,000 students between 1992 and 1998.



VAN SUSTEREN: Many of the national polls survey likely voters via the telephone. Internet polls gather information naturally from the interactive community.

Keating, let me talk about the Internet polls. If you do a poll on the Internet, you are likely to get the upper income people, the more highly educated, you may get someone who votes more than once in the poll, and worst yet, you might get my eight-year-old niece who can't even vote.

Why in the world should we give any credit -- credibility to the Internet polls.

HOLLAND: We don't, CNN doesn't. When we put them on our Web site we make a point of not calling them a poll, number one. And number two, putting a little note that it is not scientific.

VAN SUSTEREN: But it's even worse by saying it is not scientific. I mean, it could also be absurd. I mean, if you've got a bunch of eight-year-olds sitting with their computers and voting I wouldn't be so polite as to say it is not scientific, I would it's ridiculous, on the issue for instance, of voting.

HOLLAND: You'll get no argument from me on that one. I like to do polls scientifically. It bugs me sometimes when I see them done other ways.

COSSACK: Bruce, my issue with this, and we were talking a little bit at the break, you said, I'm afraid sometimes these polls take on such magnificent lives of their own that they become biblical. And I think, somehow, that when you have that possibility of that kind of impact on our elections, that something has to be done, perhaps to at least limit the way they are disseminated.

Now, I know that there's a voluntary way of -- organization that says we're not going to disseminate it until later. But, isn't there something else that could be done?

FEIN: Well, it's very difficult given the Supreme Court's decision to suggest that because you think people will react to polls in various case, the government has a paternalistic role and say all right we have to withhold this information from you because we think you will act unwisely. You will go with the crowd, you won't vote at all. However, I do think the experience abroad, I think countries like France and Mexico, in fact, prohibit campaigning or polling at all in the last two or three days before election don't seem to have warped the result.

VAN SUSTEREN: That would actually be, I think, constitutional in a sense, that when you talk about releasing information, you're really talking about a First Amendment right. But the first amendment isn't an absolute, and putting something that might be reasonable restriction, for instance suppose that a poll really did impact, suppose release of information may impact an election three hours behind us here in the United States. I mean, putting a reasonable restriction may not be a First Amendment violation.

FEIN: Well, let me tell you, Greta, why I think that's dubious. The Supreme Court's doctrine is if the reason for the government action is they think mature adults will act unwisely on information that's truthful, that's not a good -- that's not a constitutional reason to do that. There have been ideas of requiring delay in polling but they've never taken root, in part because the doctrine is so clear.

COSSACK: You are talking, Steve, talking about a real short delay here, possibly. Maybe we're talking about election day.

WAYNE: Don't change things unless you know one thing affects the other, OK. There's no proof that getting the results three hours early from the East Coast is going to do anything but depress West Coast voting.

VAN SUSTEREN: But isn't depressing the vote, I mean, isn't that important enough? I mean, even if you say it's depressed evenly, both Republican or Democrat or Nader, whatever the case may be, I mean, isn't that -- is that enough of a reason?

WAYNE: No, I don't think so, because I think part of the voting decision is whether to vote or not, and whether to vote or not relates in part to whether your vote's going to make a difference. Fortunately, we have a lot of elections on the same day. We have initiatives on the West Coast. So, the likelihood is that it is not going to have a real major effect on the outcome.

COSSACK: Now, it is your position that it depresses both sides equally, but if we -- go back to what I brought up earlier -- but if candidate A is described as everything a landslide victory, doesn't that mean that candidate B's voters and supporters won't show up?

VAN SUSTEREN: Yes, if you've got a choice, go to the movie or go vote and your candidate already lost, you go to the movie.

WAYNE: If your candidate already won you would still go to the movie.

FEIN: No, I think Roger's point is this one, Steve. It's a little bit different. That if, suppose you have a coattail effect that you're counting on. People vote by party. They have slates, so to speak. Gore is reported to have a landslide victory but it's thought that if people in California turn out in large numbers, Gore will carry along Democratic candidate for Senate or House of Representatives. If they decided not to show up because Gore is going to win, that means the subsidiary votes that would go to Democrats aren't going to be there because they don't...

COSSACK: One lever, one vote then.

VAN SUSTEREN: You all have the last word. You are all talking together. But that's all of the time that we have for today. Thanks to our guests and thank you for watching.

Today on CNN's "TALKBACK LIVE," the clock ticks down to election day, should vice president Al Gore include Bill Clinton in his campaign or avoid him? Send your e-mail and tune in at 3:00 p.m. Eastern time.

COSSACK: Maybe we should take a poll on that.

And join us again tomorrow for another edition of BURDEN OF PROOF. We'll see you then.



Back to the top  © 2001 Cable News Network. All Rights Reserved.
Terms under which this service is provided to you.
Read our privacy guidelines.