Skip to content
Who's in the Video
Andrew Kohut is the president of the Pew Research Center. He also acts as director of the Pew Research Center for the People & the Press (formerly the Times Mirror[…]

Kohut talks about all things poll-related.

Question: What do you do?

Andrew Kohut: I’m a public opinion pollster. I’ve been an opinion pollster all of my life, or a survey researcher all of my life – not always in the area of public opinion since I did do a lot of market research in my early years. And I now run the Pew Research Center, which is a . . . I think it’s not an under . . . or an overstatement to say that it’s a unique organization. We study public opinion about politics, about media, about international issues. We also have a group that studies the Internet. We have an effort on religion in public life. We have a social trends unit, and probably I’ve left some . . . We have a news organization that studies policy developments at the state level. So we do a . . . We have a broad effort that examines the most important public policy trends, political trends, and media trends. And we do it as a public service. We do it to provide information to the people, to the political leaders, to the media about . . . about American society and global public opinion and those sorts of things. And I’m delighted to be heading this. And it’s been quite successful, and I’m very proud of it.

Question: Why is polling important?

Andrew Kohut: I think it’s important because it gives a voice to the people. It gives a . . . a quantitative, independent assessment of what public . . . of what the public feels as opposed to what experts or pundits think the public feels. So often it provides a quick corrective on what’s thought to be the conventional wisdom about public opinion. There are any number of examples that I could give you about how wrong the experts are here in Washington, in New York and elsewhere about public opinion that are revealed . . . that are revealed by public opinion polls.

Question: What projects are you working on right now?

Andrew Kohut: Well I think some of the most important ones are . . . are in the realm of the Pew Global Attitudes project. We’ve been principle chronicles of the rise of anti-Americanism all around the world since 2002. We’ve conducted 165,000 interviews, and we’ve . . . in 60 nations. And we’ve observed the ways in which American policies and the American approach has alienated many people around the world and brought the esteem of the United States down in . . . in many places.

Question: How do you conduct a poll?

Andrew Kohut: Well global opinion is a little more complicated than the typical public opinion poll. Let’s do a typical public opinion poll. I think that would be easier. A nationwide survey is conducted in the following manner. There’s a sample of . . . of working banks of telephone numbers are drawn. Numbers are . . . Telephone numbers are randomly generated within those banks, and a group of interviewers working on . . . on . . . on . . . on computers with . . . with questionnaires on their screens begin to call these numbers that have been randomly generated from all over the country, soliciting the views of adult men and adult women . . . if that’s the . . . all adult men and all adult women. And they have a prescribed questionnaire. In our case it generally runs about 20 minutes, sometimes less. You can’t keep people on the phone for much more than that unless you make an appointment with them about some particular thing that is close to their lives. And we write the questionnaire. It’s based upon many long-term measures that we have about . . . about public opinion. Do you approve or disapprove of the way President Bush is doing his job? Or congressional or political preference questions. But it also includes a mix of questions about current . . . about current policy. Did you . . . Have you followed what General Petraeus has said about developments in Iraq since the surge? Do you generally agree . . . approve or disapprove of his recommendations? Or in many specific questions, staying on the Iraq example, we’ve been trending since 2003 how well is the war effort going? Did we do the right thing in going to war? Should we keep out troops there ‘til the situation is stable, or should we get out? And we have long-term trend lines on these things as well as what people know about what’s going on in Iraq. And we’re able to assess the ways in which public opinion about Iraq or other issues have evolved.

Question: Is polling the best way of gauging public opinion?

Andrew Kohut: I think it’s the best way of gauging public opinion – doing something that’s independent, that’s quantitative, that doesn’t give just the loud voices’ say in . . . about how things are going; or doesn’t give expert . . . so called experts . . . the notion that they know what public opinion is. I think those are . . . those . . . that . . . that’s what makes public opinion polling pretty important. Quantitative . . . Qualitative assessments of public opinion; going out and talking to people and understanding the nuance to what’s behind the numbers. I think it’s awfully important as well. And the other thing is . . . Another measure of public opinion is behavior – how many people are doing this and how many are doing that. You can see those reflections of public opinion in official statistics, and market trends and all kinds of things.

Question: How does polling fit into the American political system?

Andrew Kohut: Oh in lots of ways. First of all, public opinion polling is used . . . not the kind of polling that I do, but the kind of polling that the political consultants do has been used now for two decades to help candidates get elected. It helps shape messages. It helps . . . Most importantly it helps monitor the great expenditures that political candidates make these days in media. You can’t go out and spend millions of dollars with no sense of whether or not it’s working until you wake up in the . . . wake up on Election Day to find out. So with the advent of telephone polling, which began extensively and became the norm in the 1970s, that enabled political people to go into business as political pollsters as opposed to doing what I do, and what Gallup does, which is doing independent public opinion polling for the media, and for . . . in our case for the public itself through the Pew Research Center. But you have the independent media pollsters. You have the . . . You have the partisan pollsters. They do different things.

Question: How is your field changing?

Andrew Kohut: When I first started there were two or three major polling organizations: Gallup, and Roper, and Lou Harris. Since then, with the advent of opinion polling, the news organizations were able to go into opinion polling themselves. And they started their own polls and all of the partisan polling that we’ve talked about. So there’s much more polling. There’s now polling that’s done in using the kind of high-tech methods – polling that calls people with auto dialers. They’re not real interviewers, but recordings – so called robo-polling. There’s Internet surveys. I’m not a great fan of robo-polling. And as far as Internet surveys, you can do a very good survey on the Internet if you can get a good sample of people. But drawing a random sample of . . . of . . . of the country, or the state, or of the town is very difficult because no place to go to draw a systematic or a random sample of e-mail addresses. And I’m not a big fan, and I’m very suspicious of these polls that are predicated upon the harvesting of people who volunteer to be polled. People who volunteer to be polled are essentially different than the people we . . . we call up and have to root out of their dinnertime to take some questions about President Bush, or how they feel about health care reform.

Question: How should the public evaluate poll results?

Andrew Kohut: I don’t know. It’s very hard . . . It’s increasingly hard. A number of media organizations, for example, are taking the least expensive road . . . route, and they will do some of these surveys that I think aren’t as methodologically sound as what the CBS, New York Times poll does, or what Gallup does, or what we do. One of the most important things to . . . for people to do in evaluating an opinion poll is look at the question that was asked. Looked at the inferences that are being drawn from the results of the question, and also say to themselves, “Well how would I have answered that question? And what would that say about my opinion about what to do next in Iraq?”

Question: Do we rely too much on polls?

Andrew Kohut: Polling is a tool for leadership. It’s not the . . It’s not a program for leadership. And you can abuse a tool. You can overuse it. A leader who looks to the latest poll finding and says, “Well that’s what I should do” . . . that’s not a very good leader. I mean that’s someone who is not taking this poll and saying, “Well what am I gonna have to do to get public acceptance of my policies?” It’s someone who is interested in . . . in . . . in their own election or re-election, and their own popularity rather than serving . . . genuinely serving the public interest.

Question: All polls self-fulfilling?

Andrew Kohut: No, because so often you see polls . . . poll trend lines reverse themselves. There’s a lot of popularity for a particular figure, and all of a sudden it goes in a different direction. Look how popular President Bush was for a number of years. If polls were fulfilling . . . self-fulfilling prophecies, his approval ratings would have never come down.

Question: Do election day polls have a negative impact on voters?

Andrew Kohut: I think that the networks have done a pretty good job of policing – except to insiders – what those early takes on the exit polls have shown. You know every . . . every . . . every person . . . every member of the media political community in Washington by four o’clock has known what the early exit polls have been showing in the presidential elections. And that has a different set of problems. But in terms of the voters themselves, the networks and the media have generally cleaned up their act, and they’re not reporting at six o’clock or seven o’clock what the polls are suggesting. And I don’t think there’s much risk that voting is being discouraged by these polls.

Question: Did you ever get it wrong?

Andrew Kohut: Well we have a very good record, and the polls that I’ve done have been pretty good. I’ve had a couple of bullseyes on presidential elections where the numbers are absolutely right. And every presidential election that I’ve directed, I’ve been within the margin of error so to speak; and sometimes even well within the margin of error, right on the nose. But the one that I’ll always remember is I got the New Hampshire primary wrong in . . . in 1988. And there’s nothing worse than being publicly wrong. It is a . . . It is a . . . It is a humbling experience because everyone says, “Hey, I thought you said that so-and-so was gonna win the New Hampshire primary?” And so that, you know, I . . . I sincerely hope that somewhere along the line it doesn’t say in my obituary that this, this and this, but he got the New Hampshire primary wrong in 1988. I’m just teasing there. I’m sure it won’t. I’m hoping it won’t, at least.

Question: If you get it wrong, how do you regain the public's trust?

Andrew Kohut: Well it’s easy for us ‘cause we go out and keep polling, and we then show that . . . that we . . . we’ve learned a lesson. I mean Gallup, who again I started with back in the late ‘60s, you know he was wrong in ‘48. It shaped views about public opinion polling. It certainly shaped his notion, shaped his business. And a lot . . . You know and that was an extraordinary event in his life, and in his career, and in the course of his polling. God, there were congressional hearings about how could the polls get . . . how could all the polls get it wrong? And what he did is he turned his methodology over to a great statistician at Gallup, and they revamped what they were doing. I mean you learn from these experiences. But you know to get back to this point, I think having to get it right not only with regard to the presidential election, but to have your polls conform to . . . to what the census is showing about the distribution of the population; or what official statistics are showing about, you know, certain patterns of behavior or ownership of things. You have to work hard to make the measures, and the questions, and the sampling be accurate.

Recorded on: 9/14/07