MICHELE NORRIS, Host:
Okay. We know the New Hampshire primary is so over. Candidates have pressed ahead to focus on other contests, in places such as Michigan and South Carolina. However, two days after New Hampshire, pundits, pollsters and the press corps, still looking back, trying to figure out why so many polls were so wrong. Consider this sampling from today's newspapers.
ROBERT SIEGEL, Host:
Pundits were the big losers in New Hampshire primary.
NORRIS: Where did pollsters go wrong?
SIEGEL: Polls ain't perfect.
NORRIS: And this from The New York Times op-ed page.
SIEGEL: The failure of the New Hampshire pre-election surveys to mirror the outcome of the democratic race is one of the most significant miscues in modern polling history.
NORRIS: That was the first sentence in op-ed written by Andrew Kohut. He's the president of the Pew Research Center. He's also one of our contributors. And he joins me now in the studio.
Welcome, Andy.
ANDREW KOHUT: Thank you, Michele.
NORRIS: Now, perhaps with two days now to look at these results, you can tell us what went wrong. Why did so many polls overestimate that wide lead for Senator Obama, underestimate support for Hillary Clinton, yet it seemed to get just about right in the Republican side? What went wrong?
KOHUT: Well, it wasn't general methodology or else the McCain race would not have been called correctly. It didn't have to do with the subtleties of survey methodology because the really rigorous polls, CBS and Gallup, had the same kind of overstatement, and as the polls that have not as much rigor in the way they pursue respondents and interviews. It wasn't, I don't think, a matter of a last-minute trend because only 17 percent of voters said they decided that very day, and Hillary Clinton only had a three percentage point margin over Barack Obama. That could not possibly explain the overstatement of Obama in the pre-election polls.
NORRIS: So we talked about some of the things that it didn't explain, what does explain it?
KOHUT: Well, we're not sure. But one of the things that you can't preclude is race.
NORRIS: The so-called Bradley effect.
KOHUT: Well, the so-called Bradley effect or that some whites are reluctant to vote for black candidates. And there are some reason to suspect that that might be in play, given the character of Hillary Clinton's support. Her support is disproportionately among poor people and less well-educated people. And we know that surveys underrepresent poor and less well-educated people. Statistical adjustments are made to bring the percentage of them up in samples. But we suspect that there are, I suspect that the people who we are not interviewing are different from the people who we are interviewing in that category. And I don't think it's a surprise that the people who are less well- educated and poor have more negative opinions of African-Americans. So that's a possible explanation.
NORRIS: That's interesting, because you suggest that perhaps they're not participating in these polls or these surveys. But in previous cases, this so- called Bradley effect is named for former Los Angeles mayor, Tom Bradley, where the polls show that he had significant support among white voters, and when they actually were in the (unintelligible) to the voting booth, a much smaller number of them actually voted for him. So, same thing with David Dinkins in New York, the same thing with Governor Doug Wilder many, many years ago.
KOHUT: Yeah. Well, you can believe that that's the problem. That respondents are lying to the pollsters in the Dinkins race. I was the victim of this. But from my point of view, it wasn't so much that people were lying, it's that the people we weren't getting had different opinions than the people we were getting.
NORRIS: Since we're talking about a problem or an issue that was specific to Hillary Clinton and Barack Obama, I wonder if there is something that is unique to historical candidates, people might get last-minute jitters or something tantamount to that when they get ready to cast their vote.
KOHUT: You're right. I mean, I think we're a long way for understanding the motivation behind this. We're a long way from really saying that this is what's going on. But we cannot rule it out; we have to consider. The last thing pollsters should say is the people that we don't get are always like the people we do interview because we know that in some cases, they're not. And this may be one of the instances.
NORRIS: Sounds like that's the most important lesson you are taking from this.
KOHUT: Yeah.
NORRIS: Thank you, Andy.
KOHUT: You're welcome.
NORRIS: That was Andrew Kohut. He's the president of the Pew Research Center for the People and the Press.