'Bradley Effect' And The Obama Presidential Bid As the presidential candidates head into the general election some analysts, citing the "Bradley Effect," question whether polls showing Sen. Barack Obama's lead over Sen. John McCain will shrink at the voting booth. The term was coined after L.A. Mayor Tom Bradley lost his bid for governor of Calif.
NPR logo

'Bradley Effect' And The Obama Presidential Bid

  • Download
  • <iframe src="https://www.npr.org/player/embed/95929466/95929460" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript
'Bradley Effect' And The Obama Presidential Bid

'Bradley Effect' And The Obama Presidential Bid

  • Download
  • <iframe src="https://www.npr.org/player/embed/95929466/95929460" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript


We have more now on the presidential campaign. The latest polls show Barack Obama with a clear lead over rival John McCain. The latest Washington Post-ABC News poll, for example, shows Obama leading McCain 53 to 44 percent. But throughout the campaign, analysts have questioned whether some voters are actually being honest with pollsters when they say they will vote for an African- American candidate. It's been called the Bradley effect. The term was coined after then-Los Angeles mayor Tom Bradley lost his bid for governor of California in 1982. Bradley, who's African-American, had a solid lead in the polls going into Election Day but then lost by a slim margin.

Was it because he was black? Was it that people were not honest with pollsters? Could the same dynamic be at play in this election? Blair Levin worked on the Bradley campaign. He recently wrote a New York Times opinion piece arguing that Bradley's loss had actually nothing to do with his race. He's here with me in the studio now. Also with us is veteran Democratic pollster Ron Lester. He is on the phone. Welcome to both of you.

Mr. BLAIR LEVIN (Op-ed Contributor, New York Times): Happy to be here.

Mr. RON LESTER (Democratic Pollster): Good to be here, Michel.

MARTIN: Blair, let me start with you. You worked on Tom Bradley's 1982 gubernatorial campaign. You say that race wasn't the reason. So what led to the discrepancy between the polling data going into the race and what actually happened on Election Day?

Mr. LEVIN: Yeah, well, just to be clear, what I'm saying is that the so-called Bradley effect of people lying to pollsters, that wasn't the reason. Race mattered. Tom Bradley was an African-American. He was by then very popular in Los Angeles, where he had been mayor for several terms, but in the rest of the state, you know, I don't know the extent to what race is the reason, but I'm simply saying that the so-called Bradley effect wasn't true.

And the reason I say this is I was actually one of the people looking at early returns, and what we saw were two phenomenons that reflected - that the kind of the polling estimates which, of course, you have to make certain assumptions about who's showing up. One is, there was a much greater turnout in the Central Valley of California because of a gun initiative. And generally, that meant that there were pro-gun voters who tended to vote Republican showed up, if I recall correctly, a hundred thousand,a couple hundred thousand.

And secondly, it was the first year where people could vote absentee without having to give an excuse like you're going to be out of town. The Republicans, and we didn't know this at the time, I think spent about a million dollars getting Republican absentee voters, and I think had about 250,000 extra votes. That was more than the margin of victory, so Bradley actually won the day, he actually won for who voted on Election Day, but it wasn't enough because of those absentee voters.

MARTIN: Ron Lester, what's your take on this?

Mr. LESTER: Well, I think too much has been made of the so-called Bradley effect. I thought that was a very good article in the New York Times yesterday by Mr. Levin. There was another one in the Wall Street Journal by Sal Russo, a respected Republican consultant who worked with Deukmejian at that time, who was the candidate who defeated Tom Bradley. Both of them essentially said the same thing, that the dynamics of the race, the turnout in the Valley, the handgun initiative, all of those affected the race at the end. And particularly, Mr. Russo had indicated that their polling showed that it was a single-digit race all the way up until the end.

MARTIN: But where does this come from, though? I mean, what about - the other example of the so-called Bradley effect is the Doug Wilder's race for Governor of Virginia, which he did win, but going into the race he had a five-point lead, and then there was a half-a-percentage point lead. What do you think about that?

Mr. LESTER: Pollsters at the Bradley race had it right. And Governor Wilder's pollsters would tell you that they knew the race was going to be neck and neck. There was a Washington Post poll that came out the Sunday before, but which was actually done about a week before the election, and at the end the numbers moved. So it wasn't so much people giving socially acceptable answers, but that the Marshall Coleman campaign spent about a million dollars at the end on negative ads, and it brought Governor Wilder's numbers down. But I don't think that his pollsters were expecting a 15-point victory.

And I think polling has come a long way since those days. What we do nowadays when we ask the horse race question, if the election were held today, would you vote for Democrat Barack Obama, Republican John McCain? And if someone says, Barack Obama, they will ask another question. Is your support for Barack Obama very strong, somewhat strong, or not strong at all? And we might discount the not strong at all or look at the demographics to see who they are. So there are other indices that we build into polling to calibrate and measure responses now.

So with a race like Harold Ford Jr.'s race against Bob Corker for Senate last year, in 2006, in Tennessee, which was a racially charged race, the polls were dead on. We had it at about five or six percent margin, and that's about the margin that Harold Jr. lost by.

MARTIN: We need to take a short break. We're going to ask both of our guests to stay with us so we can talk about this for a couple of more minutes. Our guests are Democratic pollster Ron Lester and Blair Levin. He worked on the 1982 Bradley campaign. He wrote recently about the so-called Bradley effect for the New York Times. We're going to talk a little bit more about whether there really is a Bradley effect or whether this is some kind of urban legend. Please stay with us. Our conversation about race and opinion polls continues next on Tell Me More from NPR News. I'm Michel Martin.

(Soundbite of music)

MARTIN: I'm Michel Martin and this is Tell Me More from NPR News. Coming up, everyday ethical dilemmas. Unsolicited advice from an in-your-face friend, a couple getting a little too cozy with each other in your presence. It's our monthly visit with the O Magazine ethics panel. That's still ahead.

But first, we want to continue our conversation about race and the presidential campaign. We're talking about the so-called Bradley effect, whether people are really honest with pollsters and maybe even themselves about how race may factor into their voting decisions. Our guests are veteran Democratic pollster Ron Lester and Blair Levin. He's a former campaignee to Tom Bradley who recently wrote about the issue for the New York Times.

Blair Levin, where do you think this concept came from? This idea, the Bradley effect, it surfaces at major races every time there's an African-American candidate.

Mr. LEVIN: Well, politics is such an inexact science. You can't really run experiments and determine what caused anything, but I think you look at a couple of things. First of all, something that was just very special to California, the most respected pollster in 1982, Mervin Field, had some polls that showed Bradley significantly higher or with a much greater gap than I think our internal polls had and certainly the Republican polls had. So it looked like more.

Second, because of the exit polls, you actually had the phenomenon of 3,000 people that - I was in the hotel where we were celebrating, and you had 3,000 people celebrating and then suddenly losing. I think it was more dramatic, but it was followed up by a series of other races, one that I note in particular. In 1990, Harvey Gant running for the Senate against Jessie Helms, where there was also a gap there. And I think North Carolina is a different state. It's a different race, Jessie Helms a very different kind of opponent than George Deukmejian, and there may well have been that effect there.

MARTIN: But there were explicit racial appeals in that case, that white-hands ad. I don't know how people consider that a coded racial message. It wasn't coded at all. It was very explicit.

Mr. LEVIN: Well, it wasn't a coded message, but I think North Carolina is, it - there may have been more of a reluctance to talk to pollsters honestly. But you had a couple of different events that people put together and thought they saw a pattern, but I'm not sure that pattern is really there. In fact, the Bradley effect might be more accurately described as the Gant effect because I think in Gant's case it may have been true more so.

MARTIN: Ron Lester, what's your take on that?

Mr. LESTER: Well, I think that in any poll you do, you get what we call socially acceptable answers, but in the case with a black candidate and a white candidate, you know, this has been characterized from time to time as the Bradley effect. I think you have to take a careful look at the dynamics of each race. I think that there may have been more of the so-called Bradley effect in the Harvey Gant race. I mean, it was up by about five or six points, and the white-hand ad came on and he lost by about one. So that probably caused some movement, I think it's fair to say, probably turned some white voters around who considered working for Harvey Gant.

But you know, if you look at the Bradley race, you had that handgun initiative that really boosted turnout in the Valley, you know. Another example of this is Hillary Clinton versus Barack Obama in New Hampshire of 2008. Barack was up 13 points, and some people thought there was a Bradley effect when Hillary ended up winning. What happened there is Barack had a strong lead among Independents, but when Independents thought that Barack had won, they went ahead and voted for McCain in the Republican primary. So there are a lot of factors that you have to look at. It's just not, you know, a very simple case. You can't just come up with a broad kind of solution and say that the Bradley effect - you know, perhaps sometimes it works to some extent, perhaps sometimes it doesn't. It definitely wasn't there with Howard Ford Jr. race because we called that race right on the money.

MARTIN: Do you take special measures to account for whatever you want to call, sort of, politically correct answers when it comes to a race where there's a racial difference?

Mr. LESTER: Yes. What we'll do is in addition to asking the horse race or what's called the valid question of the election world(ph) today, and the candidates are Democrat Barack Obama, Republican John McCain. If they say Obama, we'll say, is your support for Obama very strong, somewhat stronger, or not strong at all? And we'll discount the not strong at all for both candidates. So we'll measure and calibrate support that way.

MARTIN: But why do you do that?

Mr. LESTER: To measure the strength of support, to see if a barrage of negative ads come along, who are the voters who are most likely to be driven away from the candidate, and they will be the weakest ones. They are not-so-strong supporters.

MARTIN: But do you assume that there is - their support for the African-American candidate is by definition more susceptible to manipulation by negative ads?

Mr. LESTER: Well, it's not just African-American support for American candidates. We do this for everybody. We assume that not-so-strong support can be taken away by an effective negative ad.

MARTIN: Blair, what's your take on that?

Mr. LEVIN: I'm not a pollster, and it's been a long time since I've really worked full time...

MARTIN: But do you think, though, that this is perhaps is the time to just retire the term entirely?

Mr. LEVIN: Actually, what I was writing and part of what motivated me to write the article, I'd like to change the meaning of it. I also worked for Bradley as a young guy. I was his driver in his 1973 race for mayor, and I saw the effect that he had on a lot of different crowds. I think it is a huge problem for this country to have the kind of leadership that appeals to a multi-cultural society, and that's what we need. We have to move forward.

And we are a multi-cultural society. I get very disturbed when I hear people saying things like, some Americans or real Americans and other Americans aren't real Americans, and I think what we ought to talk about in terms of the Bradley effect is the effect that the guy had as 20 years of mayor of Los Angeles. An incredibly diverse city, yet he was able to forge constant alliances. He didn't divide the city. Everybody was welcome at City Hall. That's what I want the Bradley effect to mean.

MARTIN: I understand. Ron Lester, what about you? As a pollster, do you think it's time to retire the term?

Mr. LESTER: I think that the term will become less significant over time as voters see that polling has come a long ways. The art and science of survey research has really made a lot of progress over the years, and I think that voters, the public, the press will see after this campaign that most of the polls - that the good polls were done right.

MARTIN: But is it...

Mr. LESTER: In the polls that were done right, the good polls will hold up.

MARTIN: But is it possible that people are just more honest than people - or maybe we're willing to give them credit for being?

Mr. LESTER: Well, I think, as I said earlier, you're probably going to have some people that give socially acceptable answers in the poll, but whether or not they're enough to be suggestively significant, that's the question that really matters.

MARTIN: All right. Ron Lester is a veteran Democratic pollster and a regular contributor to this program. He was kind enough to join us on the phone. Blair Levin served as a staffer on the Tom Bradley campaign in 1982. He recently wrote an opinion piece for the New York Times about the so-called Bradley effect, which he'd like to recast into something that means something very different. He was kind enough to join us in our studios in Washington. Gentleman, thank you both so much.

Mr. LEVIN: Thank you.

Mr. LESTER: Thank you, Michel.

Copyright © 2008 NPR. All rights reserved. Visit our website terms of use and permissions pages at www.npr.org for further information.

NPR transcripts are created on a rush deadline by Verb8tm, Inc., an NPR contractor, and produced using a proprietary transcription process developed with NPR. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.