< Ghost of Bradley Effect


Friday, July 03, 2009

BOB GARFIELD: And I’m Bob Garfield, with more cultural correctives from our archives. We recorded this piece in what was the latter part of the last election cycle. As Barack Obama continued in his march toward victory, pundits grew increasingly skeptical. Could he pull it off? They asked, or would he be thwarted by the so-called Bradley effect? BROOKE GLADSTONE: The term refers to the 1982 California gubernatorial race between Republican Attorney General George Deukmejian (who’s white) and Democrat and five-term Los Angeles Mayor Tom Bradley, who’s black. Even though the polls projected a Bradley win, he lost, supposedly due to hidden racism. In October 2008 political watchers obsessed over the Bradley effect, or whether there could be a reverse Bradley effect (where whites lied about supporting Obama) or even a Bradley effect-effect whereby voters might be influenced by all the blather over the Bradley effect. Blair Levin worked for Bradley in the 1982 campaign, and he says we're reading the wrong draft of history. BLAIR LEVIN: I'm not a pollster, I'm not a statistician, but I was a witness to history. And to hear the history kind of inaccurately reflected caused me sometimes to laugh and sometimes to get a little upset. LANCE TARRANCE: I was the chief pollster in that election and part of the strategy team. BROOKE GLADSTONE: Lance Tarrance was working for Bradley’s opponent, Deukmejian but, like Blair, he’s had enough of all the references to the effect. LANCE TARRANCE: In fact, I took it personally because so many people were quoting what happened in ’82 that probably weren't even alive then, number one, and, number two, had their facts messed up. But what really determined that it was a close election was the tracking. That was where we did interviews every single night, and from that we knew we were doing better and better. We started out a double-digit behind in early October and by mid-October had climbed within seven points or so. By the end of the election, we knew it was within one percent, either way. The Republicans did an absentee vote program, stronger than the Democrats. The Democrats probably took that election a little bit for granted. BLAIR LEVIN: That day we all had the emotion we thought we were going to win, but we also knew there was a significant risk we wouldn't. It was going to be a close election. BROOKE GLADSTONE: Bradley campaign staffer, Blair Levin. BLAIR LEVIN: The night of the election, when the exit polls had declared Bradley the victor, there were 3,000 people celebrating downstairs at the hotel where we were having the party. And I was upstairs looking at the early polling results and realizing that, first, there were a huge number of absentee ballots that were not in our calculations and that they must have come from the Republican side, and then, second, that there was a turnout in Central Valley that was significantly greater than we had anticipated. And that had to do with the fact that there was this gun control initiative, which we knew would be a problem. But we had not anticipated the extent to which people in the Central Valley would show up, and they were largely Republican voters, as well. BROOKE GLADSTONE: That blew a hole in the Democrats’ arithmetic. LANCE TARRANCE: And they lost by a small margin, but they still lost. BLAIR LEVIN: I think it was less than 100,000 votes. BROOKE GLADSTONE: But Levin says hidden racism did not cause Bradley’s defeat. BLAIR LEVIN: Well, to be clear, race definitely played a role in the election in a way that I suspect we'll never know. But what I feel very confident of is that those who say that there’s this Bradley effect in which people lie to pollsters, that was not the cause. Bradley actually won Election Day. That is to say, if you take out the absentee ballots he would have won. BROOKE GLADSTONE: Lance Tarrance. LANCE TARRANCE: Now, the suggestion that race was a factor – and this is very important – really came from the preeminent public pollster in the state, Mervin Field, who ran the California poll. And he'd proclaimed rather loudly right before the election that Bradley would win handily, in fact, by seven or eight points. The press accepted this as probably going to make history. It was Merv Field, after the election was over, after the embarrassing results, who said, race must have been a factor; that’s the only reason I can think of why our data was so off. But he never once really proved his case. And so, from his comments after the election, the Bradley effect was born. BROOKE GLADSTONE: And Tarrance, still a Republican strategist, says it persists because it’s politically marketable. LANCE TARRANCE: This is purely a suggestion on my part of how it would be used in a marketing sense. You would tell particularly the black community that there is a lot of hidden racism that’s going to somehow turn the election around and that you need to make sure you vote. Why else would a theory as porous – that is, not solid – be carried forward so long in American politics, unless people wanted to keep race front and center? And it’s quite historical to have a black man running for President of the United States, and a lot of people want that to happen. And, indeed, Obama’s favored to win this election. So it becomes one of these shadowy mysteries that maybe he might be denied this election at the last moment by invoking this Bradley effect, and I think that is a disservice to the American voting public. BLAIR LEVIN: The thing that struck me personally most - BROOKE GLADSTONE: Blair Levin. BLAIR LEVIN: Tom Bradley was really a wonderful person and really practiced the politics of inclusion. He presided for 20 years over an incredibly multicultural city. He helped bring the city together in many, many ways. And the notion that Tom Bradley is now remembered to a generation as part of this incident in which race played a very significant role, or at least that’s the urban legend, saddened me tremendously. I mean, Tom Bradley ought to be remembered for something else – how he governed. BROOKE GLADSTONE: The Bradley effect had more or less dropped out of the political lexicon when returned with a vengeance after the New Hampshire primary, famously lost by Obama after he was leading in the polls. In fact, in a Nexis search, we found only 87 mentions of the Bradley Effect in the five years prior to that primary, but in the nine months that followed, it was mentioned more than a thousand times. But polling wunderkind Nate Silver, who blogs at fivethirtyeight.com, told us there was no Bradley effect in New Hampshire. He says Obama’s numbers went up after a decisive victory in the Iowa caucus, but a lot happened in the days leading to New Hampshire that the polls didn't have time to reflect. NATE SILVER: In that interval between – I think it was a Thursday when Obama won the Iowa caucus and the Tuesday when New Hampshire voted, you had a debate, Hillary had a very dramatic kind of teary moment at a press conference, and you also had a Republican race. And in New Hampshire they have an open primary. Literally, you kind of go in and decide, do I want a red ballot or a blue ballot, you know, Democrat or Republican? About half of the voters are independents, so I think a lot of independents who liked both Obama and McCain might have looked at those polls and said, Obama’s eight points, nine points ahead. McCain’s in a very tight race with Mitt Romney. I'm going to have more leverage if I vote for McCain ‘cause I want to see those two candidates nominated. BROOKE GLADSTONE: So you refer to a study by a professor at Harvard as the definitive study on the Bradley effect. What did it find? NATE SILVER: Yeah, it was a professor, I think, named Daniel Hopkins, who’s a post-doc at Harvard. And he went back and looked at the polling in, you know, all sorts of races - gubernatorial races, Senate races, as well as this year’s primaries, going back to 1980. When he found there did used to be a Bradley effect, which was pretty noticeable – maybe a five- or six- point margin - where a black candidate would underperform his polls. In Chicago, Harold Washington won his mayoral race but was supposed to win by a larger margin. You know, David Dinkins in New York, you know, Wilder in Virginia; it’s sometimes called the Wilder effect, as well. So there were enough cases in that window from about 1982 to 1991. But sometime in the nineties you stopped seeing this where the polls are accurate. You know, for example, for Harold Ford in the Senate race in Tennessee, he lost, but he [LAUGHS] didn't lose by any more than was expected. So why it disappeared is the interesting question to me. The Harvard study speculates maybe it was the fact that issues like crime and welfare and affirmative action, which were, you know, really hot button issues during the eighties, are not really kind of at the forefront of the discussion today. I also think the fact that the electorate has changed, where racism is inversely correlated with education levels – right now I think something like 75 to 80 percent of the electorate has been to at least some college where you'll integrate with students of different backgrounds, whereas in 1988, ’84, that number was more like 60 percent. It’s generational as well. I mean, it’s probably just these things turning over slowly at a time, but there’s no evidence that the Bradley effect’s existed for the past 10 years or so. We're kind of relying on quarter-century-old examples to kind of prop this idea up. BROOKE GLADSTONE: So it’s a combination of essentially two things – our society has changed and polling methods have also changed, right? NATE SILVER: I, I think so. I mean, you know, I also think people’s attitude toward polls may have changed. If a pollster calls you, you see them as a stranger, whereas before you might see them almost as a neighbor or a friend and then you might gravitate toward socially desirable behavior. That’s where the whole notion comes from, the idea that people want to conceal their racism, I suppose. But if you see the pollster as just an anonymous nobody on the end of the phone - and some of these polls now actually use automated calling scripts, so-called robo-calls. You'll press a touch-tone, literally, so there’s no human interaction at all. So that might be another factor, is that people feel [LAUGHS] more free to express themselves. BROOKE GLADSTONE: Nate, thank you so much. NATE SILVER: Yeah, thank you. BROOKE GLADSTONE: Nate Silver is the founder of the blog, FiveThirtyEight.com.