Not long ago, it was considered political gospel: Be wary of polls when an election involves an African American candidate, because many whites will voice support but vote for the white opponent.
Now, poll-watchers are asking whether that could be skewing the numbers as Democrat Barack Obama, the first African American presidential nominee, moves ahead of Republican John McCain.
Most experts say they do not believe that the phenomenon, known as the “Bradley effect,” is at work in this election. But some disagree. And if the effect has disappeared, it is not clear whether that is because polling techniques have improved or because the country has become more tolerant about race.
“The Bradley effect may have been an artifact of the country 20 years ago, but I don't think it's a factor now,” said GOP pollster Neil Newhouse. “Polling has gotten better, but I think, more importantly, the country has changed.”
The phenomenon got its after former Los Angeles Mayor Tom Bradley, an African American, lost the 1982 gubernatorial race in California despite leading his white opponent in the polls. Some experts suspected at the time that a portion of white voters, reluctant to appear biased, had lied to pollsters about which candidate they were backing.
But whether Bradley, a Democrat, lost because of hidden racism has never been clear. A post-election analysis by Mervin Field, whose California Field Poll showed Bradley up 7 points in the campaign's final stage, attributed the late shift to an unusually large number of GOP absentee voters, relatively low turnout among nonwhite voters and the coincidence of a handgun initiative on the ballot.
He also highlighted the role of race, which may have been enough to tip the balance to Bradley's opponent, Republican George Deukmejian, but emphasized that that alone would not have been enough to turn around Bradley's lead.
Finding hard evidence for or against a Bradley effect today is difficult, given the relative rarity of black candidates facing a white opponent before a majority-white electorate. Obama's performance in the primaries doesn't clarify the issue since he did worse than the polls predicted in some states and better than projected in others.
Still, some academics – mainly African Americans – say the country should not be so quick to dismiss the theory.
“I'm one of those who believe the Bradley effect is alive and well,” said Michael Dawson, a University of Chicago political science professor.
There is little doubt that the willingness of white people to vote for a black candidate has grown measurably. A December 2007 Gallup poll found that 5 percent of white respondents said they would not vote for a black candidate for president.
In 1958, Gallup determined that 58 percent of whites would not cast a ballot for a black presidential candidate, and as late as 1989, 19 percent said the same.