Bradley Effect May Not Hold On Tuesday

While election polls show Barack Obama leading in tomorrow’s presidential election, a key question throughout the election has been whether polls overstate the support of black candidates. Fortunately for the Obama campaign, a new Harvard study finds that the senator may not have much to worry about.

In November 1982, Los Angeles Mayor Tom Bradley, a black Democrat, was running ahead in the polls for governor of California and was feeling confident going into election day. Pre-election polling showed that he had a significant lead over his Republican opponent, and exit-polling conducted on the day of the election predicted that Bradley would emerge victorious.

But when all of the ballots were finally counted, the pollsters were proven wrong—fewer white voters had actually cast their ballots for Bradley than polling had anticipated.

Daniel J. Hopkins ’00, a post-graduate fellow in government and social studies, said that the “Bradley effect” is “a systematic difference between how black candidates poll before the election and how they perform in the election itself.” But Hopkins said that the Bradley effect is unlikely to play a major role in future elections.

In his study, which was released earlier this year, Hopkins analyzed gubernatorial and U.S. Senate elections between 1989 and 2006 that featured either a black or female candidate. Using three pre-election polls per candidate, Hopkins found that the Bradley effect was last seen in the late 1980s and early 1990s. In recent years, he said, the effect has gone away.

“In the last ten years of elections, I can’t find any evidence that there’s a systematic difference in how candidates poll and how they perform in the election itself,” Hopkins said.

He added that he does not anticipate that there will be a significant Bradley effect in Tuesday’s election.

But Evan P. Apfelbaum, a doctoral candidate at Tufts, who recently authored two articles on how people deal with race in social situations, said that he had his doubts about whether or not a Bradley effect exists.

“I’m not entirely convinced that such a discrepancy [between pre-election polls and election results] would necessarily be the result of individuals who are afraid of appearing racist,” Apfelbaum said.

A co-author of the paper, Harvard Business School professor Michael I. Norton, said there is little empirical evidence for the Bradley effect, since it is based on a very small sample size of black politicians who have run for public office.

“Any time you’re making conclusions about the influence of one variable like race in a small sample, you have to be very cautious,” he said.

Although these scholars predict little or no Bradley effect in the upcoming election, Hopkins said that “there are lots of ways that race can influence voter decision-making. The Bradley effect is only one of them, and it’s a relatively fragile form of racial influence.”