The “Bradley effect” is named after Tom Bradley, a black Los Angeles mayor who ran for California governor in 1982. Pre-election polls gave him a sizable lead over his Republican opponent George Deukmejian, who is white. But the GOP candidate won by a narrow margin in the end.
Analysts suggested that white voters may have concealed their preference for Deukmejian in public opinion polls, telling interviewers they were undecided or favored Bradley in order to avoid the perception of racism. And subsequent polls overestimating the performance of black candidates like Virginia Governor L. Douglas Wilder and New York Mayor David Dinkins only bolstered the theory.
Political scientists say there has been no systematic study of whether the phenomenon carries over into races involving gay and lesbian candidates. But professor Patrick J. Egan of New York University has conducted research on the polling and voting patterns surrounding same-sex marriage and domestic partnership ballot measures in 33 states.
Egan’s study found that pre-election polls are generally accurate in predicting Election Day support for same-sex unions, but consistently undercount the opposition. Hence, the surprise passage in 2008 of a ballot measure in California that banned same-sex marriage and another in Maine last year that overturned a gay nuptials law.
Those election results fed speculation around a gay Bradley effect. But Egan, for his part, doesn’t buy it. Among his arguments: a gay Bradley effect should become more pronounced over time as homophobia becomes less acceptable. In fact, he says, the gap between survey results and election results has narrowed over the years.
Donald P. Haider-Markel, a professor of political science at the University of Kansas and the author of Out and Running: Gay and Lesbian Candidates, Elections, and Policy Representation, says he too has seen little evidence for a gay Bradley effect.
And the Bradley effect itself, he notes, is disputed terrain. Some analysts say that pollsters in Bradley’s race may have simply failed to account for a sudden surge in conservative voters drawn to the polls by a ballot initiative focused on handguns.
Others argue that, if there was a Bradley effect in that California race and subsequent contests, it has not been a significant factor in polling for some 15 years. Surveys in the 2008 presidential election were generally accurate in predicting Barack Obama’s performance.
But Haider-Markel says Egan’s central finding — that polls consistently underestimate opposition to same-sex unions — could point to another, equally vexing problem: a growing body of evidence that conservative voters, the sort who might object to gay marriage or a liberal gay candidate, are less willing to participate in public opinion surveys than other voters.
Not the Bradley effect, perhaps. But either way, pollsters are undercounting the right-wing vote. That could mean an unexpected bounce for Loughlin on Election Day. If he wants to make a go of the race, though, he needs to demonstrate all the support he can right now.
OFF THE MARK A Brown survey predicted a Cicilline win, but didn’t gauge the support of the also-rans.
If the lack of public polling in the Providence mayoral race added to the wow factor in Taveras’s landslide victory, it was an impressive feat in other ways, too.