I thought this had been put to bed by now, but a number of comments on diaries concerning the raft of marriage equality referenda coming up in the 2012 elections have brought up the "Bradley Effect" to support discussions of social desirability bias (h/t ebohlman). The Bradley effect, according to Time, which had to bring it up three weeks before the 2008 election,, is a theory that suggests
that voters have a tendency to withhold their leanings from pollsters when they plan to vote for a white candidate instead of a black one. In 1982, Tom Bradley—the African-American mayor of Los Angeles—ran for governor of California. On the eve of the election, polls anointed him a prohibitive favorite. But on election day, Bradley lost to his white opponent, Republican George Deukmejian. Some experts chalked up the skewed polling to skin color.
Occam's razor, right? Simplest explanation? Not so fast. I'm not going to pretend that social desirability bias doesn't exist, since we certainly saw it at work in 2008 in the Prop 8 vote in California, but this is NOT a good way to explain the 1982 California elections. Follow me below the great orange sticky bun for the real story.
It's not that the theory isn't seductive. The Time article went on to see evidence of it in elections in 1983 (Harold Washington in Chicago) and 1989 (Doug Wilder in Virginia and David Dinkins in New York City) -- Washington and Wilder won, Dinkins lost. It suspects that it MAY surface with Obama, but it eventually concludes that the effect has gone away:
There is no question that racial bias is a powerful force to overcome and a slippery one to quantify. But with Obama propelled by panic over shrinking nest eggs and the wilting Dow, the Bradley effect may be this fall's paper tiger: an old theory re-heated by the media because there's not much left to talk about.
The above is called setting up a straw man. It hasn't gone away;
it never existed at all. We knew that at the time. Here's Blair Levin,
in an op-ed piece that ran in the New York Times October 19, 2008:
On election night in 1982, with 3,000 supporters celebrating prematurely at a downtown hotel, I was upstairs reviewing early results that suggested Bradley would probably lose.
But he wasn’t losing because of race. He was losing because an unpopular gun control initiative and an aggressive Republican absentee ballot program generated hundreds of thousands of Republican votes no pollster anticipated, giving Mr. Deukmejian a narrow victory.
And then there's, well, ME!
It's protected by a pay wall now, but you get to believe me because the letter also ran in the Los Angeles Times two months later and
it's cited in this article! Here's the portion of the letter that ran in both places:
Yes, the poll numbers in the 1982 elections said Tom Bradley, the African-American mayor of Los Angeles would be the next governor, but he lost the election. The commentators who continue to say it was because racist white voters lied to the pollsters overlook the fact that this election was the first election in California where anyone could request an absentee ballot for any reason, and the Republican Party of California understood the implications of this decision.
As the KNBC news reporter Conan Nolan describes the events on his blog, rawstatepolitics.com, "Of course pollster Mervyn Field didn't see this coming. So when Field did his exit polling he didn't factor in the hundreds of thousands of Republican votes that had already been cast by mail. So although his conclusion (that Bradley had won) was wrong his exit polls were actually correct. The final totals from the secretary of state's office indicated that Bradley had won the statewide precinct vote on election day . . ." He also notes that Jerry Brown (white, and running against Pete Wilson, also white) lost his Senate election because of the massive number of Republican absentee ballots, and Field's exit polls didn't reflect the margin correctly, either. Has anyone ever cited a “Brown effect”?
So there you have it. The Bradley Effect is a piece of political fiction, and you may now view anyone who cites it as someone who at best has not done his or her research on the subject.