Is there a 'Bradley Effect'? Or perhaps the reverse?

With most opinion polls now giving Barack Obama a substantial lead in the presidential race, the big unknown is the possible 'Bradley Effect' - a reference to what seemed to happen back in 1982 when the black mayor of Los Angeles, Tom Bradley, ran for the Governorship of California.

Before that election most polls gave Bradley a substantial lead, but then he lost. In short, the effect was identified as a distortion in opinion polls caused by voters who didn't want to vote for a black candidate, but didn't want to admit that to pollsters.

A similar effect was cited by some analysts - rightly or wrongly - when Hillary Clinton confounded expectations, and the polls, and beat Obama back in the New Hampshire Democrat primary earlier this year.

But my former Newsnight colleague Peter Kellner, now president of the polling company Yougov, has written an article for the Fabian Society suggesting that the Bradley Effect may be a bit of a myth these days. Indeed, he argues that current US polls may in fact UNDERSTATE Obama's lead, simply because, on the basis of past voting behavior, they may underestimate how many black voters will turn out. But with Barack Obama in the running this time, the argument goes, black people may be a lot more inclined to cast their ballots than ever before.

Intriguing stuff.