One month ago, an Associated Press/Yahoo News poll indicated that Barack Obama would struggle in his bid to become president. His problem wasn’t a foot-in-mouth incident or his tax policy. His problem, the poll concluded, was the dark hue of his skin. It turns out a significant number of white Americans have negative views of blacks, including Democrats, and, the poll found, it might cost Obama as many as six percentage points on Election Day, which is now less than two weeks away.

That’s a formidable problem given that the last two presidential elections ended with margins much smaller.

But wait a second, says University of Washington professor Anthony Greenwald, who studies the so-called “Bradley effect,” which holds that polls overestimate black candidates’ chances because of the elusive nature of racism. It appears a “reverse Bradley effect” is in the making this election–good news for Obama.

As Greenwald says, Obama is “likely to do better than polls indicate.”

To understand all this, pull out your political history books and turn to the chapters dealing with 1984. That’s when Los Angeles’ first and only black mayor, Tom Bradley, was destined, polls predicted, to become California’s first black governor. Instead, Bradley lost narrowly to Republican George Deukmejian.

Afterward, researchers scrutinized the data and found that the polls failed to account for negative views about blacks. Voters had avoided pollsters’ questions or provided what they thought was the right thing to say to pollsters. Some even appeared to be lying to themselves.

If the Bradley effect holds true for this election, it would mean major national polls, which in recent weeks have indicated without exception that Obama will be the next president, are overestimating Obama’s chances. In other words, Obama’s average 7.6 percent lead over Republican John McCain isn’t quite that strong.

But Greenwald and his colleague Bethany Albertson conducted a detailed analysis of more than half the nation’s states during the primaries earlier this year. They identified a reverse Bradley effect in 12 states where Obama and Senator Hillary Clinton battled. In those states, Obama won by wider margins than polls predicted, Greenwald says, referring to a graph full of state names colored in red, blue, and purple.

In fact, the Bradley effect only held true in three states: New Hampshire, California, and Rhode Island.

As for Colorado, which favors Obama by 5.4 percentage points, it remains unstudied, Greenwald says, adding that many factors could apply, including the small size of Colorado’s black population, and the views of Latinos, another area he has yet to analyze.

Age, Greenwald notes, also plays into the Bradley effect. Older voters tend to favor McCain because of his age and because he’s white. As Greenwald says, “Older voters grew up in a time when it was more acceptable, even legal, to express discrimination.” And age, he adds, was even an indicator of the effect in the primaries: Obama’s supporters tended to be younger, part of a different generation with different views about race.

So, if Greenwald’s research is solid come Election Day, it appears race won’t be a factor. Voters, he notes, seem more concerned about the Bush administration’s handling of the faltering economy–another indicator that the reverse Bradley, or an underestimation of Obama’s support in polls, may be what defines 2008.