With the economic crisis center stage, Barack Obama has solidified his lead, pulling ahead of John McCain by 10 points among likely voters, 53-43 percent, in a recent ABC News/Washington Post poll.
But the question remains: Is that margin enough?
"We've seen these types of leads disappear at election time," said Michael Dawson, professor of political science at the University of Chicago.
The concern is particularly at play for black candidates. It is called the Bradley effect, named after former Los Angeles Mayor Tom Bradley, an African-American, after he ran for California governor in 1982. Some pre-election polls showed Bradley with a lead of 9 points or higher; but Bradley lost to Republican George Deukmejian by a little more than a point.
As the theory goes, people lied to pollsters, saying they would vote for Bradley when they did not.
The Bradley effect could just as easily be called the Wilder or Dinkins effect.
In the 1989 Virginia governor's race, one poll showed Douglas Wilder with an 11-point lead. He ended up winning, but by less than a point. In the 1989 race for New York City mayor, one poll showed David Dinkins with a 14-point lead. He won by just 2 points.
Some analysts suggest that voters will tell pollsters one thing, because they don't want to seem prejudiced, and vote differently.
"When you are in the voting booth, nobody is there and you can express what you really believe," Dawson explained.
Across the country, many voters expect that Obama's race will be a factor in the election.
"My father is kind of a racist guy," said one voter from South Carolina. "He's not going to vote for someone who's even light-skinned."
But the Bradley effect points to race being a bigger factor. It says voters are misleading pollsters, but some experts said there is simply no reliable evidence to prove that.
"If people wanted to lie to us, it would be much simpler for them simply to decline to participate in the poll in the first place," said Gary Langer, ABC News' polling director.
Modern history seems to disprove the Bradley effect as well. In 2006, there were five statewide races with black candidates for U.S. Senate or governor that the polls got right each time.
"When a pre-election poll goes bad, usually the reason is it made a bad estimate of who was going to show up to vote, not that its respondents lied," Langer explained.
Analysis from ABC's polling unit found 25 polls that understated support for Hillary Clinton. However, there were just as many that understated support for Obama; not a single poll overstated support for Obama outside the margin of error.
Moreover, there are key players who were around at the beginning of the reputed Bradley effect, who say it was never true to begin with.
Ken Khachigian, who ran the campaign that defeated Bradley, said that the candidate's loss was about absentee ballots, not latent racism.
"There were well over several tens of thousands of absentee ballots that no exit poll would ever pick up because absentee balloting was not a big deal back in 1982, so they missed all of those," Khachigian said.
Michelle Obama also suggested that the nation had come a long way since 1982 in terms of racism.
"If there was going to be a Bradley effect, or it was going to be in play, Barack wouldn't be the nominee," she said in an interview with CNN last week. "We have to focus on the country as it is. That was several decades ago. And I think that there's been growth and movement."
Despite the lack of empirical evidence, the Bradley effect lives on, fueling anxiety and nervousness among many Democrats that Obama's lead will disappear on Election Day.
But as the old adage in polling goes, the only votes that count will be the ones cast Nov. 4.