Election Day question: Will mathematical odds prevail over volatile feelings?

The polls point toward an Obama victory. But what if the polls are wrong, asks Slate's John Dickerson. The most famous wrong poll, of course, was Gallup's in 1948, which said Dewey would defeat Truman. But Gallup's last tracking in the 1948 race was more than a week before the election, after which Truman gained a lot of ground and, of course, ended up winning.

Sixty years later, polls are more finely tuned. There are also more of them, and they track opinion closer to the Election Day. Still, Obama has not succeeded in pushing his percentage that much above 50% -- except in the last Gallup poll -- from Nov. 2 -- where he is given a 55 to to 44 lead when undecided votes are allocated.

McCain's problem is that he has to come from behind in so many states that used to be reliably red. Mathematically, the odds are way against him. But the potential for surprising results in this election is high.

Will a Republican base energized by McCain's pick of Sarah Palin as his running mate offset a big turnout by young voter weighted toward Obama? What about the "Bradley effect" where white voters, once they're in the secrecy of the voting booth, supposedly decide not to back a black candidate? Will the disproportionate high percentage of Catholics who are reportedly undecided -- as many as 11% -- tend toward McCain, whose pro-life position mirrors the church's? What if voters who initially leaned toward Obama as the financial crisis exploded, express their swirling fears in a sudden new direction -- toward McCain?

So many questions, for which, before the votes are counted, there are no sure answers.