If in fact 30-35% of the voters will get to the polls early, and if they favor Obama by margins higher than the national polls suggest for the electorate at large (a function, if nothing else, of Obama’s well-funded campaign organization), then the math is hard to ignore. Suppose turnout is 150 million (a conservative estimate, I think). If 30% vote early, that’s 45 million. If Obama gets 60% of the early voters, he starts election day with a 9 million vote lead. At 55%, he starts with a 4.5 million vote lead. Under the first scenario, McCain would have to win 54% of the votes on election day to pull even. Under the second, he’d have to win 52.5% of the votes. Only the second seems even a remote possibility.
Of course, the national result doesn’t matter, except in the minds of those who bear bitter memories of 2000. Not all states permit early voting. Among the battleground states that make it easy are Colorado, Florida, Indiana, Iowa (I’m being generous to McCain here), Nevada, North Carolina, Ohio, and Wisconsin. McCain has a tiny lead in the RCP average in only one of them (Indiana). I don’t have time to do the math for all these states, but, again, McCain would have to perform significantly better than the polls to overcome a lead Obama established in the early voting.
This is not impossible, especially if the Obama campaign got a higher proportion of its voters to the polls early, leaving the McCain a somewhat larger share of the Election Day electorate. But if you give a well-organized and well-funded campaign a couple of extra weeks actually physically to get its voters to the polls, they’re going to take advantage of it. The turnout of likely Democratic voters will be higher than ever, since there’s more time to drive busloads and vanloads of voters to the polls.
An eleventh hour break in McCain’s direction is highly unlikely to overcome the advantage Obama has established. Hillary Rodham Clinton would, on some level, sympathize.