Heading into the final stretch of one of the closest presidential elections in recent history, many are beginning to handicap the potential behavior of undecided voters and Obama’s vaunted ground operation. But while undecided voters could make a difference in the few tightest states, as William Galston argues elsewhere on this website, turnout is the most critical outstanding question.
Although undecided voters represent 3 to 6 percent of the electorate in most swing states, they're unlikely to break overwhelmingly toward one candidate. Historically, candidates usually exceed their final share of the vote in the polls and the exits show late deciders breaking roughly evenly, at least in comparison to supposed and unsubstantiated rules about undecided voters breaking entirely for the challenger. In 2004, Bush actually performed better than the final polls, which showed Bush leading Kerry by just 1.5 points in 2004 with less than 49 percent, compared to Bush's eventual 2.4 point victory with 50.7 percent of the vote. And although the exit polls showed Kerry winning late deciders, there is a distinction between “undecided” in a final poll and deciding late in the exits. Many polls push undecided voters to enunciate their preference, and their un-polled counterparts don’t finally “make up their mind” until the final week, even though they knew their decision deep down and would have said so if they had been pushed hard enough.
In this election, the number of undecided voters is so small that there are only few states where a clear break would be sufficient to flip the outcome. In Wisconsin and Nevada, Obama already exceeds 49 percent, suggesting that undecided voters could only influence the outcome if Obama supporters turn out at lower rates than the polls anticipate. One state where Romney still retains a narrow path to victory through undecided voters is Ohio, where Obama holds a very slight lead of just 2.1 points in the RealClearPolitics average, 47.9 to 45.8. But if Romney won 55 percent of undecided voters and one percentage point vote for a third party candidate, Obama would still win Ohio by a 1.6-point margin, 50.3 to 48.7. Romney would need nearly 70 percent of undecided voters to carry the state—an exceptional performance. Colorado and Virginia are the two states close enough for undecided voters to more realistically make a difference, but, even there, turnout is a more critical question.
According to national polls, Obama is performing four points better among registered voters than likely voters. That’s well above the more typical 1 or 2 point gap and the main culprit appears to be strong Republican enthusiasm combined with low enthusiasm among young, Latino, and Democratic-leaning independent voters. Since Obama’s coalition is unusually dependent on low-frequency voters, Obama has more to gain from a strong turnout operation than previous candidates. Although it’s unclear whether Obama’s vaunted ground operation can rejuvenate turnout among infrequent Obama '08 voters, the difference between a modest and high turnout among young and minority Obama supporters could easily decide the election. And it's not just that turnout is important, it's that Obama's larger advantage among registered voters makes it an open question whether Obama could actually lose if minority and youth turnout rates approach '08 levels, even if undecided voters broke in Romney's direction.