Last week's debate might have starred Barack Obama and John McCain, but it wasn't really about them. Rather, it was about an odd and extremely powerful creature in American electoral politics: the Undecided Voter. It was the Undecided Voter whom Gallup asked to submit the questions. It was the Undecided Voter who filled the audience. It was the Undecided Voter who turned the dials controlling CBS' squiggly reaction lines and recorded his (or her) responses for CBS' postelection survey.
It's a bit odd that we give the Undecided Voter such a privileged place in American elections. Because from a civic standpoint, few creatures are as contemptible. This election has dominated every form of American news media for the better part of two years. Newspapers, magazines, networks, cable, radio, blogs, people on street corners with signs - it really has been rather hard to miss. Further, it pits two extremely different candidates against each other. Whether your metric is age, ideology, temperament, race, funding sources, health-care plans or Iraq strategies, it would be hard to imagine two men presenting a starker contrast.
But despite this, the Undecided Voter wakes up each morning and says, in effect, "I dunno." And the political system panders to him. Undecided voters are believed to be the decisive slice of the American electorate, so they get the debates and the ads and the focus groups (assuming, that is, that they live in a battleground state).
Political-science research on undecided voters remains a bit sparse, in part because it's hard to pin down who, exactly, they are. It's not a static population. At least initially, we're all undecided voters. More of us are undecided in May than in October.
And even now, when most of us have settled into our preferences, change remains possible. If Obama personally collars Osama bin Laden during a fact-finding trip in Afghanistan, a lot of die-hard Republicans will switch their vote. Similarly, if McCain emerges from one of his 12 garages and explains that he was doing some tinkering and came up with a cheap, clean, renewable, plentiful energy source, the polls will swing.
Undecided voters fall, by definition, into the larger category of swing voters - meaning that they could end up voting for either party. We do know a few things about swing voters generally. Jeffrey Jones of the Gallup Organization, for instance, reported after the 2004 election that they tended to be less educated, more rural and somewhat older than most voters.
Swing voters can be measured in various ways, but there are no solid numbers on undecided voters - in part because the numbers change with every election and, within every election, with every successive month and event and even poll. Right now, if you look at the three main tracking polls - Gallup, Hotline and Rasmussen - they show that between 5 percent and 12 percent of the electorate is undecided.
Many of the people who claim to be undecided are not. Some don't want to admit their preference. In their paper, "Swing Voters? Hah!" political scientists Adam Clymer and Ken Winneg amassed substantial data suggesting that very few undecided voters are truly indecisive. Examining the 2004 election, Clymer and Winneg found that even the most hard-core of undecided voters were fairly predictable.
They asked the 4 percent of their sample that claimed to be undecided to rate the two candidates in early October. When they went back to the same people after the election, more than 80 percent in fact had voted for whichever candidate they'd rated most highly a month earlier.
But campaigns need something to do in September and October. Most of the electorate has chosen a side, and the small sliver that claims still to be puzzling over the pronunciation of the Democrat's last name could prove decisive.
Or could it? A provocative paper from James Campbell, a political scientist at the State University of New York, Buffalo, comes to a different conclusion. Examining nine presidential elections, Campbell compared the size of the swing vote (defined here as voters with weak leanings before the heat of the campaign) with the size of the nonswing vote. Swing voters are known to be a minority of the population, but it turns out that they're not a particularly decisive minority. "In only one of the nine elections, the 1976 race between Gerald Ford and Jimmy Carter," writes Campbell, "did the swing vote majority override an opposite majority among nonswing voters."
In other words, in eight of the last nine elections, the winner could have lost swing voters but won the race. In a second test, which examined voters who were undecided at a later point in the race, Campbell found that the last campaign in which they were decisive was 1960.
Campbell concludes by quoting Paul Lazarsfeld, a political scientist from the 1940s who posited that campaigns are essentially over before they have begun. The outcomes are structural - they are decided by events and party identification and satisfaction with the incumbent and other predictable indicators. Campaigns, he said, are "like the chemical bath which develops a photograph. The chemical influence is necessary to bring out the picture, but only the picture pre-structured on the plate can come out."
With all this evidence militating against a sharp focus on swing voters and undecided voters, the reader might ask why campaigns even bother. The answer is probably best given by Ruy Teixeira and William Mayer in the conclusion to their book, "The Swing Voter."
"What's the alternative?" They ask. "If many swing voters are, in fact, not very easy to persuade, committed voters are even more immovable." Campaigns now raise hundreds of millions of dollars. They can afford both ads aimed at swing voters and "get out the vote" operations meant to motivate base voters. And why begrudge them their efforts? The campaign is long, and people need to keep busy.
Klein is an associate editor at the American Prospect. He blogs at EzraKlein.com.