One of the ideas that refuses to die in American politics is that there are "independent voters". These are usually described as voters who are not Democrats or Republicans and who can't be counted on to support a party because they refuse to adopt a label to describe themselves.
The problem is that while people may self-describe as independent, in practice, they vote like partisans. In a scathing review of Linda Killian's new book, The Swing Vote: The Untapped Power of Independents, Ruy Teixeira lays out the evidence:
In 2008, according to the University of Michigan’s National Election Study (NES), 90 percent of independents who leaned Democratic voted for Obama, actually a higher level of support than among weak Democratic partisans (those who said they were “not very strong” Democrats), 84 percent of whom voted for Obama. Among Republican-leaning independents, a still-high 78 percent voted for McCain, compared to 88 percent support among weak Republican identifiers.
Evidently, these two groups are quite different animals. On the one hand, we have a group of “independents” who voted 90 percent for Barack Obama. Moreover, as Alan Abramowitz and others have shown, the policy views of Democratic-leaning independents look just like the policy views of Democratic identifiers. On the other, we have a group of “independents” who voted 78 percent for John McCain and have policy views that look just like Republican identifiers. Clearly it does tremendous violence to the data to lump these two disparate groups together and give them a label—“independents”—that implies they do not have partisan inclinations.
To be sure, there are some voters who are genuinely not aligned with either party and who are closer to what the platonic form of an independent voter might look like. According to Teixeira, such voters make up only 7% of the entire American electorate. Not enough to build a coalition on.