There’s this typical mode of election analysis where you see that voters who said “change” is important voted for Obama, and voters who said “experience” is important voted for Clinton, so then Clinton won because more people were interested in experience, or else Clinton lost in another state because more people were interested in change. I can’t quite prove it, but I’m pretty sure this is wrong. It seems to me that people are drawn to one candidate or another for a murky series of reasons, and then they come to pick up their favorite candidate’s themes. So people who like Clinton develop an appreciation for the importance of experience, whereas Obama supporters decide that the Iraq War is the most important issue.
Larry Bartels has a paper about this phenomenon that I recommend to one and all. Certainly, I think pundits ought to do a better job of at least keeping the possibility of this sort of thing happening. In general election terms, for example, Andrew Gelman has observed that the behavior of most voters is pretty consistent and predictable from year-to-year. But the candidates are always changing. So when George W. Bush is the nominee, the sort of people inclined to vote Republican claim that Bush’s personal characteristics are really crucial, but when John McCain becomes the nominee the sort of people inclined to vote Republican like his personal characteristics.