We’ve emphasized this enough times on this blog,but it bears repeating: lots of research shows that people cannot accurately identify the reasons for their actions. Here is Lee’s “old post”:https://themonkeycage.org/2008/05/post_84.html on the subject.
Remember this when you read analysis of exit polls and other polls where respondents were asked how important such-and-such factor was in their choice, or whether their vote was a vote for or against Obama or health care of whatever.
A better way to know whether any factor is associated with voters’ choices is via statistical modeling on raw polling data. You can’t do this with the toplines from an exit poll. And cross-tabulations of any single factor and vote choice doesn’t tell us whether that factor mattered, controlling for other things.