Matt Yglesias makes a great point in response to this Marc Ambinder post:

Certainly one easy opportunity Clinton has to expand her coalition is that versus John McCain she would pick up all these African-American Obama supporters. But conversely, I don't see Hillary Clinton's feminist supporters suddenly deciding that they want to see John Paul Stevens replaced with an abortion-banner. The theory here seems to be that Clinton's strength among white working class Democratic party loyalists will translate into strength among white working class non-loyalists. But there's no evidence for this theory -- both Al Gore and John Kerry formed "beer track" primary coalitions and then went on to perform terrible among white working class voters overall.

It's easy to fall into the trap of thinking, as the Clinton campaign argues, that Hillary is a better general election candidate because the groups she's winning in the primary more closely resemble the groups Democrats need in November. But, even if you accept that premise for the sake of argument*, the relevant question isn't: Which demographic groups is each candidate winning the primary? The relevant question is: Which candidate is most likely to win the general-election version of their primary coalition (assuming they more or less hang on to the Democratic supporters of their primary opponent)?

In concrete terms, Hillary's primary coalition consists of working-class people, seniors, and women. Obama's consists of African-Americans, younger voters, and affluent/educated voters. Set aside African-Americans, who aren't really a swing group. The question then becomes: 1.) How likely is Hillary to win non-Democratic working-class people, non-Democratic seniors, and non-Democratic women? 2.) How likely is Obama to win non-Democratic young people and non-Democratic affluent/ educated people? With the possible exception of women, I'd say the likelihood of 2.) is greater than the likelihood of 1.).

You can obviously disagree with me. But you should understand that, if you think Hillary is more electable, you're basically saying that likelihood 1.) is greater than 2.). (Unless, of course, you think Obama will suffer big defections among working-class Democrats, Democratic women, and Democratic seniors if he's the nominee. But that's a different argument, and I'm skeptical of it for the reasons Matt lays out.)

*There are really two broad swing groups: one working-class, the other affluent. In principle, you could win the general by winning one or the other, or some combination of the two.  

--Noam Scheiber