Inferences

Patrick Healy has an excellent article in the Times making the point that it's illegitimate to make inferences of the form "A lost State X in the primary, therefore A will lose State X in a general election" or "A lost Demographic Y in the primary, therefore A will lose Demographic Y in a general election." If Clinton loses Demographic Y that could be because their preference is Obama > Clinton > McCain or it could be that they think Obama > McCain > Clinton and their behavior in the Obama/Clinton race doesn't give us any evidence.

The best evidence we do have to test these claims is provided by the early general election polling matchups, we can at least illustrate broad trends. According to Gallup, Clinton and Obama are both tied with McCain:

In head-to-head matchups against presumptive Republican nominee McCain, Clinton and Obama perform almost exactly the same. In Gallup's latest tracking of the general election, based on interviewing conducted April 18-22, McCain has a one-point lead over both Clinton and Obama. In the April 18-20 USA Today/Gallup poll, both Clinton and Obama were slightly -- but almost identically -- ahead of McCain among likely voters. In neither instance is there any meaningful difference in how the two candidates stack up against McCain.

Obviously, there's no way Clinton could be tied with McCain without picking up the lion's share of Obama supporters, and there's no way Obama could be tied with McCain without picking up the lion's share of Clinton supporters. Basically, there's nothing to see here.