Reply #27: yes, I overstated that

[OnTheOtherHand]

and I didn't mean to imply that Bush would need every point of that 43/37 retrospective margin in order to win the popular vote.

It might have been better not even to mention the exit polls, and just to frame the problem as: if Bush2K and Gore2K voters turn out in roughly equal proportions, and they "defect" to the other party at roughly the same small rates, and Other2K and DidNotVote2K voters break for Kerry, then Kerry wins the popular vote, period. Which is apparently what most people on DU believe happened. (This view does not appear to be widely held among political scientists, and I find this difference in perspective interesting in itself -- but that would be a different thread.)

If I were going to sort out your last point about party id shift, I would look closely at two things. First, I would expect more Dems to vote for Bush in the south than elsewhere, and I wonder whether that affects your analysis. Second, there is a literature on whether exit poll calls affect West Coast turnout, and I think it is pretty dismissive. Even if that literature holds up, that wouldn't necessarily refute your idea in 2004.

I don't think I will comment separately on your other post, but I think it proceeds in a reasonable direction. It's sort of a hybrid scenario (maybe Bush won the popular vote -- or didn't, hard to say -- but stole the election anyway in one or more battleground states). I don't think we are ready for hybrid scenarios on this thread. At this point I will be amazed if we can make it through one scenario. But the investigations in battleground states should continue. I'm not convinced that Bush stole any states (which I mean in the sense of actually stealing or inventing votes), but I'm certainly not in a position to whistle the all-clear.