It probably says more about the changing nature of PID from a stable long term force to a less stable more short term affiliation that is more closely tied to candidate evaluation (in a reciprocal fashion) than ever before.

Surely part of this is the full conversion of the conservative Southerner from being a Democrat who votes Republican in presidential contests to a Republican who vote Republican.

My working theory is the rise in self-identified independents coupled with the lack of a real “draw” third-party candidate in 2000 or 2004 (sorry, kids, Ralph and Pat don’t count) who could peel off solid partisans.

However, without sitting down with the NES data, my sense is that if you throw in the independent leaners you won’t get the 90+% predictive accuracy that Barry Burden reports (putting aside Larry Bartels’ independent-leaners-are-really-partisans argument). Data-free speculation, always fun for the holidays!

I didn’t realize that he hadn’t included leaners (DRTFA). But I seem to recall a good bit of ink being spilled in the past demonstrating that Leaners are often more consistently partisan in their voting than self IDed partisans. If this has changed, it still goes to my argument that the nature of partisanship has changed.