|
I've run Excel models with dozens of permutations. The partisan split is more significant than undecideds. In fact more than three times as decisive, given the tiny number of undecideds. Even if we get a huge 2/3+ share of undecideds a 50/50 Dem-GOP split will nullify that and beyond.
You're right about '92 and that is troubling, since the dynamics of that election are more similar to 2004 than '96 or '00 were. I wasn't following political math closely until '96, so it's difficult to backcheck and backfit and understand what changed. It's weird because that was a high turnout election with hostility toward an unpopular and out of touch GOP incumbent, so you would think the Dem partisan edge would be equal or higher, not zero. I may research that more tonight.
My base model has us with a 2-2.5% partisan edge nationwide, adjusted in various states due to historical trends and my best guess among new registers. When I look at my projected statewide raw numbers, the high populus states with a huge Gore edge in 2000 are the ones that drop significantly -- California, New York and New Jersey for example. We have lost registration percentage in states like that compared to 2000, and especially compared to battleground state numbers from 2000 to 2004. So a national partisan change may not be completely representative. In California, for example, our registration and party ID numbers jumped after '92 with the Pete Wilson decisions and Hispanic upswing.
You ask a very significant question, truly vital to this result but seldom queried on DU. Has there been a party identification shift in the direction Gallup detects? Certainly their projections are wildly exaggerated. But the GOP taking control of so many state houses and local offices makes you wonder. Again, I would think a fever pitch presidential race would draw out our numbers in full force. A GOP shift is more likely to be felt in midterms and obscure elections. If forced to estimate, I tend to believe our edge will be less than '96 and '00, making independents even more critical. My friend who is very astute in politics has looked at state registration numbers and theorizes a 10% block of newly registered independents is really more Dem leaning than standard independents, but for whatever reason has chosen not to officially identify with us.
|