The interesting thing is that they all line up with Gallup, who has never used Party identification as part of his model. So you would expect that to the extent that Party ID is the strongest determining factor on whether you vote for 0 or R/R they should show what Gallup shows when the partisan edge is factored out.
Now, here is the problem. On the one hand, in 2008 Dems had a +6 edge at the polls, so if you use the most recent presidential year as a way of modelling likely voters, we are in some trouble, and Gallup's method (minus 6) shows essentially an even race. On the other hand, in 2008 Republican Party self-identification was at historic lows, so D+6 is probably too high for 2012. Right now, Rasmussen shows Republican Party self-ID is a little over 1% HIGHER than Democrat self-ID, so it looks like Gallup might be right. The thing that disturbs me is that even with all that in mind, Rasmussen is calling this a dead even race. I sure don't like that...
My stomach hurts.
:(
Rasmussen is not using his own party affiliation breakdown model, where Rpublicans are now in the majority. He's using a hard D+3 for the national poll. (That's according to a freeper - I forget who - who has a Ras full subscription.) Maybe someone can confirm that.
As for Ras's state level polls - I don't know what model(s), if any, he's using. Anyone know?
I find it hard to believe Rasmussen will go down to the wire using D+3, seems to me that guarantees he'll be at least 5 points off.