Rasmussen is using a D+5 model, and here's the math.
In a prior thread, I reverse engineered the Ras breakdown percentages to 34.4%R / 39.3%D / 26.6% I (D+5 gap).
Since there's some controversy on whether Ras is using 3D or 5D model, I thought I'd show the math. Here's profit_guy's internals (thanks, BTW!) as of Sunday:
Confirming these calculations are easy enough by picking a few party breakdown crosstabs and seeing if they match the answer. I picked the following three:
Total Approve (50%) = 0.14R 0.87D 0.42I
Wrong Direction (57%) = 0.89R 0.25D 0.62I
Romney Leaners (49%) = 0.89R 0.10D 0.54I
Here's the math:
Total Appove = 0.14*0.344 + 0.87*0.393 + 0.42*0.266 = 0.5018 (50%)
Wrong Direction = 0.89*0.344 + 0.25*0.393 + 0.62*0.266 = 0.5693 (57%)
Romney Leaners = 0.89*0.344 + 0.10*0.393 + 0.54*0.266 = 0.4891 (49%)
So, I think my original calculations of
Rasmussen using D+5 gap are correct after all. Furthermore, note that Rasmussen is predicting a D turnout equal to or slightly greater than 2008, where exit polls showed that 39% of the voters self-identified as Ds. Take what you will.
BTW, note the IDs won’t add up to exactly 100. This is a result of Rasmussen’s roundings not giving enough precision needed, but the ratios should be very very close to what he’s really using. Safe to assume D+5.