I have drilled down much further with my model, assigning each competitive house race a probability based on things like polls, the make up of districts, etc. Interestingly, I'm still about where I was. The projection is now a pick up of 68 seats for the Republicans.
There are still many unknowns. For instance, many races haven't been polled at all, and who knows how many might be competitive? If Scott Brown's election in Massachusetts hadn't been polled, would you have thought he had a chance? I count at least 25 races of obvious interest where there has been no polling. One of these (Indiana 8th) is a Republican lay up. Of the others, my probability weighting suggests the Republicans will pick up 4. This may well be conservative.
What we do know is that there are currently 48 Democrat seats where the Republican has a polling lead. The Dems in these seats are dead men walkin'. The model says 44 will lose. There are 42 more seats where the Democrat has a lead but is polling under 50%. These are where the real action is if you're a betting man. I have the Republicans picking up a probability-weighted 19 of these. If the Gallup turnout models turn out to be correct, this number could be low.
No comments:
Post a Comment