The trend is registrations (in states where voters register by party) tend to lag AND CONFIRM the trend in “affiliation.” Pollsters usually use the subjective concept of affiliation.
2004 Exit Poll 37 Dems 41 Reps
2006 Exit Poll 36 “ 39 “
2008 Exit Poll 37 “ 34 “
2010 Exit Poll 36 “ 36 “
According to the Exit Polls in Florida, affiliation shifted from the Pubbies to the Demos from ‘04, and shifted to even in ‘10.
The media polls are using 2008 turn-out model, when there is reason to think the Republicans will have an advantage in affiliation this year.
the Republican # .93% enthusiasm (probably low, but: 3,881
the indie # at 66 R 34 D (I know, this is questionable, but no one doubt Romney has a big advantage in indies for 1699).
That left R 5.5 m to D 5.356m but latest polls of Jews in FL say that the shifts here could cost the Dems 50,000 votes, so maybe the final ration will look like this: Romney 5.5m to Obama 5.29 M
I think that's even outside the automatic recount.
Oh, and this assumes that NO Dems will vote for Romney-—which I don’t think is the case. I think we’ll see 5% to maybe 7% Dem voting R. But likewise I assume NO Republicans will vote for Obama, and I think that’s absolutely true.