Ok let’s go with those 2008 numbers — what does “overperforming” mean, does it mean they returned a higher percentage of AB’s than R’s did, or that they returned them earlier, or that more D’s voted via AB than R’s, or what.
Knowing nothing about NC registration except what I’ve seen on this thread, which is almost certainly why I see a dozen variables instead of just one or two, I don’t see any conclusion that can’t be refuted.
Over and underperforming based on 2008 turnout.
Guess you need a primer on polling, and why all the polling that is out there is (if these numbers are accurate) badly off:
All polls assume a D/R split based on some turnout model. So far, not only are all the polls except Rasmussen using the 2008 model, some are actually INCREASING the turnout from 2008. This is just ludicrous and unconscionable. Nevertheless, if, say, a county went for Obama by 60-40, then the model used for a poll would consist of interviewing 60 Ds and 40 Rs-—and they would have a split for Indies, but let’s table that for now.
The dynamics of OH are that if the Rs turn out and vote R, Rs win. No exceptions. In 2008, Rs turned out and 25% voted Dem. I don’t know one analyst, even the most lib, who thinks that this time around Romney won’t get almost all Rs and Obama will get almost all Ds.
So when we see a county that went 60-40 for Obama and it now has absentee early splits coming in at 66-33 REPUBLICAN (say, Hamilton County) the Dems are in huge trouble.
You MIGHT say, “Well, x county is an outlier.” Yes, that would be valid. But when it’s across the board, even in the very Dem heavy counties (Obama will be down 70,000 votes in Cuyahoa, for ex.-—count on it), then barring something really unusual, Obama is screwed in OH.
Based on the heavy “overperformance” of Rs in NC, he’s screwed there too. If we find numbers are similar in VA and FL, then it’s going to be an early night on Nov. 6.