TPMDC reported the poll:
From the pollster's analysis: "14% of those who say they voted for Barack Obama in the 2008 Presidential election say they will vote for Republican McDonnell in the gubernatorial contest; 9% of John McCain voters say they will cross over and vote for Democrat Deeds."
I did the math. Barack Obama won Virginia 53%-46% last November.
If Deeds is winning Dems 86-14 and losing the GOP 9-91, then I most certainly don't get McDonnell leading 55-40. Rather, I have it much closer.
Here's the basic math I came up with:
86% of 53% plus 9% of 46% = 50.2%
14% of 53% plus 91% of 46% = 49.8%
Throw in the 5% undecideds, and we should be looking at 48%-47% in Deeds' favor.
Now, are they going to try to tell me that Obama's voters are going to turnout at such low levels compared to McCain/conservative voters while new voters are going to be so Republican-friendly that a 1-point advantage is going to turn into a 15-point deficit?
I smell something fishy there. I think this poll is not up to snuff. Not saying this is deliberate, but I just don't trust their turnout model and the sampling. I'm willing to bet this is an outlier.