Taking just those states and adding them up from 2008 I come up with 7.08% advantage for Obama, nearly the same as the 7.27% national advantage.
As I understand. Ed Goes one of the two founders of the poll, said last night there model they run, taking poll results and other metrics shows Romney winning 52-47. They are generally one of the most accurate groups out there.
If you go to the full internals, by region, you'll see that the full poll had respondents in numerous non-battleground states (like Texas and California). The 10 competitive states is a subset (and one they have Obama leading in, though I think there's a serious margin of error problem there due to the much smaller sample size).