Your analysis is a bit off. It’s not enough for O to win indies, if he loses more Ds and Rs outnumber Ds. Consider it this way.
O gets 50% in a 38% weighted D poll.
Assume he gets all the Ds, then hes got 12% or 50% of indies.
M gets 46% in a 36% weighted R polls
Assume he gets all the Rs, then hes got 10% or 42% of indies.
In a R 39/ D 35 / I 26 race, matching 2004, this would be
39+ 11 (.42x.26) = 50% McCain
35+13 = 48% Obama
Polls also show the ‘undecideds’ to be mostly white and not in NOVA, meaning a likely mccain/but-wont-state vote.
So that would be 52%/48% McCain, or a 4point margin.
I dont know what will happen, because there is fluidity in who is calling themselves R and D this year. I think that is screwing up polls perhaps moreso than ‘bradley effect’ (which imho is showing itself in the high undecideds).
If you can ‘fix’ the internals on this poll to go from narrow victory for Obama to the same for McCain, then it is telling you that THE RACE IS STILL TBD.
See #34....for late breaking news on Obama that could change some states.