I think that we are missing something here. The D+7 is plainly wrong unless you assume that the high Indy number includes many who would identify as Republican tomorrow in final party ID based upon how they vote. This would bring the Indy number down to a reasonable level and close the gap between D and R. It also is consistent with Romney being up big with the Indies. Unfortunately, the top line result thus would not have to change to produce a more reasonable looking party split. In other words, I am no longer convinced that the exclusive focus on the D+ number is enough to rebut all polls where that number appears out of whack. There may be other things to consider.
I am afraid that the tide of support that Romney recently enjoyed is going out and we will not like what tomorrow brings. I hope that I am wrong.
Let’s just hope that Ras and Gallup are right and that Romney pulls this out.
You make no sense to me. If the Dems are oversampled (relative to Rep) because (as you appear to say) the Independents who favor Romney are actually self identified Republicans then there is no point in polling at all.
These pollsters are using the respondents own self id’s Party affiliation in their results, If out of 100 people taking the survey 38% say they are Dems, 31% say they are Rep and 31% say they are Ind then that is the turnout they predict will occur. It doesn’t appear to make a difference if they must contact 100 people to get even 5 to take the survey and there os no line drawn to cut off where common sense leaves and absurd takes over. By that I mean if of the 100 taking the survey 44% id’d as Dems they would use that number while in another poll it would be D or R + or - absolutely anything, thereby making the polls worthless.
Disregard most of what I said in my just posted response. I understand what you are saying is if we take the Ind lead of 6 for Romney and apply it to the Rep turnout (making Ind even for both candidates) we would still have Obama with a tiny lead albeit within the MOE and with a small sample.