Its pretty clear that these national polls showing Obama +7 to 9 are no longer outlyers as 4 straight surveys have show this - cnn, fox, ap, pew. But its also clear the gains are (almost) solely due to increased dem party ID. No one except the pollsters knows what changed, but something gave them reason to increase tbe Dem Id. However none of the tracking polls ( ras, gallup, kos, yougov) - taken oven the exact same time and presumably among the same sample - show any change. The average of the 4 is a dead heat. One of these groups is wrong, we just dont know which one. It does seem quite bizarre that there iz such a desparity btw the two groups.
Party ID is notoriously volatile - like ideology, it's one of two major characteristics that respondents volunteer, but unlike ideology it tends to fluctuate pretty wildly. Off the top of my head, 2008 saw a hue nationwide shift in party ID, but ideology stayed
exactly the same, about 20-45-35.
Note that party ID questions rarely, if ever, ask about registration - the most common phrasing is "Do you consider yourself a Republican, Democrat, or independent?" And as we see with the shy partisan phenomenon (there are many fewer true independents than you think), self-professed partisanship is unstable. If people start to lean towards one candidate or another, they become more likely to identify with his or her party.
In short, weighting by party ID is putting the cart before the horse. That's why (most) professionals don't do it.