Here's Nate's latest musings on the Pew poll which set everyone's hair on fire:
A great poll for Romney, in perspective
As of today, the Nov. 6th projection gives Obama a 74.8% chance of victory. That's the same value as ... September 3rd. And if I recall correctly, most of us were feeling pretty good about this election in early September.
The evidence is that the Romney debate bounce is fading. There's a reason why they call them "bounces". Romney picked up 2.5%, more or less, from a quality execution of the Gish Gallop. That erased somewhat more than half of the President's lead. Aside from yesterday's Pew poll, it looks to be fading.
Nate is dismissive of those who criticize the Pew Poll because of the crosstabs showing a higher percentage of Republicans than they had seen in the past. Party ID is fluid, and one of the ways that an improvement in the polls manifests itself is that more people identify with the improving candidate's party. Also, Rasmussen adjusts poll results to match a pre-determined party ID distribution. That ought to seal the deal.
The post-debate state polls show a mixed bag. Aside from a +7 change for Romney in Michigan, the results look better for Obama than they do for Romney. This suggests to me that perhaps the months of advertising in the swing states have sealed Romney's image in those states in a way that one good debate cannot undo. Romney narrowing the gap in California and widening it in Utah doesn't mean diddly-squat (a precise mathematical quantity representing a milli-tad bit) for the Electoral College.
Conclusion: we've lost a month of gains, Romney has some momentum, but the fundamental picture is what it was a month ago. We're winning, but we can't let up.