UNC’s Jim Stimson on the evaporating GOP bounce and the state of the race (posted on Sunday):

I wrote in Tides of Consent that convention “bounces” were permanent effects, persisting through election day unless counteracted by some later events. The Palin bounce of 2008 appears to contradict this conclusion. It appears that Obama was ahead by about two points going into the Democratic Convention. A 4 point bouce put him at 6. Then the more dramatic Republican/Palin bounce appeared to put McCain in the lead by appout two points, a net 8 point shift. Now, more than two weeks later, it appears that about half of the Palin bounce was transient and the other half almost exactly offsets the Democratic bounce of four points. That leaves the race exactly where it began in late August, with a small Obama lead. But a post-convention lead is much more solid than a pre-covention lead, because it represents mainly real voting intentions, as opposed to the much fluffier stuff of pre-convention polls, where large portions of the electorate have yet to think about the contest. Thus, after a period of unremitting second-guessing and hand wringing, it appears that the Obama campaign is the net beneficiary of the convention season, if not by much.

Stimson’s poll-filtering algorithm (Excel spreadsheet) puts Obama at roughly 51.5% of the two-party vote, which is back in line with the fundamentals.