New DailyKos/Research 2000 tracking poll

[pepper11]

No way there is a 10 point disparity in party ID.  Perhaps there is a 7-9 point margin in the 'generic D vs R ballot for congress' but actual party ID is more like Dem + 4-6.

[pepper11]

How convienent that the night a Obama +3 sample rolls off a new one comes on. Couldn't be any bias in there could it?

I don't think you can call that bias. But when an Obama + 3 replaces an Obama +3 you would expect there to be no change in the 3 day average.  Instead Obama somehow gained 2 points in the average - a result that would require a + 9 Obama day given the +3 falloff. Sounds like fuzzy math.