This is also a little over half the lead that Parnell boasts in the PPP poll. Is it me or is Rasmussen becoming more Dem-friendly than PPP?
Rassumssen changes their LV model monthly(or more often as we get closer) and then I believe adjust state and national polls accordingly, which uses a bigger sample, but tends to mean outliers are more likely to show up the later in the month. At least thats what I infer from his reweighting party id each month, since no one knows exactly how he translates that into LV.
PPP on the other hand does it dynamically in each poll. My guess is that PPP's system gives a better picture at the moment, which is why they have done so well in actual races(ie. Florida Gov, MA Senate) with last minute polls, but it probably also makes them very vulnerable to news cycle based samples(hence the trouble with NY-23), and becoming out of date rapidly.
It also matters a bit on when it was conducted since both seem to be doing one-day robo-polling. From what we can tell from the daily tracking polls for both Ras and Gallup, Obama started moving up late last week, peaked over the weekend, and while still higher, had a less good day yesterday. The Alaska poll was conducted during the highest Ras daily sample for Obama. Today's Washington poll was conducted yesterday, when Obama had a sample that was probably quite a bit lower than previous two days.