Sam, why would Ras's monthly national polling be better than his daily national polling or whatever. What makes up his monthly national polls? How is it done differently than his shorter term national polls? I was never aware of this distinction before. In any event, the same issues of bias arise, lack of cell phone contact, etc.
Well, the first point that I should make is that Rasmussen's partisan trends polling
is not done in the same sample in which he does his other political polling.
Basically, Rasmussen does two daily samples - one for the daily tracking poll (the 500 LV) and one for the Consumer Index (the 500 adults). I have to assume that the likely voter sample is done differently than the adults sample (but have no proof on that). Anyway, the party ID question is asked in the Consumer Index sample in order to insulate that result from political questions (which would skew it).
I agree with the way Rasmussen does this, and the proof is in the pudding - go look back in 2006/2008/2010 and the three-month average of Dem/GOP/Ind in Rasmussen is very much in line with the exit polls.
As for daily/weekly/monthly distinctions, there are none except that the Presidential approval poll is 1500 LV, any weekly polls are 3500 LV (like say Generic Ballot) and any monthly polls are close to 15000 LV. I'm just noting that the internals of October's monthly poll, if not the poll of the week before look a hell of a lot like the internals of the 2010 exit poll, if not skewed a tad Democratic because of turnout issues. Which makes why the Generic Ballot was off very interesting, if not confusing or maybe a conspiracy?
Rasmussen's state polling, OTOH, is only 500/750 voters on
one day. Ever since he went to this in 2006, I have questioned it because one-day samples tend to be much less reliable (go ask Vorlon) for a whole host of reasons. Good support in this theory would be found in this election in that only two Rasmussen polls were conducted over two days - North Dakota and Indiana were quite accurate this election (even though Indiana sucked in 2008 (the only poll then, undoubtedly done because of time zone splits).
Also, the terms you use are inaccurate IMO. Rasmussen is not "biased", rather Rasmussen, over the last three cycles, has had a "clear Republican bias" in its state polling. Cell phone contact may be an issue, it may not - jury's still way out there. I should add that West Coast polling this year was way off because turnout was quite high among Dems, especially minorities.
