I've caught a fair bit of flak on my earlier post on Rasmussen's poll showing Obama upside down on his approval ratings (same result today, by the way).  I thought it was worth fleshing out my thoughts a little bit more.

The goal of my post wasn't to criticize Rasmussen's methodology or polls.  I think experience demonstrates that they are quite good and usually end up on the nose.  That said I think it was important to at least acknowledge that Rasmussen is the only pollster showing Obama upside down in the polls.

This discrepancy is what the remainder of the post is dedicated to exploring:  Why Rasmussen finds Obama upside down where others don't.  He is an outlier, but there are a couple of very good reasons that he's an outlier.  The first is that he samples likely voters, when everyone else is polling registered voters or adults.  This typically results in findings a few points better for Republicans.

There is a lively debate among pollsters about what the best "screening question" for likely voters is, and when the best time to implement that screen is.  Some think you should just ask if voters have voted in the last election, others believe that you should also ask if you plan to vote in the next election.  Obviously, attitudes toward voting in the next election are unreliable at this point, so if that is how you screen, you probably don't want to put your likely voter screen into place yet.  And regardless, there's a question of whether, this far out from an election there's any reason to divide our citizenry into likely and unlikely voters, since you're not trying to say how an election is looking as of this minute.  As a personal matter, I tend to believe it is early to screen likely voters (I go back and forth on this), but I don't think screening likely voters at this point is a bad thing, either.

The bottom line is, Rasmussen doesn't really have a skew here.  He's just testing something different than other pollsters are testing, and reasonable minds can disagree over whether its too early to test for what he's polling just yet.

The other difference for Rasmussen is his questioning.  Again, he allows for more nuanced answers, which can change the likelihood of declaring one's self "undecided," and might even change whether you give an approve or disapprove answer.  Again, we can debate the relative merits of this questioning, but you have to know its there, and that its probably one of the reasons that Rasmussen has fewer undecideds than others.

Ultimately, this exonerates Rasmussen, rather than attacks him.  If Rasmussen were polling registered voters or adults and asked more standardized questions, we'd have to wonder if his methodology was hinky.   But he isn't, and his methodology isn't.  That was the real point of my post.

--------------------------------------------
Follow the RCP Blog on Twitter.
Become a fan of RCP on Facebook.
--------------------------------------------