« Douthat on Romney's Big Religion Speech | Main | In Hindsight »


Contaminated Sample

A staff post here yesterday mentioned an Iowa State University poll of likely Iowa caucus-goers that showed a relatively big lead for Hillary Clinton, along with a surprisingly poor third-place standing for Barack Obama. We noted that the polling data was a bit stale. But now, via Chris Bowers, we learn that the ISU poll also had a strange sample bias: self-identified independents were excluded. Since these voters are expected to provide about one-fifth of the Democratic Caucus participants, this is a pretty important distortion.

To be clear, no amount of indie love for Obama would enable him to overcome HRC's 11-point lead in this poll, but it's a good reminder that all polls are not created equal, even if their headlines get reported that way.