We've been saying for the last year that Democrat polling firm Public Policy Polling (PPP) has a sampling problem — they oversample Democrats and undersample Republicans in their Colorado polls, delivering skewed results. Now The New York Times polling expert, Nate Silver, finds that PPP, on average, has a "house effect" that leans to the Democrats by 3.1 points. Rasmussen, by comparison, only shows a 1.3 point Republican lean.
Here's the NYT's finding, visually:
This means PPP's results are showing Obama doing three points better than the NYT's "consensus of surveys." Without diving into the statistical weeds, what this basically does is create results that paint a rosier reality for Obama and Democrats than actually exists according to a roundup of the best publicly available polling.
The Grey Lady endorsement of a point we've been making for some time now is especially painful for PPP, who in their April Colorado poll ended the press release with this sentence:
PPP is a Democratic polling company, but polling expert Nate Silver of the New York Times found that its surveys in 2010 actually exhibited a slight bias toward Republican candidates.
No one noted it at the time, but this sentence was conspicuously absent from their last press release on Colorado poll results. Maybe PPP should update their press release template with this:
PPP is a Democratic polling company, and polling expert Nate Silver of the New York Times found that its surveys in 2012 have exhibited a strong bias toward Democratic candidates.
This is a fact we hope reporters covering future PPP polls note in their coverage, as PPP's previous New York Times spin has ended up in print (cough*Tim Hoover*cough).
(Graph Credit: FiveThirtyEight)