It certainly has been an educational week for those of us interesting in polling.
First came the illuminating disclosure that liberal polling company Public Policy Polling sometimes does not release the results of polls with unexpected results.
Then came this interesting article by TNR’s Nate Cohn, showing that PPP’s polling methodology includes a variety of questionable practices such as “random deletion” of respondents.
Then came Nate Silver’s statement that he will continue relying on PPP data even though he thinks PPP’s polls are crap.
Drew Linzer, a Ph.D. political scientist who knows his stuff, likened random deletion to “intuitive ‘shrinkage’ weighting.”
Sounds like PPP is using an intuitive "shrinkage" weighting, between the sample and known populatn. That's defensible. @ppppolls @Nate_Cohn
— Drew Linzer (@DrewLinzer) September 12, 2013
We’re no experts, but it sounds like he’s saying this is a method used to get the sample to mirror the population at large.
If that sounds like smoke and mirrors, don’t worry, because all pollsters do it:
@Nate_Cohn Have you ever worked at a polling firm? When you have small samples, there's judgment involved. There has to be.
— Drew Linzer (@DrewLinzer) September 12, 2013
Sorry but they all do: "No pollster weights to whatever electorate it chooses, undisclosed and subjective" @Nate_Cohn http://t.co/uMU3OQ9rjY
— Drew Linzer (@DrewLinzer) September 12, 2013
Well, that’s certainly reassuring.
Related:
Join the conversation as a VIP Member