The Recurring--and Misleading--Focus on Party Identification [Gallup says we are full of crap]

Gallup ^ | 9/27/12 | Gallup's Editor in Chief

[SoFloFreeper]

The discussion of the party identification composition of poll samples comes up in every presidential election with which I've been involved. Interested observers often opine that when a given poll shows that Candidate X is ahead, it cannot be correct because there is a higher percentage of voters who identify with Candidate X’s party in the sample than there should be, based on comparison to some previous standard.

There are several reasons why this is a faulty approach to evaluating a poll's results.

Party identification is basically an attitudinal variable, not a stable population parameter...

Party identification changes as political tides change.

[Intolerant in NJ]

So much BS - party identification should be determined independently of the voting breakdown, and can be done dynamically by registration totals kept by the states or separate polls (Rasmussen says party identification for Republicans grew in both July and August and now is at its highest point ever at about 37% percent) - voting samples are then fit to the party identification data as adjusted for the expected turnout - contaminating party makeup with voter preferences is a mistake even a first year statistician would not make, and tells more about the methodolgocal sloppiness of Gallup than the state of the election.......

[montanajoe]

“I remember when in 2004 liberals where whining about polls oversampling republicans, this discussion happens every election”

True, its human nature, it reminds me of fans of a sports team blaming the refs for a loss and folks point out that the game should never have been so close a decision of a ref could decide the outcome.

This race should not be close but the GOPe insisted on nominating a liberal, its not the polls fault the the politician running.

[BfloGuy]

that is an extremely twisted logic here by Gallup and they could not explain anything at the end

You're right. A very, very unconvincing argument from Gallup.