You are viewing an obsolete version of the DU website which is no longer supported by the Administrators. Visit The New DU.
Democratic Underground Latest Greatest Lobby Journals Search Options Help Login

Reply #15: Thanks Mike. [View All]

Printer-friendly format Printer-friendly format
Printer-friendly format Email this thread to a friend
Printer-friendly format Bookmark this thread
This topic is archived.
Home » Discuss » Topic Forums » Election Reform Donate to DU
Febble Donating Member (1000+ posts) Send PM | Profile | Ignore Wed May-04-05 02:03 PM
Response to Reply #14
15. Thanks Mike.
I probably can't address your questions though as I am in no way a polling expert, and do not know much about how the projections are made.

However, I am fairly sure that the WPE is simply a quality control measure - it isn't used to make the projections. Although voters are randomly sampled from precincts, and precincts sampled from states, voters are nonetheless treated as though randomly selected from states, and a factor is applied to the standard error in order to allow for the "clustered sampling" (the notorious "design effect"), otherwise, of course the variance will be too low, and the MoE too small.

The problem I identified is that the WPE is a very poor dependent measure for quality control because of its confound with margin. I have devised a transform (actually simple algebra) to convert the WPE into a measure that is simply the log of the numerator of the ratio between the response rate for one group of voters and the response rate of the other. So it is independent of vote margin, and also independent of overall response rates. This would seem to be a far better variable on which to regress factors that are hypothesised to be associated with variance in bias (bias, to make it clear, that can be equally bias in the polls or bias in the count - if the ratio between the response rates is not equal to 1 it could be because either votes didn't exist or voters were phantoms - or because one group of voters was more ready to respond than the other).

Anyway back to the point - if my transform is applied, there should be no confound, and the factors associated with bias more validly identified using the GLM.
Printer Friendly | Permalink |  | Top

Home » Discuss » Topic Forums » Election Reform Donate to DU

Powered by DCForum+ Version 1.1 Copyright 1997-2002
Software has been extensively modified by the DU administrators

Important Notices: By participating on this discussion board, visitors agree to abide by the rules outlined on our Rules page. Messages posted on the Democratic Underground Discussion Forums are the opinions of the individuals who post them, and do not necessarily represent the opinions of Democratic Underground, LLC.

Home  |  Discussion Forums  |  Journals |  Store  |  Donate

About DU  |  Contact Us  |  Privacy Policy

Got a message for Democratic Underground? Click here to send us a message.

© 2001 - 2011 Democratic Underground, LLC