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
Google

Reply #54: Thanks - another possibility [View All]

Printer-friendly format Printer-friendly format
Printer-friendly format Email this thread to a friend
Printer-friendly format Bookmark this thread
Home » Discuss » Topic Forums » Election Reform Donate to DU
Febble Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Feb-10-07 09:48 AM
Response to Reply #52
54. Thanks - another possibility
worth bearing in mind is that the weights themselves (the ones that gave the Gore-Bush shares in the "Final" cross tabs" may not have been actually correct, even though they summed to a Bush win.

The assumption behind the reweighting is that the poll is not a random sample, i.e. that there are people "missing" from the poll. If the raw poll numbers appear to have too few Bush voters, then reported Bush voters will be up-weighted in the poll to match the incoming vote. TIA and others (including me, and of course you) have rightly questioned the assumption that the incoming vote was correct. But if we hypothesise for a moment that the vote count was correct, and that there were indeed groups of voters who were systematically undersampled, thus skewing cross-tabulations, then one possibility is that the group of "missing" voters was more heavily populated with Bush-voting 2000DNVs than the actual respondent sample. This would not only tend to skew the sample Kerry-wise, but would also skew the sample voted-in-2000 wise. And as the past vote question was not one from which weights were derived (because it was only on 25% of questionnaires anyway, quite apart from the fact that there would be no data with which to inform the weights) then a blanket upweighting of 2004 Bush-voting respondents, regardless of whether or not they were 2000DNVs, Gore voters, or Bush 2000 voters, is not going to produce sensible results from that particular crosstab, even if we assume that 2000 vote recall was perfect.

It really all boils down to how realistic one's views are regarding the representativeness of the NEP sample. I see no reason to assume that it was random, and plenty of reason to give credence to the possiblity - actually the probability - that the sample itself systematically under-sampled voters in various categories tapped by the questionnaire. It would not surprise me, for example, if infrequent voters ("reluctant voters"?) were also "reluctant" exit poll responders. But the fundamental point is that there is no reason, especially in a poll with a 53% response rate, to assume that the responders were "drawn from the same population" as the non-responders. TIA's entire analysis assumes that this assumption is valid, and it isn't.

And as it is difficult to make his inference consistent with my own swing-shift findings, that raises considerably the probability that his assumption is false.
Printer Friendly | Permalink | Reply | Top
 

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

Powered by DCForum+ Version 1.1 Copyright 1997-2002 DCScripts.com
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