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 #18: still some problems here [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
OnTheOtherHand Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Oct-04-08 07:56 PM
Response to Reply #17
18. still some problems here
Edited on Sat Oct-04-08 07:57 PM by OnTheOtherHand
While an exit poll is intended to be a random sample, as a practical matter, the interviewers often cannot choose at random from all voters at a polling place. Even if they could, the voters retain discretion to participate or not to participate. I would never propose to "divorce exit polling from statistical science," but I do insist that random sampling error is only one source of error.

Steven Freeman is not a statistician. Or, if you prefer, he is no more a statistician than I am. His Ph.D. is in organization studies.

I doubt Freeman ever claimed that there is "only 1 chance in 250 MILLION that the 3 main exit polls (OH, FL, PA)... could all be wrong." Although I disagree with Freeman about many things, I'm pretty sure you are misstating his view. What Freeman said was that there was a very small chance that the discrepancies "could have been due to chance or random error." (See here at page 13 -- this is a revised edition with somewhat different odds.) The probability that the polls "could all be wrong" can't be computed without heroic additional assumptions -- for instance, that non-response bias is impossible.

As I mentioned, the exit poll data indicated that Kerry won Pennsylvania by about 14 points. The final pre-election polls ranged from Kerry +4 (Zogby) to a tie (Quinnipiac). It seems rather selectively credulous to assume that the exit poll was correct.

The mean WPE reported for the 40 hand-count precincts in the sample was -2.2. It's true that that value is not statistically significantly different from 0, there being only 40 precincts to analyze. However, the mean for hand-count precincts isn't significantly different from that for precincts using other voting technologies in either urban areas or rural/small town areas. (Freeman complains that this is an unfair comparison, but the truth is, there just aren't enough data to draw a solid inference.)

I'm certainly not against SPC. I'm not even against trying to use exit poll data to support that effort, as long as we are accurate and honest about what we can infer from it. I advocate stringent post-election audits of hand-marked paper ballots, or failing that, voter-verifiable paper records.

(edited subject for clarity)
Printer Friendly | Permalink | Reply | 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