Reply #105: What bothers me [View All]

is people's willingness - even people I admire enormously, like you - to assert that a piece of evidence - the exit polls - is indicative of fraud, and yet be unwilling to discuss the question as to whether it is. I know the subject bores you, but as I said downthread, getting stuff wrong can seriously warp a debate, and while it hasn't put you off your stride, in many ways I think it put an important sector of the Election Reform movement off its stride, which is why, although frankly, the subject bores me now, I feel honour bound to keep trying to present what I see as a more veridical account of the exit poll narrative.



In that case, why not the competence in their judgement as to why there was a greater than unusual "redshift"?



Including me. Which is why I have spent so much time on the darned things.



Of course they were aware. They actually have their own people at the "decision desks". The networks know how the exit polls work. That's why they don't "call" states until the "adjustments" have reached a given confidence level. Please read my DKos diary (linked in the previous post) for more information.



He did not "go into hiding". I read interviews with him myself. He may not have answered your email, but he did answer very many others. He also conducted an in depth investigation into the poll, and made the report publicly available for scrutiny in January 2004. It wasn't until April or so of 2005 that he contacted me, and that was because I'd written a piece criticising his own report (and the USCV analysis of it). He thought my criticism was valid, and hired me to redo the analyses.



But you disrespect his own opinion, formed before a single result was in, that the polls had an greater than usual pro-Democratic bias in several states, together with his own conclusions from his own post mortem of his own poll.



There is a great deal of evidence presented in that report. What you have done (reasonably, I did it myself) was to equate "participation rates" with "completion rates". The two are not the same thing. What the report does is give strong evidence that differential participation rates resulted from factors likely to make strict Nth voter protocol more difficult. But even if we consider differential response rates: it is perfectly possible that Kerry and Bush voters were equally likely to respond if approached, but that Kerry voters were more likely to be approached by/approach the interviewer, and the evidence supports this hypothesis. You also need to avoid the ecological fallacy (committed by Freeman, among others) in thinking that because total completion rates were not lower in precincts with stronger Bush support that relative completion rates were not lower among Bush voters for any given completion rate. If completion rates were lower in urban precincts, but lower still among Bush voters than among Kerry voters, because urban precincts tend to be more Democratic, you'd see lower completion rates in Democratic precincts, but greater response bias in those same precincts, because the relative completion rate differential was greater.



Well, not many survey researchers. Can you name one? (NB Freeman is not a "survey researcher").




Well, you missed the point. The point was not that they were proprietary - Mitofsky has always insisted that the data be publicly available, and indeed it was available for free download to anyone for over a year. What was NOT released were the precinct identifiers for the simple reason that given the extraordinary level of personal detail in the released data, in some instances, if precincts were identified, so could individuals, together with very personal information, including, inter alia sexual orientation. Nonetheless, "blurred" data was carefully prepared for Ohio, and made publicly available by ESE, and both the "Best Geo" and "composite" estimates made at close of poll (i.e. adjusted only by geographic stratification, and adjusted by pre-election expectations, respectively, and not by incoming returns) for each state was published in the Evaluation document.



And yet again the close of poll data was available for anyone to download, and many did. No cover-up, no conspiracy. And the fact that there was also a discrepancy in 2006, as there has been in every year since at least 1988, with a large one in 1992 is as consistent, if not more consistent, with an explanation lying in polling methodoligy as it is with an explanation lying in fraud. Note that the Dems won, and won in line with pre-election polling, even winning a couple of expected marginals by a hair.



Well, I have no quarrel with you over the unreliability and insecurity of electronic voting, nor with your case that malice has been aforethought. What I take issue with, as I said previously, is concluding that because A and B both suggest C, that A and B necessarily corroborate each other. IMO, they don't. What the exit poll narrative does is scale up the evidence for skulduggery to a scale of millions and place the mechanism on electronic fraud. Closer inspection shows that the exit poll evidence actually contra-indicates electronic fraud on a scale of millions. It says nothing about voter-suppression, little about high residual votes in Democratic precincts, or places I am genuinely suspicious of (New Mexico). In other words, it misdirects. And I don't like misdirection whether it is done through honest misevaluation of the evidence or deliberate sleight of hand. Sometimes I honestly think that Rove was behind the exit poll story, in order to distract attention from the real methods used to ensure Bush's victory. (That is almost not a joke).

Cheers

Lizzie


Edited to fix tags, and to say - my name is Febble :)