This is just getting started, and work is being done to accually quantify the reluntant GOP responder effect, but the interesting thing to me was the fact that even where the Dem won, the exit poll seemed to indicate the margib rather than being say 1%, should have been reported as 8%. No doubt part of this is due to the reluctant GOP responder, but not nearly enough to make a fraud assumption unlikely(read about the central office "working on the final numbers so they could not give out photo copies of the raw data as reported to them"). Seems exit polls do work to detect fraud - if large enough relative to the voting population.
http://electionintegrity.googlegroups.com/web/EI+EVEP+Operations+report+061125.pdf?gda=H2qJ9VoAAADNEvMuFI2kxJITVDxIXJwmC1vZ6eNCpYJg-WLfQ1WIBRVJVT3hrgKrn_IySc8la3KSQPqrAdgTWV44idjWfVlNNkou5-XdSxrnUr5tn8Q9wouZlWRJeUWGpPDAX33A_EoThe Inaugural Election Integrity Election Verification Exit Poll
Operations (Phase One) Report
Steve Freeman Saturday, November 18, 2006
Director, Election Integrity Revised (slightly): November 25, 2006
<snip>
The pollsters who conducted the 2004 national media exit poll attributed the 2004 exit poll discrepancy to non-response bias, in particular to selection bias among interviewers who were younger and more highly educated (see “Evaluation of Edison/Mitofsky Election System 2004” pages 43-47). In Chapter 5 of “Was the 2004 US Presidential Election Stolen: Exit Polls, Election Fraud and the Official Count” (Seven Stories Press, 2006), Joel Bleifuss and I systematically refute these attributions, but nevertheless, they have been accepted by most casual observers as explanatory.
In future efforts, we hope to rigorously study so-called “interviewer effects” through recruitment of a more diverse population of interviewers. The wide variation in our interviewer response rates in this election may also afford us a basis for analyzing interviewer effects. <snip>
6th Congressional District, Lois Murphy conceded the race the morning after election day when she reportedly trailed Murphy by 3,000 votes (1.2%) with 100% of precincts counted (but prior to counting of provisional ballots).6 Our initial calculations indicate a disparity of eight percentage points as compared to these unofficial numbers, meaning that our raw polling data indicated that Murphy won the race by nearly 7%. Election Integrity counsels candidates to never concede close races before forensic analysis is done on the results.
The other Democrats did very well in the election, easily wining the Senate, Gubernatorial and House 7th district races. In our polls, however, they did much better yet. For the Montgomery and Chester County precincts, there are disparities between the count and exit poll results in all four races of between five and eight percentage points.
Five to eight percentage points is a large disparity. This was a carefully done survey in which we eliminated most of the potential sources of error. Moreover, response rates were high and our pollsters took their tasks very seriously. In his two decades of exit polling, Ken Warren has only observed a disparity of this magnitude in one race, and in that race, mass scale fraud had been subsequently uncovered. And the disparity we observed is apparently consistent with those found by the media-funded National Exit Poll and Zogby’s post election telephone poll.
On the other hand, debriefings with our interviewers indicated some doubt about the representativeness of their samples. Several interviewers had response rates in the 30% range or lower (we had a few who legitimately seemed up around 90%!), and several of the interviewers with lower participation rates subjectively felt that Republicans were disproportionately avoiding participation. In most shifts, a small number of those approached, typically one or two per shift, reacted antagonistically, for example questioning the interviewers’ or organizers’ motives. In a few of these cases, the complainant overtly accused us of having “liberal” motives or otherwise indicated Republican preferences, e.g., proceeding to speak with Republican candidates or committeemen. The Republican party has been openly critical of exit polls (in the U.S.), especially since the 2004 election. To at least some small extent, such rhetoric has poisoned the ability to obtain a purely random sample of respondents.
I observed interviewers both unobtrusively and openly, and although the interviewers’ performance and effort were in all cases outstanding, I did observe one subtle breech of protocol: voters who had not been selected sometimes asked to fill out a questionnaire and sometimes were in fact permitted to do so. In debriefings, it was clear that these were not isolated occurrences, although neither does it seem to have been all that widespread. Even if every antagonistic non-respondent were Republican and every non-selected respondent a Democrat, it could not account for much of the disparity. On the other hand these cases may be just the visible portion of a larger phenomenon. Many people were thrilled that we were doing what we were doing, and expressed disgust or contempt at the idea of electronic voting and suspicion of the election process. These people invariably agreed to participate; and we know from other polls that Democrats are, in general, far more concerned about e-voting and election fraud than Republican
<snip>See Stephanie Singer’s blog at www.electionintegrity.org . Astonished and frustrated by this key missing data, I called the Delaware County Board of Elections on Friday afternoon, November 17, 2006, and asked for myself if I could have the precinct tabulations. I was told that the director, Laureen Hagan, was out to lunch, but her assistant Regina said that I could have these numbers but I would have to go into their office. She couldn’t give it to over phone, fax, etc… because someone “might make a mistake.” I said that was OK, but she insisted I would have to come in. I called Stephanie (who lives closer to that office). She expressed some doubt, so I called again to make sure; I was told that neither Laureen nor Regina was there, but another woman named Lorraine absolutely assured me that if someone came there, he or she could obtain the numbers. So Stephanie went. But Laureen, the director, who by then returned from lunch, said she could not allow it. When I told her that we had been assured that we could get the numbers if we went to their office,,she said that she would have permitted it, but couldn’t because she was at that time “working with the numbers.” I asked what that meant, and she said that she was working towards certification, that there was a certification deadline and that she was “working with the book and needed to work with the numbers.” I asked again what that meant and why Stephanie getting the unofficial precinct tallies would interfere with that. She said that she just told me, but that I didn’t like her answer… Etc…
<snip>