|
First, I'll try to clarify the meaning of pre- and post-tabulator vote counts.
The pre-tabulator vote counts are marked down in each individual precinct "poll book", at the time of poll closing, I believe. I suppose that these are obtained by the electronic tabulation from each of the individual machines at each precinct. I don't know if they are computed by hand or electronically, or if these vary. I also am not clear about the legal standing of these numbers. I do know that they are supposed to serve as a check on the accuracy of the final counts.
The pre-tabulator counts are then sent to the county's central tabulator, which then comes up with a total (post-tabulator) count. So each county reports not only a total (post-tabulator) count for the county, but also the official (post-tabulator) count for each precinct. These are the official results.
Obviously, the pre-tabulator and post-tabulator counts for each county should be equivalent. How is it that one would find post-tabulator counts to be systematically less than the pre-tabulator counts? Fraud would be at the top of MY list. In fact, when I initially proposed that the central tabulator may be deleting votes, some people told me that they thought that would be unlikely because then the pre- and post-tabulator results wouldn't match. My thought in response to that was, "How do we know that they DO match?"
The issue of the relationship of all this to the exit polls is complex (and I have discussed this with Febble). This type of fraud would cause a discrepancy between the official vote and the state exit polls (and in fact, we do have a discrepancy of 4.2%), but it would NOT cause a red shift in the WPE. The reason it wouldn't cause a red shift in WPE is that the relative percentages (vote share) within any individual precinct would NOT be changed, since votes would be deleted in proportion to the relative vote shares of each candidate). BUT, if this was done disproportionately in Kerry precincts (which I believe it was, and which our initial investigation suggest was the case), then it would cause a discrepancy between the state exit poll and the official results.
Another interesting thing about this type of fraud is that it would escape the kind of detection that the Mebane and ESI analyses were meant to show, since those analyses were based on within precinct analyses. So, when those two studies came to the conclusion of "no widespread election fraud" they totally failed to consider the possibility of this type of fraud.
Very interesting.
|