Re: Here comes another fraudulent OH election

From: Richard C. Johnson <dick_at_iwwco_dot_com>
Date: Thu Nov 02 2006 - 10:52:29 CST

  This "statistical" procedure you describe is indeed a farce. Absent any examination of the VVPAT, there is no meaningful audit. Even a limitation of funds only means a limitation on the number of precincts in which one could conduct the audit. The use of volunteer students (natural in a university environment) can greatly reduce the cost of the audit, The choice to stop short of an audit but still call it that is pretty lame, IMHO. Imagine an accountant interviewing a company's officers but not bothering to examine the books and pronouncing the company in great financial health!
  By the way, CS people are not necessarily statisticians. Check the math department, political science, sociology, and the epidemiologists at the university hospital for people who have greater knowledge of statistics.
  -- Dick

Jerry Lobdill <> wrote:
  I have been consulting today with an Ohio university's CS people who have been tasked by the elections administration of Cuyahoga County, OH, to "design an audit for the county".

They have Dopp and Stenger's recent paper and wanted to know if I'd recommend their method of computing the sample size or Rivest's method. (I recommended Rivest because it is simpler.)

Obviously they are being used to add gravitas to a fraud the elections administration (EA) is planning to perpetrate. Here is what the process is going to be:

1. At close of polls the (EA) will combine the precinct vote counts using GEMS at election central HQ.
2. The results will be furnished to the university team.
3. The university will choose three races from the list of races that appeared on every ballot in the county. (They say they plan to choose the closest races.)
4. They will be given the incremental vote counts from the precincts as reported by them.
5. They will also be given the incremental vote counts that GEMS summed to get the final outcomes on a race by race, precinct by precinct basis.
6. They will use the method Lobdill and Stanislevic recommend to calculate the sample size.
7. They had some advice from a statistics prof to the effect that they should choose different samples for each of the three races. I recommended against that. Choose one sample on the basis of the margin and vote count distributions of the closest race. It will be right for the closest race and overkill on the other two.
8. Then, they said, they would compare the precinct reported vote counts with the GEMS reported vote counts for these precincts. (!!!!)

I asked why they weren't going to recount the VVPATs for the sample precincts, and they said they wouldn't be allowed to do that because:

   By the way, all the precincts can't be counted since the hardcopy long reports that are a summarized version of the ballots include every race in the county by DRE and precinct. This is a very long report to go through and each DRE will have several precincts. There can be up to 40 DREs to just extract 1 precinct though usually there will be more than 1 precinct extracted from that same set of DREs.

I explained that the practical reason they were sampling and doing a partial recount was because of the reluctance to do a full recount because of time costs. And that the only way they could get a valid audit was by doing that. I said that since they would be getting all the data they were going to ever process on a single spreadsheet there is no need to sample at all. Simply compare GEMS values with precinct reports of vote counts on the spreadsheet. The time it will take is milliseconds. There is absolutely no need to sample if that is going to be the test.

I have not heard back from them on this yet.

The fact is that they are being used, and their involvement is as part of a fraud, because this exercise will be reported to the public as an audit, and any question of validity will be deflected by a response that they had the very best advice they could get from the university CS experts.

Jerry Lobdill

OVC-discuss mailing list

OVC-discuss mailing list

= The content of this message, with the exception of any external
= quotations under fair use, are released to the Public Domain
Received on Thu Nov 30 23:17:03 2006

This archive was generated by hypermail 2.1.8 : Thu Nov 30 2006 - 23:17:19 CST