|
MTS
1% Manual Tally Analysis for LA County, 2006 On June 26, 2007 the Los Angeles Board of Supervisors approved the request by Registrar-Recorder/ County Clerk McCormack to exclude the MTS tabulating system from the Top to Bottom Review of California election systems conducted by Secretary of State Debra Bowen and on Nov. 20, 2007 approved Ms McCormack’s request to use it in 2008 without its being reviewed. I wrote to Secretary of State Bowen requesting that MTS be included because of my simple study of the 1% manual tally of 2005 that I sent to her. In 2005 I analyzed the 1% manual tally to see how accurately the hand count and the MTS-counted ballots matched exactly. They matched an average 28%: 22% for the 8 initiatives, 14% in the local elections, and 44% in 8 local issues. Now I am submitting a statistical report of the 1% manual tallies of the 2006 June primary and November general election done by Brian Dolan, a professional statistician. This report also shows how inaccurately MTS counts our votes. I summarize his report. He first did a line by line analysis of every entry on the reports for the 70 to 83 precincts. In 8869 entries, (he calls records,) 81% of the hand count matched the computer count. That means MTS does not count accurately nineteen out of every 100 votes in a line by line analysis. But that total 8869 includes 1071 zeros (12%), thus the line-by-line analysis only matches 77% of the time: 23 out of every 100 votes are not counted accurately. But at the contest level only 13% matched, that is, eighty-seven percent had discrepancies. Contest means all the candidates in each race for an office. In my 2005 report I examined the 1% manual tally at the contest level. The June, 2006 primary had more problems than the Nov. general election. The manual count shows two kinds of errors made by the MTS programmed scanners: it misses votes because the scanner does not read the ballot if the ink dot is not dark enough or is not centered: Mr. Dolan interpolated across the county from the results of the1% analysis: the rate MTS misses a vote, that is on the ballot, is 7 in every thousand votes cast go uncounted. But MTS adds votes that are not on the ballots at the rate of 3 in every thousand. He found the largest discrepancy was142 votes added in one primary contest for a county central committee. Mr. Dolan used .5%: 1 in 200 votes, as permissible error rate. The Federal Election Commission recommends an error rate of one in 300,000--.0003%. There is no set guideline for error rate in California. With Mr. Dolan’s .5% error rate, the error rate across the county is 1%: one in 100 votes counted by MTS are incorrect. In Los Angeles County has about 3 million voters, thus approximately 30,000 votes are not counted accurately. The accuracy level seen in this analysis is totally unacceptable. We have to count the votes in Los Angeles County more accurately than we see here. M.T.S. must be examined by the state experts and analyzed for its accuracy, transparency, and reliability in the same manner as the other California election systems were. I request that you and your staff members study the analysis completed by professional statistician Brian Dolan showing its serious level of inaccuracy and then find ways to improve its accuracy. Please do not replace MTS with any proprietary system now shown in the Top to Bottom Review to be poorly designed, inaccessible, and seriously insecure – easily hacked.
|