2006 Election Integrity/Voter Confidence Subcommittee Meeting Minutes

Lynda Roberts, Registrar of Voters, Elections


September 19, 2006

Registrar of Voters
Election Integrity/Voter Confidence Subcommittee Meeting
Tuesday, September 19, 2006


A meeting of the Voting Integrity/Voter Confidence Subcommittee of the Election Advisory Committee was held on Tuesday, September 19, 2006 at the Registrar of Voters’ Office. Members present were: Sherry Resen, Bob Roberts, Linda Bagnaschi Dorrance and Carl Carter and Elaine Ginnold from ROV.

Chain of Custody procedures for ballots and scanners

Committee members made suggestions for revisions to the Procedures for Deployment and Chain of Custody of Ballots and Voting Equipment. The ROV will revise the procedures and distribute them at the next meeting of the Election Advisory Committee.

Wireless capability on equipment

Ms. Ginnold advised sub-committee members that there is no wireless capability on the scanners.

The serial port connector on the back of the scanner controls the write- in diverter in the ballot box. ROV does not use this diverter, since it tends to jam with 18” ballots. The modem connection is for one-way transmission of results from the precincts on election night.

Manual Tally

Members discussed the post election manual tally and the Roy Saltman article. In order to build up voter confidence in the election results, ROV will hand count ballots from 3% of the precincts for the upcoming November election and then assess the impact of this change. Saltman has recommended a tally of 3-5%, however, there was no supporting research cited in his article to form a basis for this recommendation. Arel Cordero at UC Berkeley has agreed to study the issue of what an optimal percentage for a manual tally would be. He has also agreed to study the issue of the percent spread between winning and losing candidates that should trigger a recount of the contest.

Use of volunteers to recount ballots: Ms. Ginnold said that she would be using regular full or part time paid staff to perform the manual tally. They have experience and are known to be dependable in their attendance.

Location of tally: The location of the manual tally will be at the Registrar of Voters’ Office.

Use of videotape, cameras to record ballots: Committee discussed the suggestion to recount and videotape ballots at the precincts after the clo se of polls and decided that it would present a security challenge. ROV does not permit videotaping at the polls or in the office since it violates employee and poll worker privacy and is intrusive to the process of the canvass. However, observers are welcome to observe the entire canvass process.

Time of audit. The post election canvass in CA by law begins on the Thursday after the election and continues until it is completed. The manual tally in Marin County will continue to take place after all of the ballots are counted as recommended in the Wagner/Cordero paper on method of random selection of precincts.

Roy Saltman article: Saltman recommends a manual tally of 3-5 per cent of the precincts. ROV plans to conduct a manual tally of 3% of the precincts. See above on study on optimal percentage for this tally.

Criteria that would trigger an automatic recount : Committee discussed the possibility of the ROV conducting an automatic recount if the percentage spread between a winning and losing candidate is equal to or less than .5%. ROV will consider doing such a recount during the canvass period if such a situation occurs after the November election. Recount will be done on a different machine than the one used on election day and ROV will consider hand counting additional precincts in the contest if time permits. Goal is to do what is necessary during the canvass to give voters and candidate’s confidence that the results are accurate.

ROV Election Day incident reporting forms

Reviewed the election day reporting form used in the Office and recommended expanding the Accuvote OS portion. Committee asked to look at the form used at the polls at the next meeting.

Other business
Report to Election Advisory Committee

Ms. Ginnold will report the recommendations of the sub committee for a 3% manual tally and automatic recount if .5% margin in a contest to the full committee at the next meeting.

Meeting adjourned at 6:00 p.m.

September 11, 2006

Registrar of Voters
Election Integrity/Voter Confidence Subcommittee Meeting
Monday, September 11, 2006


A meeting of the Election Advisory Committee’s Subcommittee on Voter Integrity/Voter Confidence Subcommittee was he ld on Monday, September 11, 2006 at the Registrar of Voters’ Office.

Present were: Cat Woods, Bob Richard, Linda Bagneschi Dorrance, Sherry Reson and Elaine Ginnold from the Registrar of Voters

Committee viewed the Accuvote Optical Scanner that is used at the polls and saw a demonstration of how it is sealed before the election. There are 2 numbered, tamperevident seals place on each side of the machine over the screws of the casing. There is another seal over the door to the memory card, which is numbered and difficult to break without wire cutters. Committee members asked whether there was access to the memory card via the modem connection or the serial port at the back of the machine. Ms. Ginnold to find out more about access to the memory cards.

Members asked for clarification on how to read the statement of vote, since it is difficult to understand it. Would like to be able to see turnout by precinct as well as overvotes and undervotes. Would like statement in a form that can be analyzed in program such as excel or access.

Members discussed the draft chain of custody procedures. Ms. Ginnold will revise them according to committee suggestions and bring them to the next meeting. The plan is to incorporate most of the questions about storage of the machines, etc. into the procedures.

Ms. Ginnold told committee members that there are few academic studies of the administration of elections. More studies such as the one done by UDB on the method of random selection of precincts for the 1% manual tally would be helpful. Arel Cardoso, who is a student of David Wagner in the Computer Science Dept. at UCB, has agreed to do a study of the percent of precincts that should be recounted as part of the post election manual tally. He will also study the margin of victory that should trigger a recount.

Members discussed the idea of an expanded audit after the election. However, agreed it would be important to set up guidelines for this so that we do not waste resources. All agreed that if the margin of victory in a contest was the same as or less than the margin of error of the machine, currently estimated to be .5%, that the office should be recounted during the canvass. Cat Woods will pursue finding out if there is a way to statistically determine the number of precincts that would need to be recounted. Ms. Ginnold said she would consider doing a recount in such a close contest, most likely it would be a machine count on different scanners. The question arose as to how many scanners we would have to recount the ballots on to get a level of confidence. Will ask statistician.

1% manual tally. The precinct selected for the 1% manual tally will include the absentee ballots in that precinct. In the event that there is an unexplained discrepancy between the manual and machine tally, the ROV uses his/her discretion to determine what to do.

Write- in votes. By law, if voters do not fill in the bubble opposite the name of the writein candidate, their vote for the candidate will not be counted. Members requested that the bubble issue be placed on the list for proposed legislation.

The article by Roy Saltman was distributed by Sherry Reson. Members will read this before their next meeting.