Last month, IPAT Clinic students Prachi Mistry and Hannah Green observed Orange County’s One Percent Manual Tally (1PMT). The 1PMT is a process in which each county in California manually tallies the paper ballots in 1% of precincts as a measure to ensure that the reported election outcome is correct. California has been conducting the 1PMT since the 1960s. Any member of the public can observe these audits if they agree not to interfere with the audit process, and on March 14, IPAT Clinic students did just that. We learned a lot. We came away with a new appreciation for the importance of post-election auditing and the care that must go into doing an audit properly.
First, Operations Manager Justin Berardino used the Election Management System (EMS) software, made by DFM Associates, to select 1% of precincts for the 1PMT. Because all races were not accounted for in the initial ballot selection, EIMS was again used to select additional precincts until at least one precinct covering every contest in the election was included. (We noted that the selection of precincts is not perfectly random because it relies on software rather than a physical process such as rolling a dice or flipping a coin. Unlike the 1PMT, risk-limiting audits (RLA) generally use a 10-sided die to randomly generate a number which is then put through an algorithm to select ballots for the audit.) Click here for an overview on RLAs.
Mr. Berardino then took the observers into a secured area that required passing through three doors that required key-card access, where all vote-by-mail and vote center ballots were organized in batches (as depicted in the image on the right above). Batches of ballots were organized by date and precinct or vote center. At the top of each batch was a paper printout indicating the ballots contained in that batch. To retrieve the ballots for the 1PMT, staff were split into groups—some to identify which ballots to be pull, and others to pull those ballots. Meanwhile, other staff members would go to each batch with a cover sheet in hand and find the selected ballots. They would then return with the critical ballots from a given batch and the respective cover sheet so that that the other staff members could verify that the correct ballots had been pulled from the batch. The selected ballots were then placed in boxes organized by precinct, to be delivered later to the staff members in charge of tallying the votes on the ballots.
Mr. Berardino informed the observers regarding the procedures involved in the tally process. These procedures made clear that election staff take extensive measures to ensure the accuracy of the audit. Specifically, staff were split into teams of three where one staff member would read the votes on the ballot out loud, another would tally the votes, and the third would listen and observe. Staff members would occasionally rotate to avoid fatigue and ensure accuracy. The tallies would then be compared to the electronic tally stored on V-drives taken from the vote centers and the computers on which vote-by-mail ballots were scanned. Notably, for security purposes the V-drives (modified flash-drives) cannot be opened or accessed without the use of another drive. The meticulousness of the tally and security of the V-drives demonstrate the extensive measures taken to ensure the transparency and security of election audits. Separately, to ensure voter privacy, observers were not permitted to take any photographs that may include voter identifiable information and were always monitored .
We were impressed by the organization and efficiency of the staff in compiling and extracting ballots. The team consisted of both permanent and temporary election staff. Everyone on the team seemed enthusiastic and well-informed regarding the various staff members’ jobs and responsibilities. Additionally, it was great to see community volunteers assist in the process. As observers, we were given detailed information about the entire audit process, and everyone answered our questions to the best of their abilities. The staff prided themselves on the measures taken to ensure that each vote was counted, and on the security measures taken to protect voter privacy and election integrity. We were able to see first-hand how seriously the Registrar’s office takes election audit procedures.
We observed the ballot selection for the March Presidential Primary Election on March 14. We were advised, from our correspondence with the Orange County Registrar, that the audit itself began March 16 and was completed by March 20; Registrar of Voters Neal Kelley then certified the official results of the March Presidential Primary Election on March 23.
Observing the 1PMT audit and speaking with Orange County Registrar staff regarding the RLA audit provided invaluable insights to the IPAT Clinic’s research into post-election audits. Most importantly, we learned that an appreciation for thoroughness and attention to detail are critical when doing an election audit. In addition, our observation confirmed that virtually no audit can be totally infallible. For example, with the 1PMT, the selection of precincts is not perfectly random because it relies on software rather than physical processes such as the rolling of dice. Along similar lines, RLA pilot audits have used a 10-sided die, but the dice have not been tested to ensure that they are random. The IPAT Clinic, together with the California Voter Foundation (CVF), is developing a report comparing the one percent manual tally (1PMT), which California has used since the 1960s, and risk-limiting audits (RLAs), which are currently being piloted throughout California (click here for a review of California pilots and here for the 2018 Orange County pilot). In our report, we recommend that the Secretary of State procure, test, and certify the dice used in audits. Stay tuned for more!