Good Luck With That Hand Count, Georgia!
R. Michael Alvarez
I was surprised to learn that the Georgia State Board of Elections had voted last week to require that counties hand count the number of ballots cast on Election Day in November. After studying elections for the past three decades, one thing that I’ve learned is that hand counting of ballots is hard. Most importantly, hand counting of ballots requires careful planning and the development of strong procedures to make sure that it is done securely, accurately, and in a timely manner.
Hand counting of ballots may look like an easy task, but it’s not. This is true even for simply trying to count number of ballots cast in a location after the polls close on election night (which is what my understanding of the Georgia proposal, that they are not proposing a recounting of votes cast for candidates, instead it will be a hand count of ballots that were cast).
I’ve personally witnessed hand counting of ballots on election night in many places. Each time I’ve witnessed it, unless there are well-developed procedures and guidelines in place for the election workers, many things can go wrong. At the end of a long day, under pressure to wrap things up, election workers need training and detailed procedures to avoid making mistakes. Election workers need:
Precise and detailed procedures for how the hand count.
Specific training on those procedures and how to conduct the hand count.
Procedures for what to do when the hand count does not reconcile.
A very clear chain of custody to make sure that all election materials are secured before, during, and after their hand count.
These procedures are also necessary to ensure that these hand counts are conducted in a consistent manner throughout the jurisdiction.
To give an example, one of the very first times I observed an election night hand tally and reconciliation of ballots cast, one of the ballots was mislaid during the hand count and it was of course not counted. So the hand count did not reconcile. This led to quite a bit of confusion, a search of the location, and eventually the election workers discovered the mislaid ballot. Once they included it in the hand count, things reconciled. But what if that ballot had not been discovered? What if problems like these significantly delay the reporting of election results? What if the hand count continues to differ from other tabulations in the voting location and they cannot reconcile the ballot counts on election night? What if the reconciliation is off by more than a single ballot, say off by a handful or more ballots? What do they do then?
Back in 2006 I had a great opportunity to work with Professor Lonna Atkeson on the New Mexico Post Election Audit Study. We had the unique chance to work with nearly 50,000 cast ballots from Bernalillo County, New Mexico, and to conduct a variety of machine and hand counting studies to prepare recommendations for how New Mexico counties might conduct post election ballot audits. I encourage everyone interested in this topic to read the project report — I especially suggest that Georgia’ elections community read the project report for valuable insights into the problems they might encounter and how to design procedures to make sure that if they do a post election ballot count, that it is done securely, timely, and accurately. The academics and interested election geeks might want to also read the book that was published, containing more detailed analyses of the New Mexico Post Election Audit Study, Confirming Elections.
Of these procedures that Georgia counties will need to develop, and develop quickly, regards how they will maintain the chain of custody of the cast ballots. If election workers need to unseal ballot boxes, that will need to be documented, observed by multiple election workers (and hopefully independent observers), and resealed after the count is completed. At no time should ballots be out of the control of election workers, and the entire chain of custody process needs to be documented itself on paper. Chains of custody are important components of audits and any type of recount, and they will be critical to develop and maintain if Georgia counties implement this post election hand count.
Thad Hall and I wrote a paper back in 2008 about the importance of chains of custody. What we wrote in that paper, “Building Secure and Transparent Elections through Standard Operating Procedures” (published in the journal, Public Administration Review) should be read by the folks in Georgia, they will need to develop these procedures very quickly.
I’ve got no doubt that if Georgia counties engage in hand counting of cast ballots, that it will slow down the reporting of results in November. Discrepancies will arise, and they will need to be investigated quickly and thoroughly. As we’ve found in the past, discrepancies that arise in recounts and audits are often due to procedural or human error, but election officials will need to release reports about any discrepancies that emerge and how they get resolved. Transparency will be crucial for maintaining confidence.