The Wikipedia says: “An undervote occurs when the number of choices selected by a voter in a contest is less than the minimum number allowed for that contest or when no selection is made for a single choice contest.”1 Undervotes can be intentional or not. This paper examines the question of whether there was some systemic explanation for undervotes in South Carolina in 2014.
I was a poll worker in the 2014 General Election in South Carolina, and had first-hand experience with the setup and operation of the electronic voting machines used by the state. I also had first-hand experience with problems reported by voters trying the use the machines. There were not many problems reported, but in three instances voters who believed that they had completed their ballot found upon review that no votes at all were recorded. In one instance where I was personally requested by the voter to assist, I observed a review screen where a straight-party vote was recorded, but no individual votes were recorded (the straight party vote should have caused the machine to mark every candidate of that party in every race). In one instance a voter requested assistance in choosing not to complete a write-in vote and after pressing the Cancel button on the write-in screen it was observed that no votes appeared for any race, something which the voter stated was wrong. One voter protested that the review screen repeatedly showed no votes for races he had voted in, and gone back to correct. In each problem case, the solution of “try it again” resulted in a ballot where the review screen showed what the voter intended.
One issue with the aging voting South Carolina machines is that in some circumstances, pressing the box on the screen to record a vote does not record a vote, and no visual indicator appears on the ballot screen. In one instance, poll workers worked for several minutes to try to get one checkbox to activate during the initial setup of the machine. This was an extreme and anomalous occurrence, but during the day several voters reported that they had to press the selection multiple times.
This analysis was originally promoted by my concern over whether there was some undesirable flaw in the voting machines that could lose votes, and whether some voters who didn’t pay attention to the review screens may have lost votes.
South Carolina considers votes to be public records. The voting machines do not know the identity of the voter, and a randomized detail log of individual votes cast does not compromise the privacy of any individual voter.2 A log of every vote cast, called the EL155 file, is available in randomized order from the South Carolina Elections Commission web site for each county and election.
Documentation of the EL155 file could not be found by me on the Internet, but the layout was for the most part self explanatory. My first step was to load the raw EL155 data into a relational database to allow for easy tabulation using SQL queries.
References:
- Automated Analysis of Election Audit Logs – AuditBear project
- Audit Analysis of the Venango County 2011 Municipal Primary: Initial Report
- Auditing a DRE-based Election in South Carolina
- Automated Election Auditing of DRE Audit Logs
1Undervotes do not disqualify ballots in South Carolina.
2One voter told me that he did not believe his vote was secret and that “they” would know how he voted. I told the voter that “the machine doesn’t know who you are,” but he didn’t seem satisfied.
3In this case, Microsoft SQL Server 2008.