The Center for Voting Technology Research (VoTeR Center) at the School of Engineering of the University of Connecticut performed a post-election audit of the memory cards for the AccuVote Optical Scan tabulators that were used in the August 2008 Connecticut Primary Elections. The cards were programmed by LHS Associates of Methuen, Massachusetts, and shipped to the towns in Connecticut. For the purpose of the post-election audit, the VoTeR Center received 297 memory cards from a number of districts after the elections. The audit was performed on 280 of these memory cards (among the 17 cards not included, 14 were from the February 2008 Connecticut Primary, two were from the March 2008 elections, and one card arrived damaged, preventing data analysis). This document reports on the findings obtained during the audit. Among the 280 cards tested, 237 cards (84.6%) were found to have been properly programmed for election. These cards contained valid ballot data and the executable code on these cards was the expected code, with no extraneous data or code present. The remaining 43 cards, or 15.4%, were found to contain seemingly random data, or as we call it, “junk” data. Such cards are not accepted by the tabulators, and they could not have been used in the election. Nevertheless, this is a high percentage of faulty cards. Among the 237 cards found to be properly formatted for elections, 153 cards, representing about 80% of the districts, were used for the actual election, while the remaining cards were not used (this is expected, given that under normal conditions, only one card is used per polling place). All cards were properly programmed and contained no unexpected code. Among the 153 cards actually used, 141 cards, or 92%, were in the “Election Closed” state with non-zeroed counters, which is the intended state. Additionally, 9 cards (5.9%) indicated that printing was aborted, which suggests that election officials turned off the machine before the printing of the election results was completed. 3 cards (2%) were in the ’election closed’ state with zeroed counters. Among the 84 cards that were not used in the elections, 83 cards, or 98.8% were in “Set for Election” state with zeroed counters, which is the intended state. The one remaining card was in “Not Set” state with non-zeroed counters. Although this is not the intended state for a card that was not used, before one could use this card, it would have to set for election, which would result in the card entering the “Set for Election’ state and the counters of the card zeroed. The audit was performed on request of the Office of the Secretary of the State.
Full report: audit-post-08-08