Academic journal article The William and Mary Bill of Rights Journal

Machine Errors and Undervotes in Florida 2006 Revisited*

Academic journal article The William and Mary Bill of Rights Journal

Machine Errors and Undervotes in Florida 2006 Revisited*

Article excerpt


The 2006 election for U.S. House of Representatives District 13 in Sarasota County, Florida, attracted extensive controversy because an unusually high proportion of the ballots cast lacked a vote for that office, and the unusual number of undervotes probably changed the election outcome. Intensive technical studies based on examining software and hardware from the iVotronic touchscreen voting machines used to conduct the election failed to find mechanical flaws sufficient to explain the undervotes. Studies that examined the ballots used in Sarasota and in some other counties concluded the high undervote rate was caused by peculiar features of the ballot's format that confused many voters. I show that recorded events involving power failures and problems with the Personalized Electronic Ballots used with the machines correlate significantly with undervote rates in several Florida counties. The relationships between machine events and undervotes are sufficiently substantial and varied to make it unreasonable to discount the likelihood that mechanical failures contributed substantially to the high numbers of undervotes.


The controversial election for U. S . House of Representatives District 1 3 (CD- 1 3) in Florida in 2006 has been extensively investigated, but the basic question of what happened with the electronic voting equipment used there remains unresolved. An unusually large number of ballots cast in Sarasota County did not include a vote for that office: 1 8,4 1 2 of the 238,249 ballots cast on iVotronic touchscreen machines in the county were undervotes for that race.1 Some ofthe research regarding the high undervote rate has occurred in the context of litigation,2 some has been done by independent scholars,3 some was sponsored by the State of Florida,4 and some was conducted by the federal government.5 The unusually high undervote rate probably changed the election outcome.6 Teams of computer scientists examined several features ofthe software and hardware used to conduct the election but found nothing they considered sufficient to warrant attributing the lost votes to defects in the equipment' s operations.7 Even though the adequacy of these technical examinations has been seriously questioned,8 many read the technical reports as largely exculpating the machines.9 Hence, some argue that the high undervote rate was caused by peculiar features of the ballot's format that confused many voters.10

Frisina et al., in particular, use data from several counties in Florida to show that the undervote rate for the attorney general race was also unusually high in several counties when the choices for attorney general - for which, like CD- 13, there were only two candidates running - were placed on the same screen as the governor's race, which had many candidates. 1 1 The correlation they demonstrate between the features of the ballot and the attorney general undervote rate is clear, but the explanation entirely in terms of voter confusion is speculative. There is no direct supporting evidence about the voters' experiences, although Selker reports some suggestive results based on experiments conducted using subjects fromBoston, Massachusetts.12 Frisina et al. point out, however, that their results set the bar for any competing explanation fairly high: "[A]ny explanation for the CD 1 3 undervote in Sarasota must be capable of explaining the attorney general undervotes in Charlotte and Lee Counties and the lack of an attorney general undervote elsewhere (not to mention the lack of high undervotes in other races)."13

Recent reports examining the use and performance of touchscreen voting machines across Florida in the 2006 election open the door to meeting the standard Frisina et al. propose. 14 Pynchon and Garber document extensive problems with the iVotronic machines that go well beyond the scope of the officially sponsored technical studies. 15 They show that undervote rates for many races were higher in many counties where iVotronic touchscreen machines were used, regardless of the ballot format. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.