A U.S. House of Representatives task force has voted to authorize additional testing of e-voting hardware in Florida's 13th Congressional District to try to determine whether problems with the machines could have contributed to an 18,000-vote undercount in the 2006 U.S. House race.
In a unanimous vote Tuesday, the Committee on House Administration's Election Task Force adopted three additional testing scenarios for the e-voting machines, which were proposed in a report from the Government Accountability Office (GAO). The task force had asked the GAO to look at the e-voting hardware in the contested election and report back with its recommendations.
In its 27-page report, the GAO said that previous testing of some of the e-voting machines did not include a large enough sample to provide accurate results.
The 18,000 undervotes -- or votes that were not cast at all in the Congressional race by voters who turned out for the overall election -- occurred in Sarasota County, Fla. The losing candidate in the race, Democrat Christine Jennings, and other e-voting critics, argued that the undercount was too large to have occurred naturally and blamed problems with the e-voting machines. The 1,499 e-voting machines used in the election were direct-recording iVotronic systems made by Election Systems & Software (ES&S) of Omaha.
Jennings lost the election by 369 votes out of 238,249 votes counted. Republican Vern Buchanan was certified as the winner in the race.
"We found that some of the prior tests and reviews provide assurance that the voting systems in Sarasota County functioned correctly, but they are not enough to provide reasonable assurance that the iVotronic DRE voting systems did not contribute to the undervote," the GAO report said.
Rep. Charlie Gonzalez, (D-Texas), chairman of the task force, said in a statement that "the results of the GAO's testing will play an essential role in determining whether the machines did or did not contribute to the undervote in the contested election. The testing and its results are crucial; in their absence, we cannot answer this question with the confidence and assurance that is necessary."
The task force voted to pursue the three key tests proposed by the GAO report:
-- Firmware testing to verify that the firmware in the iVotronic machines used in Sarasota County matched the certified version of the firmware approved by election officials.
-- Ballot testing of the iVotronic machines using more than 112 scenarios, including casting votes, changing votes, changing votes again and other combinations, to confirm correct operation of the units.
-- Miscalibrating the iVotronic units to see if that has any effect on the undervote counts.
The additional testing will be conducted between Nov. 26 and Dec. 7, after upcoming local elections in Florida. The tests are expected to take about two to three weeks to complete after which the GAO will report back to the task force.
Naba Barkakati, a senior level technologist in the applied research and methods division at the GAO and the author of the report to the task force, said that the new round of testing will involve 115 of the iVotronic machines to provide a more accurate representation of the results. The three new tests will help determine whether the undervote was caused by the machines or by other possibilities, such as voter confusion over the ballot or deliberate decisions by voters not to cast votes in the race.
"The bottom line is that this will help eliminate the machines as a reason" for the undercount if no specific problems are found, Barkakati said. "We would love to find the reason, too. The machines are the only thing we can test. I guess if the machines function OK in our tests, it will still be somewhat of a mystery."
Sterling Ivey, a spokesman for the Florida Department of State, said the additional GAO tests will likely confirm the earlier test results. "The GAO is more than welcome to come into Florida and do more tests, but we still firmly believe in the tests that we've conducted" as well as those done by outside groups, Ivey said. "We don't feel that these [new] tests will give a different result."
Ken Fields, an ES&S spokesman, said in a statement that the company has "tremendous respect for independent, expert reviews. For that reason, we have cooperated with numerous such reviews related to the District 13 Congressional election. We are still reviewing the just-released final report from the GAO. But we can tell you that every test conducted by the Florida Division of Elections, Florida State University and Sarasota County has found no technology-based iVotronic system error that pointed to the cause of the undervote.
"The collective results of extensive testing conducted by independent experts has demonstrated that Sarasota County's voting system performed properly and exactly as it was designed to function," Fields said. "Based on all of that research, there is no reason to doubt the fact that the voting system accurately captured and counted all of the votes that were cast in Florida's 13th Congressional District race."
This story, "Florida Voting Machines Under Scrutiny Again " was originally published by Computerworld.