E-voting earns nod of approval
Study: Touch-screens cut 'lost' ballots

By CARLOS CAMPOS
The Atlanta Journal-Constitution
Published on: 10/24/04

Georgia's touch-screen voting machines have dramatically reduced the number of "lost" votes — a measure of the reliability of voting technology, an academic study has concluded.

The study recently released by the Caltech-MIT Voting Technology Project notes that the rollout of the machines in 2002 by Georgia Secretary of State Cathy Cox was clouded by controversy over the machines' security.
Some computer scientists, political activists and other critics contend the machines use simplistic software that easily can be rigged to manipulate election results. The study does not attempt to address those security issues.

But its author, Charles Stewart, head of the political science department at the Massachussetts Institute of Technology, said an important question was lost in the controversy: "Did the Diebold machines perform better than the collection of older voting technologies Georgia had used before? The purpose of this research is to answer this question. The answer is yes."

About 30 percent of Georgians had cast ballots on punch cards before the statewide switch to touch-screen machines. The rest voted on optical-scan and lever machines, or by paper ballots.

In the 2000 presidential election, Florida was at the center of a national controversy because elections officials could not determine many voters' intent on punch-card ballots. Chads — whether hanging, dimpled or pregnant — became household words for many Americans following the election debacle.

The Caltech-MIT study notes Georgia's rate of uncounted votes was far worse than Florida's in 2000. Georgia had the nation's second-worst rate of uncounted votes at 3.5 percent.

"Lost" or uncounted votes often result when a voter intentionally skips over a race on a ballot.

But they are often used as a measure of voting technology performance, since undervotes are also an indication that a machine may have failed to properly record a voter's intent. For example, the number of uncounted votes in Georgia's 1998 gubernatorial election was 49,777.

After touch-screen machines were put in, the number of uncounted votes in the 2002 gubernatorial election ped to 19,792 — even though 233,053 more people voted in that election.

The study notes that the machines are doing a better job of tallying votes accurately in counties with "disadvantaged populations — counties with larger African-American populations, rural counties, low-income counties and counties whose residents were less likely to have completed high school."

Cox hailed the Caltech-MIT study as "powerful new evidence that . . . touch-screen voting has improved the accuracy of the vote count in Georgia."

"Dr. Stewart addressed the question that is too often ignored by critics: What would have happened if Georgia had done nothing?" after the 2000 presidential election, Cox said. "And the answer is we would be going into this important presidential election knowing that tens of thousands of votes that were cast would not be counted, and we would still have a system that gave lower-income, rural and minority communities a less-than-equal opportunity to have their ballots properly counted."