Scientists Want E-Voting Data for Research

LANSING, Mich. – After nearly 50,000 Michigan Democrats cast ballots over the Internet in February, academics eagerly sought election data that would help them determine what types of people voted online. But scholars around the country complain that they haven't been able to get statistics from the Feb. 7 caucus.

The delay could stall important research, they say, on voting technologies and on boosting participation in U.S. elections — for example, by studying whether Internet voting could help such historically disenfranchised groups as overseas military personnel and citizens who don't speak English well.

Researchers and civil rights experts say Michigan's Democratic Party (search) is merely one of a number of organizations that are stingy with voting data — even though computerized balloting systems and registration databases make such information relatively easy to share.

"The quantity and quality of data we get on elections is highly variable and highly inconsistent, and it makes it very difficult for us as social scientists to study what happened," said R. Michael Alvarez, a political science professor at the California Institute of Technology. "It also makes it difficult for the public to have confidence in the integrity of the numbers."

The dearth of data has thwarted independent analyses of issues ranging from disenfranchisement by ZIP code to whether the layout of a ballot influences the way people vote, academics say. Some registrars are quick to send CD-ROMs, spreadsheet files and other data to universities, they say, while others won't return researchers' phone calls.

Organizations that retain voting data — counties, states, political parties and the companies that build the hardware and software for voting machines — say they're not trying to stymie academics or keep the public in the dark. They say they're simply so overrun with requests that they can't keep up.

Michigan Democratic Party Executive Chairman Mark Brewer says he's eager to have someone analyze the February caucus, the country's biggest experiment to date with online voting.

A preliminary report of the five-week voting period showed more older voters than expected. Web-based ballots also seemed to help Democratic presidential candidate John Kerry more than Howard Dean, who was initially expected to get more Internet votes because his supporters tended to be young, well-educated and cyber-savvy.

But as the head of a major political party dealing with a presidential election in a battleground state, Brewer says he simply hasn't had time to organize the data for academics. The party, which ran the caucuses and paid for them, owns the information.

"No one's trying to hide anything," Brewer said at his state party headquarters. "We want it studied. And the DNC (Democratic National Committee) wants it studied, too."

Brewer says the data — which includes each voter's name, address, gender, race and other personal information — has to be handled carefully to protect voters' privacy. There's plenty of time to get a more detailed analysis, he said.

"The data's not going away," Brewer said.

The delay aggravates Michael Traugott, professor of political science at the University of Michigan, who says he tried to approach Brewer about a larger research project involving the data even before the caucuses were held but couldn't get Brewer to call back.

"They've been promising this data for a long time, and they haven't produced it," said Traugott, who has a grant from the National Science Foundation (search) to study election administration reform and the impact of new voter technology. He says he'd use aggregated data to come to general conclusions about voting systems — not to determine how any individual voted.

Thad Hall, a social scientist and assistant professor at the University of Utah, says organizations that maintain voter data shouldn't make excuses about privacy concerns or time constraints. The data he requests is entirely anonymous and, if registrars have proper coding in election software, only takes a few seconds to send as an e-mail attachment.

"Some officials are like, 'What if the scientists find out something bad about our data?' Their attitude should be, 'We're professionals and do a good job — and if you find problems, tell us so we can solve them," said Hall, co-author of "Point, Click, and Vote: The Future of Internet Voting,"

Hall and other academics praised officials in Los Angeles County for granting requests for data. The county — which, if it were a state, would be one of the top 10 voting jurisdictions in the country — has shared data such as participation rates of military and overseas residents; the number of people who request absentee ballots in Korean, Spanish, Vietnamese and other languages; and the number of people who are permanent absentee voters versus the number who request an absentee ballot if they're sick or on vacation.

The U.S. Election Assistance Commission, a 1-year-old federal agency created to oversee election reform nationwide, acknowledges that a lack of data is one of the biggest obstacles to making U.S. elections more accurate and efficient.

The EAC last week announced plans to spend at least $5 million on basic voting research from the 2000 and 2004 presidential elections, compiled by the National Institute of Standards and Technology.

"The truth is, most people would be shocked at the paucity of data we have on elections," EAC Chairman DeForest Soaries Jr. "We're flying without instruments."

Connie McCormick, registrar of voters in Los Angeles County, said demand for data could impose burdens on election officials — particularly in small jurisdictions struggling to submit election results on time. But data will help researchers — and boost voter confidence in election results.

"This will force counties ... to adopt clearly important process of basic statistical gathering," said McCormick, who was shocked to learn that Miami-Dade County had no local requirement in 2000 to reconcile the number of ballots cast with the number of voters who came to the polls. "How can you reassure the public on any level if the counties aren't doing basic reconciliations?"