Funny Numbers

SOMEWHERE IN AMERICA,there are local elections to be won for progressive causes. There are local Democratic leaders who need to be thrown out of their jobs. There are Christians to be wooed from the anti-gay marriage tent and gathered under the banner of real moral issues, such as universal health care and minimum wage.

But first, say a number of grassroots activists, statisticians and that nattering nabob of progressivism himself, Ralph Nader, we need to make sure our Democratic candidate really lost this time, fair and square. “We never had a lot of confidence in Kerry, but we still think he’s entitled to an accurate vote count,” said Kevin Zeese of Nader-Camejo for President, which has taken on the responsibility of verifying New Hampshire’s results after receiving “2,000 complaints by fax” of strange anomalies. “We’ve got to stop the coronation before it goes too far.”

If it sounds like a mathematical manifestation of the denial stage of the grief process, that’s maybe what it is. Even so, say various investigators, some of the evidence is too powerful to ignore. To wit: In Lafayette County, Florida, only 845 of the 2,755 registered Democrats voted for Kerry. In Warren County, Ohio, Bush picked up 12,000 votes after county election officials kicked out all independent observers, citing “homeland security” concerns. And in Guilford, North Carolina, a re-tallying of the votes produced 22,000 uncounted votes for Kerry after a huge turnout crashed the machines.

To add to the suspicions, most of the technology, both touch-screen machines and optical scanners, used in the 2004 election is owned and operated by two companies, the Texas-based Diebold Election Systems, and Election Systems and Software (ES&S), both of which have ties to Republican sympathizers: Diebold’s CEO, Walden “Wally” O’Dell, once infamously announced his intent to deliver Ohio’s votes for Bush; ES&S is a subsidiary of American Information Systems, formerly chaired by Chuck Hagel, who in 1996 defeated a popular sitting governor to become Nebraska’s first Republican senator in 24 years.

And, finally, there’s the crushing defeat itself, an election result that, if not quite the mandate Bush trumpets it to be, is at least a far more solid victory than anyone expected. “It’s a statistical impossibility that Bush got 8 million more votes than he got last time,” wrote media critic and New York University professor Mark Crispin Miller in a posting on Salon.com. “I would have to hear a much stronger argument for the authenticity, or I should say the veracity, of this popular vote for Bush before I’m willing to believe it.”

The most compelling data set has been circulated by Kathy Dopp, founder of Utah Counts Votes, who has posted a comprehensive spreadsheet on the Web site UsTogether.org examining the Florida results by county, voter-registration rolls and voting-machine technology, touch-screen or optical scan. Dopp holds a master’s in mathematics from the University of Utah and has been tracking the electronic-voting-machine problem since Bev Harris hacked Diebold’s code in 2003 and wrote her book, Black Box Voting: Ballot Tampering in the 21st Century.This year, she crunched the numbers in various patterns looking for trends and anomalies.

“A few days before the election,” Dopp told me over the phone from her home in Park City, Utah, “I thought it might be useful to see if I could examine election results by voting machine and by county, and if I could see any patterns that highlighted errors or electronic failures, so I started thinking of ways that could be done. And then the day after the election, as soon as the results were available, I started putting the numbers into spreadsheets.” She was surprised at what she found. “I was sort of expecting to find an anomalous pattern in the electronic voting machines,” she said. “But instead I found it in the optical-scan machines,” which translate paper ballot results into computer code. By comparing voter-registration rolls by party affiliation to the actual vote results, Dopp unearthed some dramatic surprises. In Baker County, Florida, for instance, where 83.3 percent of voters have registered as Democrats, 5,610 people came out to vote for Bush, and only 2,392 for Kerry, meaning that 64 percent of the county’s registered Democrats crossed party lines to vote for Bush. In the 88 percent Democrat county of Calhoun, Florida, Bush got more than 55 percent of the vote.

FASCINATED BY DOPP’S NUMBERS, I did my own, much simpler calculation — comparing the votes Gore received in 2000 to Kerry’s numbers in 2004. The results were less startling: Kerry got 9 percent less of the vote in Baker than Gore did four years ago, and 2 percent less in Calhoun. Using that same formula, comparing Gore’s votes in 2000 to Kerry’s in 2004, even Warren County, Ohio, looks less suspicious. It’s true that Bush scored a 22 percent increase over his 2000 tally. But Kerry also picked up some 6,000 votes over Gore — a 33 percent increase in support for the Democratic candidate. Significantly, Democratic voter registration grew from 6 percent to 9 percent in that Republican stronghold.

When I sent Dopp my figures, she replied in an e-mail that I had erred by assuming the 2000 election was a clean one. “Do you know the error rates in the counties for the 2000 recounts versus the official results?” (In Lafayette County, Bush and Gore each gained exactly one vote in the recount.) Later, on the phone, she told me, with some measure of irritation, that she wasn’t really interested in the 2000-2004 comparison. “I’m trying to do scientific mathematical analysis,” she said firmly. “I have no way on the planet of knowing what really happened in those places.” Nor is she trying to find out. “The beauty of my method is that I have not aggregated the numbers — I’m only interested in looking at statistical patterns, looking for ways to pinpoint counties with unexplained anomalies, counties whose votes might need to be recounted. When we determine through this analysis that there are counties with unexplained phenomena, we can use the statistics we’ve got to request [Freedom of Information Act] recounts if they’ve got paper ballots. Then we can potentially put a system in place by 2006,” Dopp said, “so no candidate would ever concede an election without knowing what was possible and where to ask for a recount first.”