Voter Fraud or Flawed Elections?
The e-mails keep pouring in. Please investigate voter fraud! Here's evidence the Republicans stole the election! We're watching you cover the election irregularities! A number of Americans – is the number growing? – believe George W. Bush only won the election because the voting was somehow rigged. And each day they disseminate via e-mail what they consider to be proof – or, at the least, reasons to be suspicious. In pieces for The Nation magazine, I've noted that there is good cause to worry about the integrity of a voting system that is overseen by partisan players and that relies in part upon paperless electronic voting machines that are manufactured by companies that are led by pro-GOP executives and that refuse to reveal the computer codes they use. But I've also cautioned against declaring that the potential for abuse means the system was abused to flip the results. Exit polls that differ from reported vote counts are not necessarily proof of foul play, and statistical analyses that seem to raise questions need thorough vetting before they are waved about as signs of chicanery.
Take one of the early arguments for the "stolen election." Shortly after E Day, a former high school math teacher named Kathy Dopp sent out a chart that showed George W. Bush faring unusually well in Florida counties that used optical scan voting machines. A-ha, some folks exclaimed, this chart demonstrated the vote had been fixed. A team of political scientists led by Walter Mebane, a professor of government at Cornell, then examined the votes in these counties and found they were consistent with a years-long trend of registered Democrats in rural counties voting for Republican presidential candidates. Their findings were disputed by some "stolen election" advocates. But the Caltech/MIT Voter Technology Project released a study that reached the same conclusion as the Mebane paper. And this past Sunday, the Miami Herald published the results of its investigation of this particular voting pattern. The paper noted:
Some wondered whether Florida's tally was corrupt, with one Internet site writing: "George W. Bush's vote tallies, especially in the key state of Florida, are so statistically stunning that they border on the unbelievable."For instance, in Union County, more than 75 percent of registered voters were listed as Democrat. The official vote count was 3,396 for Bush and 1,251 for Kerry. The Herald found 3,393 votes for Bush and 1,272 for Kerry – practically no difference. The results were the same in the other two counties. This hands-on exercise demonstrated that statistical analyses that rely on predicted outcomes based on voter registration figures can only prove so much. There's no substitute for inspecting actual evidence, such as ballots.
The Miami Herald last week went to see for itself whether Bush's steamroll through north Florida was legitimate. Picking three counties that fit the conspiracy-theory profile – staunchly Democratic by registration, whoppingly GOP by voting – two reporters counted more than 17,000 ballots over three days. The conclusion: no conspiracy.
The count of optical-scan ballots in Suwannee, Lafayette and Union counties showed Bush whipping John Kerry in a region where registered Democrats outnumber Republicans 3-1.
The Herald found minor differences with official results, most involving ballots that had been discarded as unreadable by optical-scan machines but in which reporters thought the voter's intent was clear.
Of course, that's the problem when it comes to votes cast on machines that produce no physical trace of a voter's decision. After the election, three graduate students at University of California at Berkeley ran the vote counts from three heavily Democratic counties in Florida that used touch-screen voting machines – Broward, Palm Beach and Miami-Dade – through a statistical model based on past voting patterns. They found what they call "ghost votes" for Bush – vote tallies that exceeded what could be expected given previous elections in these counties. In Broward County, the grad students maintained, Bush collected 72,000 of these "ghost votes." They pointed to electronic touch-screen voting machines as the cause of this and estimated that the 15 e-voting counties in Florida mistakenly yielded Bush between 130,733 and 260,000 "ghost votes." This pattern, they said, did not occur in counties that used optical scan voting machines.
According to Florida's official vote count, Bush won by 381,000 votes – more than the total of these "ghost votes." Still, the grad student study has been hailed by election results skeptics as reason to believe skullduggery transpired. Yet other experts in statistics have not been persuaded. Charles Stewart III, a political science professor at MIT and researcher at the Caltech/MIT Voting Technology Project, examined the data used by the Berkeley researchers and found what he calls "an interesting pattern." But, he told me, "it may not have anything to do with voting machines." He explained, "It is a baroque form of regression model they're using. Almost everyone I talk to says it looks like they were fishing for results. I would hope you'd find a lot of skeptics." Stewart pointed to two sets of precincts he examined in Palm Beach County. Both were heavily Democratic, one contained many African Americans, the other set had but a few. It was the set with few black voters that shifted dramatically toward Bush, according to Stewart. And this movement, he said, may be unrelated to the e-voting machines. These precincts, he speculated, could have had more Jewish voters who shifted toward Bush this election.
Andrew Gelman, a professor of statistics at Columbia University, also examined the Berkeley study and found that the statistical anomalies only were significant in two counties – Broward and Palm Beach – not all of the 15 e-voting counties. On his weblog, he notes that the Berkeley researchers "make some pretty strong causal claims which I would think should be studied further, but with some skepticism." Gelman observes, "Something unusual seems to have happened in Broward and Palm Beach counties in 2004. One possibility, as suggested by [the Berkeley researchers] is cheating." But he is quick to add, "I don't know what was going on in these counties, what else was on the ballot, etc., but an obvious alternative explanation is that, for various reasons, 3 percent more people in those counties preferred Bush in 2004, compared to 2000 ... [S]uch a swing would be unusual (at least compared to recent history), but that doesn't mean it couldn't happen! ... It would make sense to look further at Broward and Palm Beach counties, where swings happened which look unexpected compared to the other counties and compared to 2000, 1996 and 1992. But lots of unexpected things happen in elections, so we shouldn't jump to the conclusion that e-voting is related to these particular surprises." (Pollster John Zogby says that he does not believe that the "election was stolen," but he concedes it was an odd result: "51 percent of the voters gave Bush a negative approval rating; 51 percent voted for him.")
The Berkeley study is no slam-dunk. And the-election-was-rigged activists are raising other issues regarding the Florida vote count. When Bev Harris, a prominent critic of electronic voting who runs www.blackboxvoting.org, showed up at the elections office of Volusia County – where Kerry won by 3,723 votes – in mid-November seeking poll tapes for the optical scan voting machines used during the election, she found a set of the poll tapes discarded in a garbage bag. Was this part of a cover-up? Elections Supervisor Deanie Lowe told the Daytona Beach News-Journal that these election records were backup copies destined for a shredder. Harris and others fear there is more to the tale. And today Black Box Voting sued Teresa LaPore, the elections supervisor for Palm Beach County, to force her to turn over elections records. (The group is threatening to initiate similar lawsuits against 13 other counties in Florida and up to 80 counties in Ohio.)
In Ohio – where Bush's margin of victory was 136,000 – much organizing has been conducted by activists who question the final tally. As of yet, there have been no statistical studies of Ohio similar to the Berkeley paper. But e-mailers have zapped around a chart that supposedly shows 93,000 "extra" votes were cast in various municipalities in Cuyahoga County – that is, these areas listed more votes than registered voters. As I've reported earlier, county elections officials have what sounds like a good explanation for this. They claim their software has an odd glitch that assigns absentee ballots for a group of municipalities to one of the municipalities in the group. Consequently, on the spreadsheet posted by the elections office a particular municipality can end up showing more votes than registered voters. Cuyahoga elections officials – who work in an office run by a Democrat – insist there were no "extra" votes.
A recount is set to occur in Ohio as soon as Secretary of State Kenneth Blackwell, a conservative Republican who co-chaired Bush's campaign in the state, certifies the results (which is scheduled to happen this week). The recount was requested by David Cobb and Michael Badnarik, who were the presidential candidates for, respectively, the Green Party and the Libertarian Party. The Kerry campaign has not been involved, but it is watching. The state Democratic Party, though, is supporting the recount, citing complaints from Ohio voters that they faced long lines at the polls, encountered malfunctioning machines, and never received absentee ballots. "As Senator Kerry stated in his concession speech in Boston," Ohio Democratic Party chairman Dennis White remarked, "we do not necessarily expect the results of the election to change; however, we believe it necessary to make sure everyone's vote is counted fairly and accurately."
And claiming there was "fraud and [vote] stealing" in Ohio – without detailing the charges – Jesse Jackson has asked the state supreme court to consider setting aside the election results there. Jackson has called for a "thorough" investigation of voting irregularities. There does seem to be evidence of voter suppression in Ohio – and that does merit investigation. But it is far from clear that any of the alleged suppression tactics – such as not providing Democratic precincts enough functioning voting machines – cost Kerry over 136,000 votes. Nevertheless, a coalition of public-interest outfits calling itself the Ohio Honest Election Campaign has threatened to file a lawsuit to challenge the election results, asserting that thousands of Ohioans votes were incorrectly counted or not counted. And on Nov. 26, People for the American Way filed a lawsuit to challenge the rejection of 8,000 of 24,472 provisional ballots. RedefeatBush.com plans to bus in protesters for a rally in Columbus, Ohio, on Dec. 4 to support the recount.
A strong case that the election was stolen – either in Ohio or Florida – still has yet to be made. Statistical arguments are not convincing without concrete evidence (or widespread support among statistical experts). When reporters looked at actual ballots in Florida they found the armchair analysts were way off in their assumptions. And a recount requested by Ralph Nader in a limited number of precincts in New Hampshire – after Bush received higher than expected vote tallies in those parts of the Granite State – found little change from the original results. KPFK, the Pacifica radio station in Los Angeles, was a bit ahead of the facts when it issued a statement on Nov. 23 noting it was projecting that Kerry "has won the state of Ohio and thus the presidency by a minimum electoral college count of 272 to 266."
Yet the voting system is shaky enough to warrant serious concern. The General Accountability Office was right to agree to a request from Rep. John Conyers and four other Democratic House members that it investigate election irregularities in the 2004 election. According to these members of Congress, the GAO will examine the security and accuracy of voting technologies, distribution and allocation of voting machines, and the counting of provisional ballots. "All Americans, no matter how they voted, need to have confidence that when they cast their ballot, their voice is heard," the lawmakers said in a statement. Indeed. There are Bush critics who probably never will accept the Nov. 2 results. And the systemic problems that do exist – secretive voting technologies, the opportunity for partisan hacks to engage in voter suppression – will allow these people to hang on to their worst fears and to continue to share look-at-this! e-mails with fellow believers (or nonbelievers). But the evidence to date is that the election results were not rigged but were produced by a flawed system.