A no-vote-left-behind kind of project

Proof that a PhD doesn’t actually turn the brain to mush: "This is a no-vote-left-behind kind of project, not a change-the-president project," said UC Berkeley sociology professor Michael Hout of the research he and his graduate students did on discrepancies between exit polls and election results. "We're as interested in the next election as the one just over." Tell that to the Washington Post and the Boston Globe, sneering about conspiracy theories and sore losers.

As I noted in an earlier entry, I was skeptical of the Caltech-MIT report dismissing discrepancies between exit polls and officially tallied returns because it was unclear whether in doing their analysis the professors had used the original exit poll data (released by CNN until about midnight and saved as screen shots by bloggers) or the “recalibrated” data, corrected to conform to actual poll results as they came in. That’s about as fair as betting the Red Sox would lose the World Series and then changing your bet when they’d won three games.

The Caltech study determined that the exit poll discrepancies were statistically insignificant, but made no mention of any recalibration. Turns out the Caltech crew did use the suspect recalibrated data and they’ve been widely slammed for it. They've also been slammed for not noting that computer tabulation of optically scanned ballots is electronic and just as much a source of potential fraud or computer error as fully electronic voting machines.

The Oakland Tribune and the AP advanced the story by asking one of the Caltech researchers to review the Berkeley study. MIT arts and social sciences dean Charles Stewart III, a researcher in the MIT-Caltech Voting Technology Project, found that the Berkeley team’s findings held up. He wouldn’t come to any conclusions about the cause for the discrepancies, but he said they warranted an investigation. The Berkeley study found that Bush may have incorrectly won between 130,000 and 260,000 “ghost votes” in Florida

I still have not seen any explanation for why some critics (Bev Harris, Thom Hartmann) have found discrepancies with optical scanners and others (the Berkeley team) have found them with touchscreen voting machines. So I'm staying agnostic about the meaning of all this funny business—which doesn't mean dismissal of questions about the election results. It means all the more investigation, please.

Margie Burns at the Baltimore Chronicle appears to have gotten hold of the original exit polls from polling firms Edison Media Research and Mitofsky International. Based on them, she says bluntly, “It seems unlikely that Mr. Bush won the election.” In 42 of 51 states, and in all the close states, the election results swung significantly to Bush from the exit polls, regardless of whether the state was decided for Bush or Kerry. In four states—Ohio, Florida, New Mexico, and Iowa—the shift from exit poll to published vote tally shifted the state from Kerry to Bush. These states add up to 59 electoral votes, enough to decide the election.

As Burns notes, exit polls have been widely used throughout the world to verify the legitimacy of an election. In former-Soviet Georgia, when exit polls contradicted official election results showing that Eduard Shevardnadze had been re-elected, popular outrage forced Shevardnadze to resign. Right now, the Ukraine is in crisis over its election, widely seen as fraudulent, in part because of discrepancies between official results and exit polls. Hundreds of thousands of Ukrainians have taken to the streets in protest, demanding the government annul the results of the election.