'Phantom Votes' and 'Fleeing Voters' Haunt 2008 Election in Allegheny County

Last Friday, MESHers Donna Grandinetti, Chris Mlsna, Mary Ellen Kennedy Ralph, and I participated in the audit of the 2008 Presidential Election at the County Office Building. Organized by Richard King of VoteAllegheny, the audit seeks to determine just how much of a threat the electronic voting machines used by Allegheny County -- you know, the ones that do not produce a paper ballot -- are to the integrity of the electoral process. Translation: How close is Allegheny County to becoming a Banana Republic?

Our goal was simple: find the number and location of "phantom votes", when the machine records more e-ballots than registered voters. Here's what we did: the judge of each polling place seals several documents in an envelope: the "zero tape," which shows that the voting machines (one tape is run for all machines at a polling place as they are linked with a cable) were cleared of votes from the previous election and are storing no votes); the "results tape," which shows the number of e-ballots counted in total and for each election; the handwritten and hand-numbered list of voters; compilation sheets that show the total number of voters signed in, e-ballots, absentee ballots, provisional ballots, and emergency ballots. If there is a difference between the number of e-voters and the number of voters signed in on the register, election officials are to provide an explanation.

Some of the explanations seem innocent enough. Blind voters vote with an audio ballot and braille keypad (how cool is that!), but because it's not done very often election officials thought it may have resulted in counting ballots twice. Most of the time though, election officials just couldn't explain how there could be more e-ballots than voters.

How often did this happen? Donna and I audited 165 districts covering 71,739 voters. We found a total of 18 phantom ballots at 12 different districts. No one district had more than 3 ballots and most just had one. That's an error of 0.024%, which may not seem like much, but on a million votes is the difference in the Minnesota senate race: 237 votes.

The other problem we noticed -- and noted just because it bothered me -- was the incidence of "fleeing voters" (the term used by election officials for voters who sign in to vote, but whose ballots do not get counted). Maybe some "fled" because they remembered that they left something on the stove. What probably happened was that they didn't push the button to confirm their ballots. That means that 101 ballots weren't counted in our sample. In our fictional election with 1,000,000 voters, that's 1,408 voters, which as we have seen, can be the difference between war and peace, deprivation and prosperity, death and life.

Check back to this space. Dave Brown, a member of VoteAllegheny and MESH Standard Bearer on this issue, will provide further details on the audit, which should be finished in the next couple of weeks.