This weekend, the Columbus Dispatch did an
analysis that they claim shows that the voting machine shortage disproportionately affected
suburban Republican voters. It claims that, because more votes per machine were recorded in suburban wards, these machines were busier. Busier machines lead to longer lines. The analysis, also cited by the
Election Law site at OSU, is flawed on a number of levels.
1) It assumes that the number of votes cast per machine is a good approximation of machine shortage. The problem with this is that machines were operating at full capacity all day, all over. The average number of votes per machine in Franklin County was 170. At 5 minutes per vote (not atypical given the number of ballot initiatives), voting will finish an hour after the polls close. Differences in votes/machine will not show up until then.
2) If most people are not willing to wait more than two hours to vote, and machines typically have a two hour wait, exactly the same number of people will vote at each machine, regardless of how many people looked at the line and didn't vote.
3) The Dispatch Map is based on ward level data, even though the number of precincts per ward varies considerably. For instance, Ward 10 has 22 precincts. Kerry only won 6 of them. He won those 6 precincts 1515 to 819. You could have 200 votes per machine at these precincts, 140 votes per machine in the other precincts, and still only average about 150 votes per machine in the ward.
4) There were a number of lengthy proposals on the City of Columbus Ballot. This made the average time to cast a ballot within the city limits longer than outside of the city limits. Granted, on the one hand this gives an alternate explanation for long lines. On the other hand, it makes the city/suburb comparison ridiculous. What needs to be done is to check votes/machine (or better yet, registered voters per machine) by the Kerry vote percentage, for each precinct within the city limits. Then do a separate analysis of the same type for the suburbs. That's how you provide evidence for/against discrimination.
I am sure that the Dispatch analysis will be cited often as proof that the distribution of voting machines was not discriminatory. I haven't come close to proving that they were, but the evidence presented is extremely weak and misleading.