Paper v. Electronic Voting in Ohio

There have been an interesting couple of articles about error rates in paper ballots in Franklin County Ohio.  Here is the crux of the discussion:

Votes in central Ohio’s Franklin County were 2.5 times more likely to go unrecorded when they were cast on paper ballots instead of touch-screen machines, a newspaper analysis of the Nov. 4 election found.

Critics of paper ballots say the statistical discrepancy is the downfall of Ohio’s popular early voting system, which relies heavily on paper ballots that are read by a scanning machine.  Half of Franklin County’s voters ended up making their selections on a paper ballot because early voting was almost exclusively conducted on paper.

Roughly 1.4 percent of paper ballots cast had no vote recorded for president, while the amount of ballots cast electronically without a vote for president was less than 0.6 percent, the analysis by The Columbus Dispatch found.

“That sounds like the pattern of what we found across the country,” said David Kimball, a political science professor at the University of Missouri-St. Louis who helped conduct a study comparing voting methods.

Another Brunner spokesman, Bryan Clark, noted that the rate of undervotes in Franklin County for the presidential race was about the same for both paper ballots and touch-screen machines.  That suggests voters who used both systems intentionally skipped the race, and that voters who cast paper ballots didn’t mistakenly miss it.

Now think about this for a second.  According to the quote above, Bryan Clark expects me to believe that early voters turned out to vote early in the race to fill the remainder of the term for the State Attorney General (the top race on the ballot after president)?  This explanation is incongruous with previous research; as Bob Stein has noted:  “early voters appear to be more partisan, ideological, interested in politics, and disproportionately likely to have voted in the past.”  (The attorney general race also had a residual vote race 8 times higher than for the presidential race).  One would expect early voters to have the lowest residual vote rates, not a rate twice as high as the rate for election day voters.

Charles Stewart is the guru on the blog on the residual vote issue and it will be interesting to see how his analysis of residual votes turns out, especially in a place like Ohio.