I’m attending the “How We Vote” conference organized by the William and Mary School of Law.
They’ve brought together a nice set of speakers, something of a list of the usual suspects, but a good mix of law, computer science, and political science speakers.
Walter Mebane opened the morning by presenting additional analysis of the 2006 FL 13 results. I’m not quite sure his results challenge results recently published by Frisina, Herron, Honaker, and Lewis in the Election Law Journal. They seemed complementary to me. What was really fascinating is that Mebane has gained access to what he called “machine events” such as power outages, unexpected downtime, resets, etc. What he finds, in part, are significant relationships in some counties between the level of machine events and balloting mode (early and election day). Put simply, many more problems (“events”) occurred on election day. Is this an argument for early voting, as a way to minimize machine anomalies?
Michael Herron followed with a paper (oddly enough, co-authored with Walter Mebane and Jon Wand) examining the 2008 NH primary, asking whether the final results could have been affected by the kinds of voting machines used across the state. Using an innovative ward matching method, Herron et al. show that the disparate results for Clinton by wards by machines is almost completely due to the kinds of voters who live in these different wards, not the machines.
More as the day goes on…