In my Roll Call oped calling for the House to order a revote in the FL-13 race, I make the following statements:

A recent academic analysis by Laurin Frisina, Michael Herron, James Honaker and Jeffrey Lewis notes that in Charlotte County, the undervote rate in the state attorney general race was huge. In Charlotte — but not in Sarasota — the attorney general race was paired with another contest on the same electronic ballot page, mirroring the placement of the House race on the Sarasota County ballot. Fortunately for Floridians, the race for attorney general was not close.
A separate analysis recently completed by the Sarasota Herald-Tribune bolsters the poor ballot design hypothesis. Looking at every ballot cast in the county, the newspaper found that the people you’d most expect to vote in the House race — loyal party voters who voted Republican or Democrat throughout the ballot — were the ones most likely to have skipped voting in the race.

Walter Mebane of Cornell and David Dill of Stanford sent a message to the election law list calling into question the validity of the second study, that of the Herald-Tribune. At my request, Walter posted his analysis in pdf form here. Their bottom line:

The Herald Tribune analysis (“Analysis points to bad ballot design,” Matthew Doig and Maurice Tamman, Dec 5, 2006) is seriously misleading. The claim Hasen cites, that “loyal party voters who voted Republican or Democrat throughout the ballot — were the ones most likely to have skipped voting in the race,” is factually incorrect, indeed qualitatively wrong. In fact, while straight-party Democratic voters had dramatically higher undervote rates than average, straight-party Republican voters had dramatically lower undervote rates than average.

This is a very interesting development, raising the question why straight-party Democratic voters would have higher undervote rates (an issue Mebane and Dill don’t address). But it does not undermine the main point of my article: whatever the cause of this undervote (given that no one can credibly claim this was intentional undervoting on this scale), if the courts don’t grant a new election, the House should do so.