You can find Michael’s analysis here. A snippet:

The key to understanding the difference between the two studies is that among Election Day voters, there are more strong Republicans than there are strong Democrats, 29,364 to 25,326 (also note there are fewer voters in the intermediate categories). The Sarasota Herald-Tribune reports total number of undervotes among the groups, while Mebane and Dill emphasize the rate of undervoting among the groups. It would seem that strong Democrats were nearly twice as likely register an undervote than strong Republicans. The newspaper’s emphasis on total numbers, rather than rates, masks this phenomenon simply because there are more strong Republican voters.

He concludes: “From the look at the overvote rates among types, I’d agree with Mebane and Dill that the analysis suggests that this is not simply a matter of poor ballot design affecting equally strong Democrats and Republicans. Something affected strong Democrats at nearly twice the rate as strong Republicans.”