This is a reponse to comments and analysis by Walter Mebane et al. posted as a letter on Common Dreams.

The small counties Mebane refers to were already excluded from our statistical analysis which compared medium-sized counties only. Mebane et al again documented the effects of county size on voting patterns which many people have noticed.

Comparison to 2000 data is not sufficient to make a conclusion, and further historical comparisons of other elections and other comparisons are necessary. We plan to test our analyses against real data.

We believe that the analyses we have posted raise important questions about the vote count that require further study; we never claimed or believed that they prove fraud. Vote counting errors can be from many causes, and our group is not investigating the causes of vote count errors; we are merely looking at patterns that suggest where they may be occurring.

Our statistical group plans not only historical comparisons, but comparisons between the presidential race totals and those of U.S. Senate and Congressional races to explore for consistency or inconsistency. Mebane et al are responding to an article in which the author drew conclusions that the ad hoc investigators here have not made. Mebane et al.'s declaration that there is no evidence of errors is premature. At this time we are investigating the question of likelihood of errors in the opt-scan tabulations in medium sized counties's paper ballots. Many of these counties are in the center and south of Florida.

The only way to demonstrate or disprove errors in the opti-scan tabulations is to compare our predictions with actual ballot recounts. We hope that Mebane et al will join with us and advocate that a number of the medium size counties paper ballots for the opti-scan system be recounted manually once our group finishes its analyses.

Marc Sapir MD, MPH, Executive Director, Retro Poll
with much editing from Kathy Dopp et al.