Dear SftP:   

You may not be aware of the excellent statement frm the American Statistical Association about the dubious election in Florida's district 13.  It should be helpful to anyone working for reform.  You can see it here:
http://www.amstat.org/news/pdfs/SPA_Statisticians_examine_one_race.pdf
but to make sure I'll copy it below.

Please use and circulate widely! -- John Lamperti

Here's the statemnt:


Do we really know who won one hotly contested Congressional election this November?

     Statisticians describe an electoral outcome that calls for a revote


            In one Florida Congressional district this November, nearly 240,000 votes were recorded in total for one or another of two candidates for the U.S. House of Representatives. Candidate A got 50.08% and Candidate B, 49.92%, a difference of about 400 votes overall ­ arguably too close to call even if only the usual errors were present.

              But this was not a usual race. About 18,300 people in one county, nearly 13% of those voting on touchscreens, had no vote recorded in this race. That is, they seem to have "undervoted." Legitimate undervotes occur when voters choose not to vote in some races ­ usually obscure races where they do not know the candidates. But people who trouble to go to the polls rarely skip a hotly contested federal race. For example, the undervote in this county and this race on (paper) absentee ballots was less than 2% - very much in line with both paper and machine voting in other counties, and what we historically expect.  However, machine-recorded ballots for this race ­ in this county only ­ had about 15% undervotes in the early voting and still, despite some warnings to poll workers from the Supervisor of Elections, over 10% undervotes on Election Day. Importantly, votes that were recorded here distinctly favored Candidate B.

            Faulty ballot design is a likely cause, since isolated other Florida counties had undervotes of as much as 22% in a different statewide race with similar screen placement as the House race in this county. Also, many voters in this county reported looking for the House race and not seeing it, or how to verify that their vote had registered. 

	Thus, well over 10,000 Sarasota County voters were disenfranchised by the voting machinery. How would this have affected the 400-vote "winning margin"?  Unfortunately, we cannot determine each apparent undervoter's intent because these machines had no paper audit trail.  However, we can project the likely impact of the lost votes. For example, the recorded vote in Sarasota County favored Candidate B by 5.5%; had the lost votes tracked the recorded ones, then Candidate B had more than enough votes to win.

	Which Candidate should win? Candidate A, with the most officially recorded votes, or B, who was probably the choice of more legitimate voters?  We want supporters of both Candidates to trust the outcome, but whoever wins, the apparent margin of victory is small compared to the number of lost votes. No re-examination of these deeply flawed data can fix that. 

	So how should the winner be decided? Fortunately, an entirely satisfactory remedy is available, and we strongly recommend it: "Do it over."

	Sincerely,

	David Marker, Chair
	Scientific and Public Affairs Committee
	American Statistical Association