The 2006 election in Florida's 13th Congressional district was won by the Republican candidate by 369 votes, out of about 240,000. But an examination of the ballots revealed a disturbing discrepancy: In the Republican-leaning half of the district, about 3,000 ballots recorded no choice in the Congressional race (a normal number of so-called undervotes). In the Democratic part, however, 18,000 ballots recorded no choice in the race, apparently because of a problematic ballot design. A statistical analysis showed that with a properly designed ballot, the Democrat would have won. Arlene Ash will review that analysis and discuss the potential for using statistics to improve elections in the future. One key change would be for courts to recognize the need for remedy when the statistics reveal a failed electoral process.