Since 2005, I have worked with others, including Ron Baiman, Frank Stenger and Ron Rivest to develop the new field of election audit mathematics. On Thursday, March 15, I obtained the election results from 2002 and 2004 US House and Senate races and used it to calculate cost estimatess and success probabilities for manually auditing all federal elections. A cost analysis based on real election results disproves the prior analysis of the two persons who will be testifying on Tuesday before the House Admin Cmte on election audits, who formerly claimed that the Holt audit was less administratively burdensome than more effective election audits that have higher chance to detect vote fraud: Brennan Center Letter Supporting Holt's Audit
When evaluated using 2002 and 2004 election results, Holt's HR811 is only 10% Effective at detecting vote fraud and is More Costly than Election Audits designed to be 99% effective. Holt's audit is more administratively burdensome than more effective election audits! The reasons for the lack of effectiveness and costliness of the Holt/Clinton/Nelson/Tubb-Jones audit are explained in this paper:
The data for replicating the cost analysis is available here:
