The Election Integrity Audit
FOR IMMEDIATE RELEASE: 9/25/06
ORGANIZATION: The National Election D ata Archive
The Election Integrity
The National Election Data Archive's Kathy Dopp and Frank Stenger developed a method for calculating election audit sizes that would detect any electronic miscount that could alter an election outcome, for all auditable voting systems, particularly when voters verify that their paper ballot records are correct.
Salt Lake City, Utah 9/25/06 - The elections industry is the only major industry not routinely subjected to independent manual audits. In any field, electronic mistakes or tampering would go undetected in the absence of audits. Banks, churches and businesses are subjected to routine independent audits to detect and prevent errors and embezzlement.
Virtually all U.S. votes will be counted electronically by 2008. Electronic voting systems make it possible for one person to rig a state-wide election and for electronic errors and failures to produce wrong results.
Vote counts determine who controls budgets of millions to trillions of dollars. Yet only perhaps fifteen states conduct any audits – manual checks - of vote count accuracy and their procedures are not sufficient to detect wrongfully altered outcomes in close races. In some states audit procedures are not independent; audit results are not made public; audit results are not used to correct discrepancies found between manual and electronic vote counts; or the audits are performed after election results are certified.
The lack of detailed election audits is especially alarming in light of the fact that U.S. jurisdictions publicly report vote counts aggregated in a way that hides evidence of vote count errors, machine problems, and tampering. Forensic analysis of election data - broken out by precinct and by vote type – could be used to reveal significant irregular patterns that result from partisan vote padding, under-votes, or vote-switching. However, because no jurisdictions routinely release these detailed election results, candidates would need to contest their elections; make legal open records requests to obtain the data; and have it analyzed – all prior to conceding. (See http://electionarchive.org/ucvAnalysis/US/election_officials/ElectionArchive_advice.pdf)
Solution: The National Election Data Archive (NEDA)'s Kathy Dopp and Frank Stenger developed a solution for calculating vote count audit sample sizes to ensure electronic vote count integrity in all auditable voting systems, particularly when voters verify that their paper ballot records are correct. Properly calculated and conducted audits – hand counts of paper ballots performed to check the accuracy of electronic vote counts - can be performed in jurisdictions using audit-able voting systems.
"The Election Integrity Audit" method has been supported by the work of Roy Saltman, who was instrumental in initiating the first federal engineering and procedural performance standards for voting systems and who proposed similar election audit size calculations in 1975 (See Appendix B in http://csrc.nist.gov/publications/nistpubs/NBS_SP_500-30.pdf); and by Ron Rivest of the Computer Science and Artificial Intelligence Laboratory at the Massachusetts Institute of Technology whose September 2006 paper "Estimating the Size of a Statistical Audit" proposes similar election audit size estimation that can be done with a calculator (See http://theory.csail.mit.edu/%7Erivest/Rivest-OnEstimatingTheSizeOfAStatisticalAudit.pdf); and by Howard Stanislevic who proposed a similar method in August 2006. (See http://www.votetrustusa.org/pdfs/VTTF/EVEPAuditing.pdf) The election audit calculation work of Dopp and Stenger is new because it provides a numerical method to directly calculate the optimal election audit size.
The closer a race, the smaller the amount of vote miscount that could alter the outcome; and the harder the miscount is to detect. Therefore, races with closer margins between winners and losers require larger audits. Election officials and candidates can use a program available on ElectionArchive.org to calculate the optimal audit size to verify the accuracy of any election outcome at http://electionarchive.org/auditcalculator/eic.cgi
The National Election Data Archive is a not-for-profit, nonpartisan organization devoted to promoting scientific means to ensure election-outcome accuracy. Its "Election Integrity Audit" paper can be found at http://electionarchive.org/ucvAnalysis/US/paper-audits/ElectionIntegrityAudit.pdf.
CONTACT: Kathy Dopp (435) 658-4657 kathy@ElectionArchive.org