Testing Post-Election Audit Methods

Abstract

Listen

Translate article

Three proposed methods of post-election audits were tested for efficiency and effectiveness using 200 random audit sampling trials for each method of each randomly simulated election having 20K, 100K, and 250K total ballots cast in very close, close, semi-close, and not close elections — with both correct and incorrect reported election outcomes. One risk-limiting election integrity (EI) ballot comparison method, one risk-limiting EI precinct comparison method, and two ballot sampling methods — the ballot-polling audit (BRAVO) and the split-the-difference (STD) ballot polling methods — were tested and compared. Results show that EI risk-limiting ballot comparison audits are a more effective and efficient method of auditing elections than either ballot polling method. However, the STD ballot comparison method is more effective per ballot audited than the BRAVO method.