More Style, Less Work: Card-style Data Decrease Risk-limiting Audit Sample Sizes

U.S. elections rely heavily on computers such as voter registration databases, electronic pollbooks, voting machines, scanners, tabulators, and results reporting websites. These introduce digital threats to election outcomes. Risk-limiting audits (RLAs) mitigate some of these systems by manually inspecting random samples ballot cards. RLAs have a large chance correcting wrong outcomes (by conducting full manual tabulation trustworthy record the votes), but can save labor when reported are correct. This efficiency is eroded sampling cannot be targeted cards that contain contest(s) under audit. If sample drawn from all cast cards, then RLA sizes scale like reciprocal fraction That shrinks number per grows (i.e., more contests) ballots contest decreases smaller percentage voters eligible vote in contest). States conduct contests multi-card or small dramatically reduce using information about which contests—by keeping track card-style data (CSD). For instance, CSD expected draws needed audit single countywide 4-card 75%. Similarly, 95% for an two with same margin if one every other 10% ballots. In realistic examples, savings several orders magnitude.