The most rigorous and reliable methods for checking whether electoral outcomes are correct (i.e., whether the reported winners actually won) frame the problem in the opposite way: the electoral outcome is assumed to be incorrect (in an unknown way), absent convincing evidence that the reported winners really won. These “risk-limiting audit” methods can generate affirmative statistical evidence that results are correct, guaranteeing that if the result is incorrect, the chance the result would escape correction is at most some pre-specified value α (and the chance the audit will correct the result is at least 1 − α), for any desired α between 0 and 100% (Lindeman and Stark, 2012).