An audit plays an important role as a corrective measure for verifying the election outcome. In this research, we define a ballot-polling risk-limiting audit procedure for the selection of Condorcet winner. We particularly look into BRAVO, a risk-limiting audit that examines samples per randomly-drawn individual ballot. We further conduct an experiment to investigate the effectiveness and the efficiency of BRAVO for the selection of Condorcet winner. The results show that BRAVO is effective in confirming both cases of a correct and an erroneous reported winner within 1% and 5% risk limit. To confirm a correct reported winner, BRAVO generally requires less number of ballots compared to the estimated average sample number. It also confirms that both the risk limit and the margin of victory have an inverse relationship with the number of drawn ballots.
展开▼