Enhancing Efficiency and Flexibility in Audits through Bayesian Optional Stopping

When auditors use statistical sampling, they typically plan the sample size based on the expected deviation. However, it can be challenging to accurately formulate this expectation in advance. For this reason, the expected deviations are often different than the actual deviations, leading to samples that are larger or smaller than needed, which is suboptimal. This article introduces Bayesian optional stopping as an approach that remedies this problem. In this approach, auditors start with an initial sample and then decide, step-by-step, whether to continue testing or stop based on whether sufficient evidence has been obtained. This results in more efficiency when the auditor's expectation is too high, and more flexibility when it is too low. Hence, Bayesian optional stopping ensures an optimal sampling plan that reduces unnecessary work while still achieving a substantiated conclusion.