AICPA Nonstatistical Audit Sampling Guidelines: A Simulation.

Abstract This study examines the audit sampling risks associated with the non-statistical sampling recommendations in the AlCPA's (1983) Accounting and Auditing Guide: Audit Sampling. Surveys indicate that the vast majority of practitioners use nonstatistical sampling approaches and that many look to this source for guidance. Yet, no validation of the guidance has appeared in the literature. To offer evidence on this matter, audit sampling models were designed that followed the recommendations of the Guide and incorporated varying levels of tolerable error, sample sizes that are frequently used in practice, and information from auditors on decision rules used in practice. Simulations were conducted using actual accounting populations which contained diverse error characteristics. The paper reports on the audit sampling risks of incorrect acceptance and rejection associated with the different audit sampling models used in this study. The results indicate that the selection of the decision rule used in evaluating the sample results is the most critical factor in applying the Guide's recommendations. There is a substantial difference in the mean audit sampling risks for the two decision rules that were analyzed. Even when the true error amount in the population is equal to 1.5 tolerable error, auditors may determine that the risk of incorrect acceptance with Decision Rule 1 is unacceptable at small sample size levels. The adjustment of the sample size, as suggested in the Guide, is a less important factor in controlling sampling risks.