The American Institute of Certified Public Accountants has recently set forth significant revisions in its nonstatistical audit sample‐size decision aid (AICPA 1999). In a controlled setting involving 149 experienced auditors, we test the effects of the new guidance on auditors' sample‐size judgments, extending Kachelmeier and Messier's (1990) (KM) investigation of a previous AICPA (1983) decision aid.
We find that the current decision aid results in significantly smaller sample sizes than the previous aid. Further, auditors continue to “work backward” in their choice of decision aid inputs, resulting in sample sizes that are more intuitively acceptable. An optional supplemental worksheet added to the AICPA's guidance to assist the auditor in specifying tolerable misstatement generates a marginal increase in sample sizes, but does not eliminate the working‐backward phenomenon. However, the supplemental worksheet significantly reduces sample size variability. Additional findings update the conclusions in KM by showing that the excess of decision‐aided sample sizes over intuitive sample sizes in their study no longer applies. A final extension addresses a limitation in KM by showing that sample‐size judgments are not sensitive to the variation of population size as a separate treatment factor. Overall, this study directs focus to an improved understanding of nonstatistical sampling judgments, which are of increasing importance in the contemporary audit environment.