Calibration of auditors' probabilistic judgments: Some empirical evidence

Abstract

Two sets of auditors employed by public accounting firms participated in an experiment designed to measure the calibration of their prior probability distributions (PPDs) regarding financial statement account balances. Using six realistic cases, a total of 341 usable PPDs were elicited through the cumulative distribution function (CDF) fractiles method. Relative to most of the findings reported in the calibration literature, the results of this study indicate much less overconfidence and a tendency toward underconfidence. The findings suggest that auditors may be prone to gather more objective evidence than is necessary to achieve a desired level of audit risk. The paper considers possible reasons for and implications of these findings.