An Investigation of Auditor Judgment in Analytical Review

Experiments involving variations on two analytical review task situations were used to assess practicing auditors' judgments. The first experiment required auditors to generate expected values and noninvestigation intervals given variations in the amount of audited information available and the presence or absence of unaudited information. Consistent with prior research, the auditors' judgments were biased in the direction of the unaudited information. However, this bias was moderated when additional audited information was available. The second experiment required auditors to extrapolate intuitively an expected value for an account, given six different deterministic time-series patterns. The auditors' extrapolations were more accurate for those time-series patterns that are more likely to be encountered in practice. That is, extrapolations were slightly more accurate for increasing trends than decreasing trends, and were more accurate for linear and logarithmic patterns than exponential patterns. Implications for research and practice are discussed.