Auditors' Decision-Aided Probability Assessments: An Analysis of the Effects of List Length and Response Format.

Abstract Decision makers often face problems where they are motivated to explain unexpected events or patterns of information. An early phase of problem solving involves generating explanations that might explain the unexpected event. Simple decision aids that list potential explanations are currently available and are used in a variety of settings, including auditing. This article reports the results of an experiment examining the effect on auditors' probability assessments of two characteristics of this type decision aid: (1) the number of explanations included in the list (list length) and (2) whether the individual assesses the listed explanations individually in a one-step process or in aggregate as part of a two-step process (response format). Research suggests that individuals exposed to a list of explanations underestimate the probabilities of unspecified explanations relative to specified explanations. Psychologists refer to this as an estimation bias. The results indicate that auditors exhibited estimation bias. Auditors assigned a higher probability to error explanations that were specified in a list as opposed to when they were not specified in a list. The estimation bias was smaller under the response format in which auditors made aggregate assessments as part of a two-step process. Finally, two possible debiasing techniques, an exemplar approach and a retrieval approach, were employed in an attempt to mitigate the bias. The exemplar approach resulted in a small, but statistically significant reduction in estimation bias. The retrieval approach was not effective.