A comparative evaluation of belief revision models in auditing

External auditors typically gather audit evidence in a sequential fashion a nd revise their estimates of likelihood of material misstatements based on the evidence collected. Optimal utilization of audit evidence can help cont rol audit risk and improve audit efficiency and effectiveness. This paper first shows how a typical audit risk assessment and belief revis ion task can be modeled using four theoretical models of belief revision. T hen the descriptive properties of these models are evaluated based on the a ctual judgments of experienced auditors who assessed the likelihood of erro r in a task involving inventory valuation. A realistic audit case was admin istered to 101 experienced auditors. Models based on the following theories were evaluated: a version of Bayesian inference labeled Cascaded Inference Theory (Schum 1987; Schum and DuCharme 1971), two versions of the Belief A djustment Model (Hogarth and Einhorn 1992), and a Version of the Dempster-S hafer Theory of Belief Functions (Srivastava and Shafer 1992; Shafer 1976). The experimental task involves belief revision after combining evidence abo ut the reliability of a client's internal control system with substantive e vidence relating to inventory pricing. Both the experimental and manipulati on check results show that the auditors were sensitive to differences in th e source reliability and diagnosticity of the manipulated audit evidence. T he analytical results reveal that the four models differ in the way they in terpret evidence relating to internal control system reliability and the ma nner in which such evidence is aggregated with price test evidence. Yet, al l four models correctly predict the direction of auditor belief revision. T he study reveals that important structural differences in the models result in differences in the magnitude, but not the direction of belief revision. This theoretical and empirical evidence has heretofore been unavailable in the literature. Further, the version of the Hogarth and Einhorn (1992) Bel ief Adjustment Model that views control systems reliability as negative evi dence is the only model that captures both the direction and magnitude of a uditors' belief revision. Finally, auditors' belief revision was lower than that predicted by the remaining three models, with the extent of discounti ng ranging from 31 percent to 40 percent. Under-utilization of evidential V alue with respect to any of the models implies opportunities to improve aud it performance either through training or with decision aids. The paper dis cusses implications for both audit theory and practice.