The assessment of management fraud risk is a complex decision process,
but one with which few auditors have had experience. As a result, dec
ision aids have been suggested to support this process. Unfortunately,
many times users do not rely on decision aids even when doing so woul
d improve the quality of the decision. This paper reports on an experi
ment that examines the use of an expert system decision aid created to
enhance the engagement of the user and increase reliance on the aid.
Auditors using the expert system exhibited the ability to better discr
iminate between situations with varying levels of management fraud ris
k and selected more consistent subsequent decisions regarding appropri
ate audit actions than did users of a decision aid that provided only
a suggested assessment, a legit statistical model. The legit model use
rs did discriminate better than both checklist users and unaided decis
ion makers.