Academic journal article Auditing: A Journal of Practice & Theory

Reliance on Decision Aids: An Examination of Auditors' Assessment of Management Fraud

Academic journal article Auditing: A Journal of Practice & Theory

Reliance on Decision Aids: An Examination of Auditors' Assessment of Management Fraud

Article excerpt

Key Words: Management fraud, Decision aid, Expert system, Decision aid reliance, Red flags.

Data Availability: Contact the authors.

Statement on Auditing Standards (SAS) No. 82(1) (AICPA 1997) requires that the auditor assess the overall risk of material financial misstatement due to fraudulent financial reporting (management fraud) and that the auditor respond to this risk by designing an audit to provide reasonable assurance of detection. The presence or absence of red flags, over 30 of which are identified in SAS No. 82, can influence the auditor in assessing the likelihood of management fraud and have predictive value for assessing the likelihood of management fraud (Bell et al. 1993; Loebbecke et al. 1989). The risk assessment and audit planning response are important since too weak a response could jeopardize the effectiveness of the audit (i.e., could fail to uncover existing fraud) and too strong a response could result in an inefficient audit (McDaniel and Kinney 1995).

Prior research and documented audit failures indicate that auditors have difficulty assessing the likelihood of management fraud. SAS No. 82 gives little guidance on how the information in the red flags should be used to form a judgment of management fraud risk.(2) Auditors, most of whom have never encountered management fraud, rarely have the proper background that would lead to an ability to detect management fraud (Johnson et al. 1991; Loebbecke et al. 1989). In addition, multi-cue judgments of this type are inherently difficult for unaided human judges, even those considered to have a high level of expertise in the matter being judged (see Kleinmuntz (1990) for a review). Decision aids have been shown to outperform experts for multi-cue judgements (Kleinmuntz 1990; Libby and Libby 1989) so decision aids that make the firm's knowledge available to all auditors have the potential of improving auditor judgments of management fraud risk (Hackenbrack 1993).

The purpose of this study is to examine how three distinct types of decision aids influence auditors in assessing the risk of management fraud and in determining subsequent audit actions. The decision aids, which have been proposed for management fraud assessment, are: checklists (Pincus 1989), statistical models (e.g., a cascaded logit model) (Bell et al. 1993), and expert systems (Eining et al. 1990). All three of these decision aids use red flag cues and two of them, statistical models and expert systems, provide specific assessments of the risk of management fraud. While these assessments are likely to be superior to unaided human judgment based upon the same set of cues, recent studies suggest that decision makers are often hesitant to rely on decision aids, apparently placing more confidence in their own unaided judgment than in the aid (Ashton 1990, 1992; Arkes et al. 1986; Peterson and Pitz 1986). In those studies, the non-reliance on decision aids led to worse performance.

We conducted a laboratory study with auditors using their respective decision aid to assess the likelihood of management fraud for three situations with varying degrees of risk. Once each assessment was made, the auditors were asked to identify the appropriate additional audit steps. Our research focused on characteristics that would potentially increase reliance on the aid. These characteristics, referred to as constructive dialogue, were incorporated into the expert systems but not into the other two decision aids. All three decision aids were modeled after those used in practice, but presented in a common computerized form to enhance experimental control.

The results from our study provide strong support for the positive influence of the constructive dialogue in the expert system. The only difference between the expert system and the logit model was the inclusion of the constructive dialogue in the expert system, as both provided the same risk assessment for each case. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.