|< Day Day Up >|| |
Traditionally, audits have been a team effort in which new members learn from older members. Accounting information systems can be used to reverse these roles because the younger auditors are more at ease with using technology. O'Callaghan (1994) argues that the combination of effective hands-on training with theory helps end users to develop good mental models of systems. Markovitch (1995) believes that end users which get good user support services should be more productive if the support services are planned around the user's (auditor) specific needs.
During the auditing process, auditors select audit procedures that are easiest to perform which take the least amount of time (O'Callaghan, Walker, & Sale, 1998). In reality, there are no auditing software applications for detecting and preventing fraud. Auditors may revert to word processors, spreadsheets, and calculators to decipher the firm's myriad of accounting information. However, more advanced information technology methods are available for deterring fraud. For example, artificial neural networks (ANN) could be used for data mining and detecting key indicators of fraud, decision support systems (DSS) could be used for decision making, expert systems (ES) could be used for rule bases, and discriminant analysis could be used to predict fraud. Rarely, if ever are these tools employed.
The objective of this research is to apply several of these advanced methods in an effort to detect and predict fraudulent behavior. Discriminant analysis will be applied to survey data in order to predict fraud according to key indicators such as poor internal controls, weak ethics policies, changes in employee lifestyles, working conditions, morale, and downturns in the economy (Turpen & Messina, 1997). ANN will then be used to data mine which factors are indicative of fraud. A comparison will then be made between the two methodologies.
|< Day Day Up >|| |