2.1.1 Methodology
One of the assumptions that underlie traditional auditing methods is the presence
of an intelligent human being. When an auditor checks items in a sample, he or she is
able to apply human reason and common sense to transactions. Fraud investigations
often start with the auditor conducting a routine audit task, looking at a transaction, and
saying, “that doesn’t make sense.” This approach can be seen as an inductive approach;
the auditor investigates further when anomalies are found.
Data mining routines—run by computer—do not have this innate sense of
normality. Queries and scripts do exactly what they are programmed to do. They do not
“dig deeper” unless the user specifically programs them to do so. To accommodate this
limitation, the fraud hypothesis testing approach has been proposed (Albrecht, et. al.,
2000). This approach has also been labeled the deductive or proactive approach to fraud
detection; it involves the following sixstep
approach: