Previous studies in auditing have proposed statistical analytic techniques to determine the presence of unusual fluctuations in financial data. However, these techniques use past financial data and/or other explanatory variables to compute expectation parameters. If past data were contaminated with errors or fraud, then the precision of developed expectations is questionable and this leads to an increase in Type II error. The current study introduces a new analytic technique known as the bootstrap regression (BREG) procedure in the context of Benford’s Law. The BREG procedure mitigates Type II error based on Benford parameters and exact confidence intervals to assess for the presence of unusual fluctuations in financial data sets. These parameters and confidence intervals are derived independently from the financial data subject to audit. In addition, the BREG procedure mitigates Type I error and the excessive power problem. The BREG procedure was applied to a wide range of data sets such as non-fraudulent, fabricated, allegedly fraudulent, and fraudulent data sets. The overall results demonstrate that the BREG procedure effectively and efficiently identifies the presence of data anomalies. Unlike the BREG procedure, other commonly used analytic techniques were either difficult to implement or yielded inconsistent results in the context of the fraudulent data.
Suh, Ik Seon and Headrick, Todd C. "An Effective and Efficient Analytic Technique: A Bootstrap Regression Procedure and Benford's Law." (Jul 2011).