Regression Analysis As Analytical Procedures: A Simulation Study Of The Effect Of Error Term Nonnormality On Auditor Decisions

Abstract

Regression analysis has been shown through research to be a useful audit tool. One underlying assumption of regression theory is error-term normality. Decision rules were applied to models developed using simulated data with normal and nonnormal error terms. Material errors were seeded into the audit period data, and incorrect rejections and acceptances were tabulated for both types of models. The results of this study suggest that nonnormality of error terms may not significantly affect auditor decisions.