A NOTE ON THE USE OF REGRESSION ANALYSIS IN ACCOUNTING: COPING WITH MULTICOLLINEARITY*

A major obstacle to the effective utilization of regression analysis in accounting practice and research is the presence of multi-collinearity. This paper analyzes the impact of multicollinearity on ordinary least squares (OLS) estimation, and describes a recently developed estimation procedure, ridge regression analysis (RRA), that offers the possibility of improved coefficient estimation. An auditing example is provided comparing the results of OLS to those of RRA.