Commonality Analysis: A Reference Librarian's Tool for Decomposing Regression Effects

Multiple regression is a widely used technique to study complex interrelationships among people, information, and technology. In the face of multicollinearity, researchers encounter challenges when interpreting multiple linear regression results. Although standardized function and structure coefficients provide insight into the latent variable ( ) produced, they fall short when researchers want to fully report regression effects. Regression commonality analysis provides a level of interpretation of regression effects that cannot be revealed by only examining function and structure coefficients. Importantly, commonality analysis provides a full accounting of regression effects that identifies the loci and effects of suppression and multicollinearity. Conducting regression commonality analysis without the aid of software is laborious and may be untenable, depending on the number of predictor variables. A software solution in R is presented for the multiple regression case and demonstrated for use in evaluating predictor importance.