Ridge Regression for Interactive Models

Abstract

An exploratory study of the value of ridge regression for interactive models is reported. Assuming that the linear terms in a simple interactive model are centered to eliminate nonessential multicollinearity, a variety of common models, representing both ordinal and disordinal interactions, are shown to have “orientations” which are favorable to ridge regression. Comparisons of the potential efficiency of ridge regression to that of ordinary least squares across a wide range of conditions clearly suggest the value of ridge procedures for many centered interactive models.