Testing for Collinearity using Bayesian Analysis

Testing for collinearity continues to be a controversial issue in the literature. Multicollinearity detection criteria, such as variance inflation factor, often fail detect true extent of multicollinearity. In this article, we propose utilizing Bayesian approach an attractive alternative. Under approach, recommend comparing marginal posterior regression parameters under two different priors. If difference these priors is pronounced, one can surmise that harmful. The Kolmogorov–Smirnov test also used further evidence confirm whether significant.