Using Weighted Ridge Regression as A Proposed Approach to Remedy Outliyngness and Multi-Collinearity in Mediation Analysis

Abstract

In this paper we propose to use the Robust Ridge Regression procedure to deal with some problems that the Mediation analysis suffer from, these problems include the existence of some outlier points in the independent variable or dependent variable or both of them. As well as the existence of linear dependency between independent and Mediated variables which produce in turn the existence of in-accurate estimators. So the use of Robust Ridge Regression procedure is proposed to remedy the extremeness exist in the points, in addition to solve the Multi-Collinearity problem exists between the mediated and independent variable, and to obtain an active formula that address these problems. Furthermore getting more accurate confidence intervals compared with that of ordinary least squares method. The proposed procedure implemented through simulation study and real data analysis.