Simulation Study for Evaluating a New Three Stages Procedure for Selecting the Right Multivariate Regression Model

Abstract

Problem statement: This article considers the analysis of multivariate regression experiment that is used frequently in variety of applications research such as psychiatric epidemiologic studies. Our study concerned with multivariate regression model in which the responses were correlated in particular ways for both standard and non-standard multivariate model structures. Our objective is to find reliable procedure that can be used to guide the selection of the best multivariate regression model that has the right covariance structure and in the same time has the right multivariate model structure for both standard and non-standard multivariate model structures. Approach: In this study, we were proposing and evaluating a new three stages procedure that could be used to guide the selection of the best multivariate regression model that has the right covariance structure and in the same time has the right multivariate model structure using bootstrap simulation procedure. Results: The simulation results indicated that the performance of the new procedure in identifying the right multivariate regression model that has the right covariance structure and in the same time the right multivariate model structure from both standard and non-standard multivariate model structures was excellent overall. Conclusion/Recommendations: We recommended using the new procedure as stander tools to guide the selection of the best multivariate regression model that has the right covariance structure and in the same time has the right multivariate model structure.