Sequential procedures for multiple responses factor screening

This paper considers the factor screening problem with multiple responses for simulation experiments. The objective is to identify important factors with controlled Family-Wise Error Rate. We assume a multiple-response first-order linear model, the responses follow a multivariate normal distribution, and estimated effect coefficients also follow multivariate normal distribution. Two likelihood ratio based procedures, Sum Intersection Procedure (SUMIP) and Sort Intersection Procedure (SORTIP), are proposed and verified. Numerical studies are provided to demonstrate the validity and efficiency of our proposed procedures.