Isaac Newton Institute for Mathematical Sciences

Supersaturated Designs: Are Our Results Significant?

Authors: Robert W. Mee (University of Tennessee), David J. Edwards (Virginia Commonwealth University)

Abstract

We begin by assessing the null model performance of supersaturated designs when using all-subset and forward selection regression. The propensity for model selection criteria to overfit is highlighted. We subsequently propose a strategy for analyzing supersaturated designs that combines all-subsets regression and permutation tests. The methods are illustrated for several examples.