Efficient Model-Discriminating Designs
by
William Li
University of Minnesota

Recent progress in model-robust designs has been focused on maximizing the estimation capacities. Two competing models may be both estimable and yet hard to be separated in the model selection procedure. We evaluate the commonly used orthogonal designs in terms of the model-discriminating capabilities. The best model discriminating designs are selected from the orthogonal designs with small run sizes. The performances of these designs are compared with those that optimize the estimation capacities.

Date received: September 19, 2003