Model-Robust Design for Multiresponse Linear Model with Possible Bias

This paper studies the design problem for the multi response linear model with possible bias. It is assumed that the fitted model for each response is polynomial of degree up to two, and the model bias includes the effects due to higher degree terms of multivariate Hermite polynomials. A criterion for choosing designs is proposed based on averaging the mean squared error over all possible bias. Several examples are given to illustrate the designs in R^s.