ROBUST DESIGNS FOR REGRESSION PROBLEMS**This work was partially supported by grants from the National Science Foundation.

Publisher Summary This chapter discusses robust designs for regression problems and presents a regression design setup given by Y(xi) = f(xi) + ɛi i = 1, …, n, where the errors {ɛi} are i.i.d. with mean 0 and constant variance σ2, xi ɛ [-1, 1], f is a function of class that is commonly a linear combination of specified functions f0, …, fk. The regression problem is concerned with inference about the (unknown) coefficients of these specified fjs and the associated design problem is to choose the xis in an optimal way for this inference. Numerous papers by Kiefer and others have been addressed to this problem.