Author(s)

I. Ortiz & C. Rodriguez

Abstract

In this paper we are interested in finding D-optimal designs, when there is a special dependence among the observations (correlation). We try to relate the correlated case to the uncorrelated one, dealing with polynomial re- gression models which express the relation between response and controlled variables. An example is introduced to ilustrate the usefulness of the theo- retical contents in the article. 1 Introduction In many situations, it is assumed that the observations are indepen- dent and identically distributed. But, in some contexts this is not adequate. The observations can have different variances and they can be dependent, for example, if the measurements are taken se- quentially in time. Nevertheless, these dependences are typically un- known. In the assumption of de

Keywords