SMABS 2004 Jena University
SMABS 2004 Home Organization About Jena Sponsors Links Contact SMABS Home

European Association of Methodology

Department of methodology and evaluation research

Jena University


© Webmaster

Contributions: Abstract

The theory of optimal designs for regression models usually assumes uncorrelated errors. However, there are many experimental situations in which this assumption is invalid. For example, the raw material used in an industrial experiment can vary considerably from one batch to another. To account for this variation, random block effects are included in the regression model. A similar model can be used for modelling responses from paired comparison and conjoint studies in marketing experiments. In the presentation, a general algorithmic approach will be presented to design efficient designs for this type of experiment. Also, a few theoretical results will be presented. Special attention will be paid to the design of experiments with blocks of size two.