C. Neumann
In sensory experiments it is usual that each person receives a series of products one after the other and rates them. As these designs have more applications in general they are called crossover experiments (Kunert 1998).
Most papers on optimal crossover designs consider an estimate which is corrected for carryover effects. We look at the estimate for product effects, which is not corrected for carryover effects. If there are carryover effects, this estimate will be biased. On the other hand these estimators will be more precise.
We show that there is a symmetric design that minimizes the mean square error and the variance. This implies that the variance of the estimate is fixed for all competing designs.
It turns out that the optimal designs derived by Kushner (1997), which are optimal for the corrected estimate, are highly efficient for the uncorrected estimate. We suggest designing the experiment in anticipation of carryover effects but analyzing it without these effects.
Palabras clave: Design of Experiments, Optimal Design, Crossover Experiment
Programado
M05.1 Sesión Hispano-Alemana: Diseño Óptimo de Experimentos
6 de septiembre de 2016 11:00
0.02 - Aula de proyectos 1
