Journal of Official Statistics, Vol.18, No.2, 2002. pp. 291–303
Multiple-Objective Optimal Designs for the Hierarchical Linear Model
Mirjam Moerbeek and Weng Kee Wong
Abstract:Optimal designs are usually constructed under a single optimality criterion. Such designs are not very realistic in practice because a researcher seldom has just one objective in mind when designing an experiment. This problem can be overcome by multiple-objective designs. Here, we extend previous work and construct multiple-objective designs for situations where the data are correlated using a hierarchical linear model. We present a graphical method for constructing multiple-objective designs and investigate their robustness properties.
Keywords:Hierarchical data; sample size; optimality criteria; variance components.
Copyright © Statistics Sweden 1996-2018. Open AccessISSN 0282-423XCreated and Maintained by OKS Group