On the number of support points of maximin and Bayesian D-optimal designs in nonlinear regression models

We consider maximin and Bayesian D-optimal designs for nonlinear regression models. The maximin criterion requires the specification of a region for the nonlinear parameters in the model, while the Bayesian optimality criterion assumes that a prior distribution for these parameters is available. It was observed empirically by many authors that an increase of uncertainty in the prior information (i.e. a larger range for the parameter space in the maximin criterion or a larger variance of the prior distribution in the Bayesian criterion) yields a larger number of support points of the corresponding optimal designs. In this paper we present a rigorous proof of this phenomenon and show that in many nonlinear regression models the number of support points of Bayesian- and maximin D-optimal designs can become arbitrarily large if less prior information is available. Our results also explain why maximin D-optimal designs are usually supported at more different points than Bayesian D-optimal designs.

[1]  N A Cressie,et al.  The underlying structure of the direct linear plot with application to the analysis of hormone-receptor interactions. , 1979, Journal of steroid biochemistry.

[2]  Kathryn Chaloner,et al.  A note on optimal Bayesian design for nonlinear problems , 1993 .

[3]  Linda M. Haines,et al.  Bayesian D-optimal designs for the exponential growth model , 1995 .

[4]  Weng Kee Wong,et al.  A unified approach to the construction of minimax designs , 1992 .

[5]  J. Kiefer.  General Equivalence Theory for Optimum Designs (Approximate Theory) , 1974 .

[6]  Lorens A. Imhof,et al.  E-optimal designs for rational models , 1996 .

[7]  Dieter Rasch,et al.  Optimum experimental design in nonlinear regression , 1990 .

[8]  Lorens A. Imhof,et al.  Maximin designs for exponential growth models and heteroscedastic polynomial models , 2001 .

[9]  J. Witmer,et al.  Nonlinear Regression Modeling. , 1984 .

[10]  Holger Dette,et al.  Bayesian D-optimal designs for exponential regression models , 1997 .

[11]  Linda M. Haines,et al.  A Geometric Approach to Optimal Design for One‐Parameter Non‐Linear Models , 1995 .

[12]  E. Walter,et al.  Robust experiment design via stochastic approximation , 1985 .

[13]  J J DiStefano,et al.  Multiexponential, multicompartmental, and noncompartmental modeling. II. Data analysis and statistical considerations. , 1984, The American journal of physiology.

[14]  H.-P. Liebig.  TEMPERATURE INTEGRATION BY KOHLRABI GROWTH , 1988 .

[15]  F. Pukelsheim.  Optimal Design of Experiments , 1993 .

[16]  K. Chaloner,et al.  Optimal Bayesian design applied to logistic regression experiments , 1989 .

[17]  Holger Dette,et al.  Designing Experiments with Respect to ‘Standardized’ Optimality Criteria , 1997 .

[18]  Kathryn Chaloner,et al.  D- and c-optimal designs for exponential regression models used in viral dynamics and other applications , 2003 .

[19]  Christine H. Müller,et al.  Maximin efficient designs for estimating nonlinear aspects in linear models , 1995 .

[20]  A. Cornish-Bowden.  Fundamentals of Enzyme Kinetics , 1979 .

[21]  H. Liebig,et al.  STATIC REGRESSION MODELS FOR PLANNING GREENHOUSE PRODUCTION , 1988 .

[22]  Holger Dette,et al.  Bayesian optimal one point designs for one parameter nonlinear models , 1996 .