Parameter Neutral Optimum Design for Non-linear Models
Journal of the Royal Statistical Society: Series B (Statistical Methodology) · 1997 · Volume 59 · Issue 4
Abstract
Some Bayesian approaches toD-optimum design of experiments are considered from the viewpoint of invariance under reparameterization of the underlying statistical model. An invariant criterion is proposed which does not require the detailed specification of a prior, and which is shown to be equivalent toG-optimality under a Jeffreys prior. The methods are applied and discussed in the contexts of exponential decay and quantal response models.
Keywords
Bayesian design, D-optimality, Equivalence theorem, G-optimality, Invariance, Jeffreys prior, Logistic regression, Non-linear model
Enable JavaScript for citation tools and additional paper interactions.