pith. sign in

arxiv: 1904.09232 · v1 · pith:7Z7HWNG4new · submitted 2019-04-06 · 🧮 math.ST · stat.ME· stat.TH

Analytic solutions for locally optimal designs for gamma models having linear predictor without intercept

classification 🧮 math.ST stat.MEstat.TH
keywords designsgammamodelinterceptmodelswithoutlinearlocally
0
0 comments X
read the original abstract

The gamma model is a generalized linear model for gamma-distributed outcomes. The model is widely applied in psychology, ecology or medicine. In this paper we focus on gamma models having a linear predictor without intercept. For a specific scenario sets of locally D- and A-optimal designs are to be developed. Recently, Gaffke et al. (2018) established a complete class and an essentially complete class of designs for gamma models to obtain locally D-optimal designs. However to extend this approach to gamma model without an intercept term is complicated. To solve that further techniques have to be developed in the current work. Further, by a suitable transformation between gamma models with and without intercept optimality results may be transferred from one model to the other. Additionally by means of The General Equivalence Theorem optimality can be characterized for multiple regression by a system of polynomial inequalities which can be solved analytically or by computer algebra. By this necessary and sufficient conditions on the parameter values can be obtained for the local D-optimality of particular designs. The robustness of the derived designs with respect to misspecifications of the initial parameter values is examined by means of their local D-efficiencies.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.