Parameter Estimation when Various Models are Available

Abstract

The aim of this paper is to give an answer to the problem posed by data for which various models are possible, by means of a pre-test estimator using a likelihood criteria on exponential families. In the framework of contiguity, we will investigate the asymptotic behaviour of this estimator, comparing it with the classical ones. We illustrate those results on a practical example issued from segregation analysis in genetics.

AMS 1980 subject classification::

• 62 F12
• 62 P10

Keywords:

• Pre-test estimators
• contiguity
• segregation analysis