Premium
Confidence Intervals for Scale
Author(s) -
Morgenthaler Stephan
Publication year - 1987
Publication title -
australian journal of statistics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.434
H-Index - 41
eISSN - 1467-842X
pISSN - 0004-9581
DOI - 10.1111/j.1467-842x.1987.tb00745.x
Subject(s) - estimator , computation , confidence interval , scale (ratio) , interval (graph theory) , mathematics , computer science , scale parameter , sample (material) , interval estimation , estimation theory , algorithm , mathematical optimization , statistics , physics , chemistry , quantum mechanics , combinatorics , chromatography
Summary This paper explains the approach to parameter estimation based on the idea of simultaneous models. Instead of using a single shape—as for example the normal distribution—a simultaneous model uses a finite number of distinct shapes F, G , etc. Such simultaneous systems are tools in gauging the finite sample behavior of estimators. And they can be applied in the design of an estimator with prescribed desirable properties. The problem considered in this paper is interval estimation for a scale parameter. We discuss among other things the computation of optimal estimators in simultaneous models and study more closely the case of protecting against heavy‐tailed error distributions.