Contributions to the theory of statistical breakdown

The theory of statistical breakdown is studied from two different angles. Firstly, the finite sample breakdown points of estimators are found to be inherently related to their tail performances. This connection provides persuasive new evidence for the importance of the breakdown point of estimators in the assessment of quantitative robustness. Secondly, breakdown robustness for statistical tests is studied via the power and level breakdown functions for the test functionals. This analysis clarifies and unifies various robust aspects in the context of hypotheses testing, and provides more informative robust measures for and new insights into statistical tests. Applications to location-scale problems, the one-way layout and the linear regression model are studied. Motivations and implications of these analyses are discussed.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Statistics

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2011

Author dc:creator

• He, Xu-ming

Contributors dc:contributor

• Portnoy, Stephen L.

• Statistics

Statement dc:rights

• Copyright 1989 He, Xu-ming

Language dc:language

eng

Identifier

AAI9010877
(UMI)AAI9010877

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/23662