Simple and accurate one-sided inference based on a class of M-estimators

Research output: Contribution to journalArticleResearchpeer-review

6 Citations (Scopus)

Abstract

This paper presents a simple method of calculating one-sided p-values which yield accurate inferences for a simple null hypothesis or, equivalently, accurate one-sided confidence limits regarding a scalar parameter in the presence of nuisance parameters. The method, which extends the work of DiCiccio et al. (2001) and Lee & Young (2005) in the context of adjusted likelihood estimation, is based on a test statistic analogous to the signed root of the loglikelihood, but derived from the objective function of an M-estimator from a certain class. Monte Carlo simulation is used to avoid the need for onerous analytical calculations typical of competing procedures based on Edgeworth or saddle-point approximations. The specific class of M-estimators under consideration is characterised by the requirement that the associated test statistic has constant variance to second order. This class contains, among others, the maximum likelihood estimator as well as variants of commonly used M-estimators of location, such as Huber's (1964) Proposal 2.

Original languageEnglish
Pages (from-to)973-987
Number of pages15
JournalBiometrika
Volume93
Issue number4
DOIs
Publication statusPublished - Dec 2006
Externally publishedYes

Fingerprint

M-estimator
statistics
Statistics
saddles
Test Statistic
One-sided limit
Maximum likelihood
Saddlepoint Approximation
Confidence Limits
Nuisance Parameter
testing
p-Value
Signed
Null hypothesis
Maximum Likelihood Estimator
Likelihood
Monte Carlo Simulation
Objective function
methodology
Roots

Cite this

@article{8ac07fc7db0a411297cabd806daf51c9,
title = "Simple and accurate one-sided inference based on a class of M-estimators",
abstract = "This paper presents a simple method of calculating one-sided p-values which yield accurate inferences for a simple null hypothesis or, equivalently, accurate one-sided confidence limits regarding a scalar parameter in the presence of nuisance parameters. The method, which extends the work of DiCiccio et al. (2001) and Lee & Young (2005) in the context of adjusted likelihood estimation, is based on a test statistic analogous to the signed root of the loglikelihood, but derived from the objective function of an M-estimator from a certain class. Monte Carlo simulation is used to avoid the need for onerous analytical calculations typical of competing procedures based on Edgeworth or saddle-point approximations. The specific class of M-estimators under consideration is characterised by the requirement that the associated test statistic has constant variance to second order. This class contains, among others, the maximum likelihood estimator as well as variants of commonly used M-estimators of location, such as Huber's (1964) Proposal 2.",
author = "Stern, {Steven E.}",
year = "2006",
month = "12",
doi = "10.1093/biomet/93.4.973",
language = "English",
volume = "93",
pages = "973--987",
journal = "Biometrika",
issn = "0006-3444",
publisher = "Oxford University Press",
number = "4",

}

Simple and accurate one-sided inference based on a class of M-estimators. / Stern, Steven E.

In: Biometrika, Vol. 93, No. 4, 12.2006, p. 973-987.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Simple and accurate one-sided inference based on a class of M-estimators

AU - Stern, Steven E.

PY - 2006/12

Y1 - 2006/12

N2 - This paper presents a simple method of calculating one-sided p-values which yield accurate inferences for a simple null hypothesis or, equivalently, accurate one-sided confidence limits regarding a scalar parameter in the presence of nuisance parameters. The method, which extends the work of DiCiccio et al. (2001) and Lee & Young (2005) in the context of adjusted likelihood estimation, is based on a test statistic analogous to the signed root of the loglikelihood, but derived from the objective function of an M-estimator from a certain class. Monte Carlo simulation is used to avoid the need for onerous analytical calculations typical of competing procedures based on Edgeworth or saddle-point approximations. The specific class of M-estimators under consideration is characterised by the requirement that the associated test statistic has constant variance to second order. This class contains, among others, the maximum likelihood estimator as well as variants of commonly used M-estimators of location, such as Huber's (1964) Proposal 2.

AB - This paper presents a simple method of calculating one-sided p-values which yield accurate inferences for a simple null hypothesis or, equivalently, accurate one-sided confidence limits regarding a scalar parameter in the presence of nuisance parameters. The method, which extends the work of DiCiccio et al. (2001) and Lee & Young (2005) in the context of adjusted likelihood estimation, is based on a test statistic analogous to the signed root of the loglikelihood, but derived from the objective function of an M-estimator from a certain class. Monte Carlo simulation is used to avoid the need for onerous analytical calculations typical of competing procedures based on Edgeworth or saddle-point approximations. The specific class of M-estimators under consideration is characterised by the requirement that the associated test statistic has constant variance to second order. This class contains, among others, the maximum likelihood estimator as well as variants of commonly used M-estimators of location, such as Huber's (1964) Proposal 2.

UR - http://www.scopus.com/inward/record.url?scp=33845725111&partnerID=8YFLogxK

U2 - 10.1093/biomet/93.4.973

DO - 10.1093/biomet/93.4.973

M3 - Article

VL - 93

SP - 973

EP - 987

JO - Biometrika

JF - Biometrika

SN - 0006-3444

IS - 4

ER -