Martingale-based residuals for survival models

Terry M Therneau, Patricia M. Grambsch, Thomas R. Fleming

Research output: Contribution to journalArticle

640 Citations (Scopus)

Abstract

Graphical methods based on the analysis of residuals are considered for the setting of the highly-used Cox (1972) regression model and for the Andersen-Gill (1982) generalization of that model. We start with a class of martingale-based residuals as proposed by Barlow & Prentice (1988). These residuals and/or their transforms are useful for investigating the functional form of a covariate, the proportional hazards assumption, the leverage of each subject upon the estimates of ß, and the lack of model fit to a given subject.

Original languageEnglish (US)
Pages (from-to)147-160
Number of pages14
JournalBiometrika
Volume77
Issue number1
DOIs
StatePublished - Mar 1990

Fingerprint

Survival Model
Martingale
Cox Regression Model
Proportional Hazards
Graphical Methods
Leverage
Covariates
gills
Hazards
Transform
Model
Estimate
Survival model
methodology

Keywords

  • Cox regression
  • Deviance
  • Influence function
  • Martingale
  • Outlier detection
  • Proportional hazards
  • Regression diagnostic
  • Residual

ASJC Scopus subject areas

  • Statistics, Probability and Uncertainty
  • Applied Mathematics
  • Mathematics(all)
  • Statistics and Probability
  • Agricultural and Biological Sciences (miscellaneous)
  • Agricultural and Biological Sciences(all)

Cite this

Martingale-based residuals for survival models. / Therneau, Terry M; Grambsch, Patricia M.; Fleming, Thomas R.

In: Biometrika, Vol. 77, No. 1, 03.1990, p. 147-160.

Research output: Contribution to journalArticle

Therneau, Terry M ; Grambsch, Patricia M. ; Fleming, Thomas R. / Martingale-based residuals for survival models. In: Biometrika. 1990 ; Vol. 77, No. 1. pp. 147-160.
@article{ed1ad97ad64a41b7a7bafe635cd1e410,
title = "Martingale-based residuals for survival models",
abstract = "Graphical methods based on the analysis of residuals are considered for the setting of the highly-used Cox (1972) regression model and for the Andersen-Gill (1982) generalization of that model. We start with a class of martingale-based residuals as proposed by Barlow & Prentice (1988). These residuals and/or their transforms are useful for investigating the functional form of a covariate, the proportional hazards assumption, the leverage of each subject upon the estimates of {\ss}, and the lack of model fit to a given subject.",
keywords = "Cox regression, Deviance, Influence function, Martingale, Outlier detection, Proportional hazards, Regression diagnostic, Residual",
author = "Therneau, {Terry M} and Grambsch, {Patricia M.} and Fleming, {Thomas R.}",
year = "1990",
month = "3",
doi = "10.1093/biomet/77.1.147",
language = "English (US)",
volume = "77",
pages = "147--160",
journal = "Biometrika",
issn = "0006-3444",
publisher = "Oxford University Press",
number = "1",

}

TY - JOUR

T1 - Martingale-based residuals for survival models

AU - Therneau, Terry M

AU - Grambsch, Patricia M.

AU - Fleming, Thomas R.

PY - 1990/3

Y1 - 1990/3

N2 - Graphical methods based on the analysis of residuals are considered for the setting of the highly-used Cox (1972) regression model and for the Andersen-Gill (1982) generalization of that model. We start with a class of martingale-based residuals as proposed by Barlow & Prentice (1988). These residuals and/or their transforms are useful for investigating the functional form of a covariate, the proportional hazards assumption, the leverage of each subject upon the estimates of ß, and the lack of model fit to a given subject.

AB - Graphical methods based on the analysis of residuals are considered for the setting of the highly-used Cox (1972) regression model and for the Andersen-Gill (1982) generalization of that model. We start with a class of martingale-based residuals as proposed by Barlow & Prentice (1988). These residuals and/or their transforms are useful for investigating the functional form of a covariate, the proportional hazards assumption, the leverage of each subject upon the estimates of ß, and the lack of model fit to a given subject.

KW - Cox regression

KW - Deviance

KW - Influence function

KW - Martingale

KW - Outlier detection

KW - Proportional hazards

KW - Regression diagnostic

KW - Residual

UR - http://www.scopus.com/inward/record.url?scp=77956890008&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77956890008&partnerID=8YFLogxK

U2 - 10.1093/biomet/77.1.147

DO - 10.1093/biomet/77.1.147

M3 - Article

VL - 77

SP - 147

EP - 160

JO - Biometrika

JF - Biometrika

SN - 0006-3444

IS - 1

ER -