Simpson's paradox in survival models

Clelia Di Serio, Yosef Rinott, Marco Scarsini

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

In the context of survival analysis it is possible that increasing the value of a covariate X has a beneficial effect on a failure time, but this effect is reversed when conditioning on any possible value of another covariate Y. When studying causal effects and influence of covariates on a failure time, this state of affairs appears paradoxical and raises questions about the real effect of X. Situations of this kind may be seen as a version of Simpson's paradox. In this paper, we study this phenomenon in terms of the linear transformation model. The introduction of a time variable makes the paradox more interesting and intricate: it may hold conditionally on a certain survival time, i.e. on an event of the type [T>t] for some but not all t, and it may hold only for some range of survival times.

Original languageEnglish
Pages (from-to)463-480
Number of pages18
JournalScandinavian Journal of Statistics
Volume36
Issue number3
DOIs
Publication statusPublished - Sep 2009

Fingerprint

Simpson's Paradox
Survival Model
Covariates
Survival Time
Failure Time
Linear Transformation Model
Causal Effect
Survival Analysis
Paradox
Conditioning
Range of data
Survival model

Keywords

  • Cox model
  • Detrimental covariate
  • Linear transformation model
  • Omitting covariates
  • Positive dependence
  • Proportional hazard
  • Proportional odds model
  • Protective covariate
  • Total positivity

ASJC Scopus subject areas

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Cite this

Simpson's paradox in survival models. / Di Serio, Clelia; Rinott, Yosef; Scarsini, Marco.

In: Scandinavian Journal of Statistics, Vol. 36, No. 3, 09.2009, p. 463-480.

Research output: Contribution to journalArticle

Di Serio, Clelia ; Rinott, Yosef ; Scarsini, Marco. / Simpson's paradox in survival models. In: Scandinavian Journal of Statistics. 2009 ; Vol. 36, No. 3. pp. 463-480.
@article{88fc15f747774b2690917d092718dcee,
title = "Simpson's paradox in survival models",
abstract = "In the context of survival analysis it is possible that increasing the value of a covariate X has a beneficial effect on a failure time, but this effect is reversed when conditioning on any possible value of another covariate Y. When studying causal effects and influence of covariates on a failure time, this state of affairs appears paradoxical and raises questions about the real effect of X. Situations of this kind may be seen as a version of Simpson's paradox. In this paper, we study this phenomenon in terms of the linear transformation model. The introduction of a time variable makes the paradox more interesting and intricate: it may hold conditionally on a certain survival time, i.e. on an event of the type [T>t] for some but not all t, and it may hold only for some range of survival times.",
keywords = "Cox model, Detrimental covariate, Linear transformation model, Omitting covariates, Positive dependence, Proportional hazard, Proportional odds model, Protective covariate, Total positivity",
author = "{Di Serio}, Clelia and Yosef Rinott and Marco Scarsini",
year = "2009",
month = "9",
doi = "10.1111/j.1467-9469.2008.00637.x",
language = "English",
volume = "36",
pages = "463--480",
journal = "Scandinavian Journal of Statistics",
issn = "0303-6898",
publisher = "Wiley-Blackwell",
number = "3",

}

TY - JOUR

T1 - Simpson's paradox in survival models

AU - Di Serio, Clelia

AU - Rinott, Yosef

AU - Scarsini, Marco

PY - 2009/9

Y1 - 2009/9

N2 - In the context of survival analysis it is possible that increasing the value of a covariate X has a beneficial effect on a failure time, but this effect is reversed when conditioning on any possible value of another covariate Y. When studying causal effects and influence of covariates on a failure time, this state of affairs appears paradoxical and raises questions about the real effect of X. Situations of this kind may be seen as a version of Simpson's paradox. In this paper, we study this phenomenon in terms of the linear transformation model. The introduction of a time variable makes the paradox more interesting and intricate: it may hold conditionally on a certain survival time, i.e. on an event of the type [T>t] for some but not all t, and it may hold only for some range of survival times.

AB - In the context of survival analysis it is possible that increasing the value of a covariate X has a beneficial effect on a failure time, but this effect is reversed when conditioning on any possible value of another covariate Y. When studying causal effects and influence of covariates on a failure time, this state of affairs appears paradoxical and raises questions about the real effect of X. Situations of this kind may be seen as a version of Simpson's paradox. In this paper, we study this phenomenon in terms of the linear transformation model. The introduction of a time variable makes the paradox more interesting and intricate: it may hold conditionally on a certain survival time, i.e. on an event of the type [T>t] for some but not all t, and it may hold only for some range of survival times.

KW - Cox model

KW - Detrimental covariate

KW - Linear transformation model

KW - Omitting covariates

KW - Positive dependence

KW - Proportional hazard

KW - Proportional odds model

KW - Protective covariate

KW - Total positivity

UR - http://www.scopus.com/inward/record.url?scp=68949107506&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=68949107506&partnerID=8YFLogxK

U2 - 10.1111/j.1467-9469.2008.00637.x

DO - 10.1111/j.1467-9469.2008.00637.x

M3 - Article

VL - 36

SP - 463

EP - 480

JO - Scandinavian Journal of Statistics

JF - Scandinavian Journal of Statistics

SN - 0303-6898

IS - 3

ER -