Doubly robust estimation in observational studies with partial interference

Lan Liu, Michael G. Hudgens, Bradley Saul, John D. Clemens, Mohammad Ali, Michael E. Emch

Research output: Contribution to journalArticle

Abstract

Interference occurs when the treatment (or exposure) of one individual affects the outcomes of others. In some settings, it may be reasonable to assume that individuals can be partitioned into clusters such that there is no interference between individuals in different clusters, that is, there is partial interference. In observational studies with partial interference, inverse probability weighted (IPW) estimators have been something else different possible treatment effects. However, the validity of IPW estimators depends on the propensity score being known or correctly modelled. Alternatively, one can estimate the treatment effect using an outcome regression model. In this paper, we propose doubly robust (DR) estimators that utilize both models and are consistent and asymptotically normal if either model, but not necessarily both, is correctly specified. Empirical results are presented to demonstrate the DR property of the proposed estimators and the efficiency gain of DR over IPW estimators when both models are correctly specified. The different estimators are illustrated using data from a study examining the effects of cholera vaccination in Bangladesh.

Original languageEnglish (US)
Article numbere214
JournalStat
Volume8
Issue number1
DOIs
StatePublished - Jan 1 2019

Fingerprint

Observational Study
Robust Estimation
Interference
Estimator
Partial
Treatment Effects
Propensity Score
Robust Estimators
Vaccination
Regression Model
Observational study
Robust estimation
Model
Estimate
Demonstrate

Keywords

  • causal inference
  • doubly robust estimator
  • interference
  • observational studies

PubMed: MeSH publication types

  • Journal Article

Cite this

Liu, L., Hudgens, M. G., Saul, B., Clemens, J. D., Ali, M., & Emch, M. E. (2019). Doubly robust estimation in observational studies with partial interference. Stat, 8(1), [e214]. https://doi.org/10.1002/sta4.214

Doubly robust estimation in observational studies with partial interference. / Liu, Lan; Hudgens, Michael G.; Saul, Bradley; Clemens, John D.; Ali, Mohammad; Emch, Michael E.

In: Stat, Vol. 8, No. 1, e214, 01.01.2019.

Research output: Contribution to journalArticle

Liu, L, Hudgens, MG, Saul, B, Clemens, JD, Ali, M & Emch, ME 2019, 'Doubly robust estimation in observational studies with partial interference', Stat, vol. 8, no. 1, e214. https://doi.org/10.1002/sta4.214
Liu, Lan ; Hudgens, Michael G. ; Saul, Bradley ; Clemens, John D. ; Ali, Mohammad ; Emch, Michael E. / Doubly robust estimation in observational studies with partial interference. In: Stat. 2019 ; Vol. 8, No. 1.
@article{362340e8015f461093761e8c764f75a4,
title = "Doubly robust estimation in observational studies with partial interference",
abstract = "Interference occurs when the treatment (or exposure) of one individual affects the outcomes of others. In some settings, it may be reasonable to assume that individuals can be partitioned into clusters such that there is no interference between individuals in different clusters, that is, there is partial interference. In observational studies with partial interference, inverse probability weighted (IPW) estimators have been something else different possible treatment effects. However, the validity of IPW estimators depends on the propensity score being known or correctly modelled. Alternatively, one can estimate the treatment effect using an outcome regression model. In this paper, we propose doubly robust (DR) estimators that utilize both models and are consistent and asymptotically normal if either model, but not necessarily both, is correctly specified. Empirical results are presented to demonstrate the DR property of the proposed estimators and the efficiency gain of DR over IPW estimators when both models are correctly specified. The different estimators are illustrated using data from a study examining the effects of cholera vaccination in Bangladesh.",
keywords = "causal inference, doubly robust estimator, interference, observational studies",
author = "Lan Liu and Hudgens, {Michael G.} and Bradley Saul and Clemens, {John D.} and Mohammad Ali and Emch, {Michael E.}",
year = "2019",
month = "1",
day = "1",
doi = "10.1002/sta4.214",
language = "English (US)",
volume = "8",
journal = "Stat",
issn = "2049-1573",
publisher = "Wiley-Blackwell Publishing Ltd",
number = "1",

}

TY - JOUR

T1 - Doubly robust estimation in observational studies with partial interference

AU - Liu, Lan

AU - Hudgens, Michael G.

AU - Saul, Bradley

AU - Clemens, John D.

AU - Ali, Mohammad

AU - Emch, Michael E.

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Interference occurs when the treatment (or exposure) of one individual affects the outcomes of others. In some settings, it may be reasonable to assume that individuals can be partitioned into clusters such that there is no interference between individuals in different clusters, that is, there is partial interference. In observational studies with partial interference, inverse probability weighted (IPW) estimators have been something else different possible treatment effects. However, the validity of IPW estimators depends on the propensity score being known or correctly modelled. Alternatively, one can estimate the treatment effect using an outcome regression model. In this paper, we propose doubly robust (DR) estimators that utilize both models and are consistent and asymptotically normal if either model, but not necessarily both, is correctly specified. Empirical results are presented to demonstrate the DR property of the proposed estimators and the efficiency gain of DR over IPW estimators when both models are correctly specified. The different estimators are illustrated using data from a study examining the effects of cholera vaccination in Bangladesh.

AB - Interference occurs when the treatment (or exposure) of one individual affects the outcomes of others. In some settings, it may be reasonable to assume that individuals can be partitioned into clusters such that there is no interference between individuals in different clusters, that is, there is partial interference. In observational studies with partial interference, inverse probability weighted (IPW) estimators have been something else different possible treatment effects. However, the validity of IPW estimators depends on the propensity score being known or correctly modelled. Alternatively, one can estimate the treatment effect using an outcome regression model. In this paper, we propose doubly robust (DR) estimators that utilize both models and are consistent and asymptotically normal if either model, but not necessarily both, is correctly specified. Empirical results are presented to demonstrate the DR property of the proposed estimators and the efficiency gain of DR over IPW estimators when both models are correctly specified. The different estimators are illustrated using data from a study examining the effects of cholera vaccination in Bangladesh.

KW - causal inference

KW - doubly robust estimator

KW - interference

KW - observational studies

UR - http://www.scopus.com/inward/record.url?scp=85064528375&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85064528375&partnerID=8YFLogxK

U2 - 10.1002/sta4.214

DO - 10.1002/sta4.214

M3 - Article

C2 - 31440374

AN - SCOPUS:85064528375

VL - 8

JO - Stat

JF - Stat

SN - 2049-1573

IS - 1

M1 - e214

ER -