Bayesian Approach for Addressing Differential Covariate Measurement Error in Propensity Score Methods

Hwanhee Hong, Kara Rudolph, Elizabeth A. Stuart

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

Propensity score methods are an important tool to help reduce confounding in non-experimental studies and produce more accurate causal effect estimates. Most propensity score methods assume that covariates are measured without error. However, covariates are often measured with error. Recent work has shown that ignoring such error could lead to bias in treatment effect estimates. In this paper, we consider an additional complication: that of differential measurement error across treatment groups, such as can occur if a covariate is measured differently in the treatment and control groups. We propose two flexible Bayesian approaches for handling differential measurement error when estimating average causal effects using propensity score methods. We consider three scenarios: systematic (i.e., a location shift), heteroscedastic (i.e., different variances), and mixed (both systematic and heteroscedastic) measurement errors. We also explore various prior choices (i.e., weakly informative or point mass) on the sensitivity parameters related to the differential measurement error. We present results from simulation studies evaluating the performance of the proposed methods and apply these approaches to an example estimating the effect of neighborhood disadvantage on adolescent drug use disorders.

Original languageEnglish (US)
Pages (from-to)1078-1096
Number of pages19
JournalPsychometrika
Volume82
Issue number4
DOIs
StatePublished - Dec 1 2017
Externally publishedYes

Fingerprint

Propensity Score
Bayes Theorem
Measurement errors
Bayesian Approach
Measurement Error
Covariates
Causal Effect
Heteroscedastic Errors
Parameter Sensitivity
Confounding
Treatment Effects
Complications
Estimate
Substance-Related Disorders
Disorder
Drugs
Therapeutics
Simulation Study
Scenarios
Control Groups

Keywords

  • Bayesian hierarchical model
  • differential measurement error
  • inverse probability of treatment weighting
  • propensity score

ASJC Scopus subject areas

  • Psychology(all)
  • Applied Mathematics

Cite this

Bayesian Approach for Addressing Differential Covariate Measurement Error in Propensity Score Methods. / Hong, Hwanhee; Rudolph, Kara; Stuart, Elizabeth A.

In: Psychometrika, Vol. 82, No. 4, 01.12.2017, p. 1078-1096.

Research output: Contribution to journalArticle

@article{093415481901478c9ce00235351336e0,
title = "Bayesian Approach for Addressing Differential Covariate Measurement Error in Propensity Score Methods",
abstract = "Propensity score methods are an important tool to help reduce confounding in non-experimental studies and produce more accurate causal effect estimates. Most propensity score methods assume that covariates are measured without error. However, covariates are often measured with error. Recent work has shown that ignoring such error could lead to bias in treatment effect estimates. In this paper, we consider an additional complication: that of differential measurement error across treatment groups, such as can occur if a covariate is measured differently in the treatment and control groups. We propose two flexible Bayesian approaches for handling differential measurement error when estimating average causal effects using propensity score methods. We consider three scenarios: systematic (i.e., a location shift), heteroscedastic (i.e., different variances), and mixed (both systematic and heteroscedastic) measurement errors. We also explore various prior choices (i.e., weakly informative or point mass) on the sensitivity parameters related to the differential measurement error. We present results from simulation studies evaluating the performance of the proposed methods and apply these approaches to an example estimating the effect of neighborhood disadvantage on adolescent drug use disorders.",
keywords = "Bayesian hierarchical model, differential measurement error, inverse probability of treatment weighting, propensity score",
author = "Hwanhee Hong and Kara Rudolph and Stuart, {Elizabeth A.}",
year = "2017",
month = "12",
day = "1",
doi = "10.1007/s11336-016-9533-x",
language = "English (US)",
volume = "82",
pages = "1078--1096",
journal = "Psychometrika",
issn = "0033-3123",
publisher = "Springer New York",
number = "4",

}

TY - JOUR

T1 - Bayesian Approach for Addressing Differential Covariate Measurement Error in Propensity Score Methods

AU - Hong, Hwanhee

AU - Rudolph, Kara

AU - Stuart, Elizabeth A.

PY - 2017/12/1

Y1 - 2017/12/1

N2 - Propensity score methods are an important tool to help reduce confounding in non-experimental studies and produce more accurate causal effect estimates. Most propensity score methods assume that covariates are measured without error. However, covariates are often measured with error. Recent work has shown that ignoring such error could lead to bias in treatment effect estimates. In this paper, we consider an additional complication: that of differential measurement error across treatment groups, such as can occur if a covariate is measured differently in the treatment and control groups. We propose two flexible Bayesian approaches for handling differential measurement error when estimating average causal effects using propensity score methods. We consider three scenarios: systematic (i.e., a location shift), heteroscedastic (i.e., different variances), and mixed (both systematic and heteroscedastic) measurement errors. We also explore various prior choices (i.e., weakly informative or point mass) on the sensitivity parameters related to the differential measurement error. We present results from simulation studies evaluating the performance of the proposed methods and apply these approaches to an example estimating the effect of neighborhood disadvantage on adolescent drug use disorders.

AB - Propensity score methods are an important tool to help reduce confounding in non-experimental studies and produce more accurate causal effect estimates. Most propensity score methods assume that covariates are measured without error. However, covariates are often measured with error. Recent work has shown that ignoring such error could lead to bias in treatment effect estimates. In this paper, we consider an additional complication: that of differential measurement error across treatment groups, such as can occur if a covariate is measured differently in the treatment and control groups. We propose two flexible Bayesian approaches for handling differential measurement error when estimating average causal effects using propensity score methods. We consider three scenarios: systematic (i.e., a location shift), heteroscedastic (i.e., different variances), and mixed (both systematic and heteroscedastic) measurement errors. We also explore various prior choices (i.e., weakly informative or point mass) on the sensitivity parameters related to the differential measurement error. We present results from simulation studies evaluating the performance of the proposed methods and apply these approaches to an example estimating the effect of neighborhood disadvantage on adolescent drug use disorders.

KW - Bayesian hierarchical model

KW - differential measurement error

KW - inverse probability of treatment weighting

KW - propensity score

UR - http://www.scopus.com/inward/record.url?scp=84991108776&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84991108776&partnerID=8YFLogxK

U2 - 10.1007/s11336-016-9533-x

DO - 10.1007/s11336-016-9533-x

M3 - Article

C2 - 27738956

AN - SCOPUS:84991108776

VL - 82

SP - 1078

EP - 1096

JO - Psychometrika

JF - Psychometrika

SN - 0033-3123

IS - 4

ER -