Simple least squares estimator for treatment effects using propensity score residuals

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

Propensity score matching is widely used to control covariates when analysing the effects of a nonrandomized binary treatment. However, it requires several arbitrary decisions, such as how many matched subjects to use and how to choose them. In this paper a simple least squares estimator is proposed, where the treatment, and possibly the response variable, is replaced by the propensity score residual. The proposed estimator controls covariates semiparametrically if the propensity score function is correctly specified. Furthermore, it is numerically stable and relatively easy to use, compared with alternatives such as matching, regression imputation, weighting, and doubly robust estimators. The proposed estimator also has a simple valid asymptotic variance estimator that works well in small samples. The least squares estimator is extended to multiple treatments and noncontinuously distributed responses. A simulation study demonstrates that it has lower mean squared error than its competitors.

Original languageEnglish
Pages (from-to)149-164
Number of pages16
JournalBiometrika
Volume105
Issue number1
DOIs
Publication statusPublished - 2018 Mar 1

Fingerprint

Propensity Score
Least Squares Estimator
Treatment Effects
Least-Squares Analysis
least squares
Covariates
Estimator
Score Function
Robust Estimators
Variance Estimator
Imputation
Asymptotic Variance
Mean Squared Error
Small Sample
Weighting
Choose
Regression
Simulation Study
Valid
Binary

Keywords

  • Binary treatment
  • Generalized propensity score
  • Multiple treatments
  • Propensity score

ASJC Scopus subject areas

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

Cite this

Simple least squares estimator for treatment effects using propensity score residuals. / Lee, Myoung-jae.

In: Biometrika, Vol. 105, No. 1, 01.03.2018, p. 149-164.

Research output: Contribution to journalArticle

@article{42b0b796ea7249e38bba08c151c65665,
title = "Simple least squares estimator for treatment effects using propensity score residuals",
abstract = "Propensity score matching is widely used to control covariates when analysing the effects of a nonrandomized binary treatment. However, it requires several arbitrary decisions, such as how many matched subjects to use and how to choose them. In this paper a simple least squares estimator is proposed, where the treatment, and possibly the response variable, is replaced by the propensity score residual. The proposed estimator controls covariates semiparametrically if the propensity score function is correctly specified. Furthermore, it is numerically stable and relatively easy to use, compared with alternatives such as matching, regression imputation, weighting, and doubly robust estimators. The proposed estimator also has a simple valid asymptotic variance estimator that works well in small samples. The least squares estimator is extended to multiple treatments and noncontinuously distributed responses. A simulation study demonstrates that it has lower mean squared error than its competitors.",
keywords = "Binary treatment, Generalized propensity score, Multiple treatments, Propensity score",
author = "Myoung-jae Lee",
year = "2018",
month = "3",
day = "1",
doi = "10.1093/biomet/asx062",
language = "English",
volume = "105",
pages = "149--164",
journal = "Biometrika",
issn = "0006-3444",
publisher = "Oxford University Press",
number = "1",

}

TY - JOUR

T1 - Simple least squares estimator for treatment effects using propensity score residuals

AU - Lee, Myoung-jae

PY - 2018/3/1

Y1 - 2018/3/1

N2 - Propensity score matching is widely used to control covariates when analysing the effects of a nonrandomized binary treatment. However, it requires several arbitrary decisions, such as how many matched subjects to use and how to choose them. In this paper a simple least squares estimator is proposed, where the treatment, and possibly the response variable, is replaced by the propensity score residual. The proposed estimator controls covariates semiparametrically if the propensity score function is correctly specified. Furthermore, it is numerically stable and relatively easy to use, compared with alternatives such as matching, regression imputation, weighting, and doubly robust estimators. The proposed estimator also has a simple valid asymptotic variance estimator that works well in small samples. The least squares estimator is extended to multiple treatments and noncontinuously distributed responses. A simulation study demonstrates that it has lower mean squared error than its competitors.

AB - Propensity score matching is widely used to control covariates when analysing the effects of a nonrandomized binary treatment. However, it requires several arbitrary decisions, such as how many matched subjects to use and how to choose them. In this paper a simple least squares estimator is proposed, where the treatment, and possibly the response variable, is replaced by the propensity score residual. The proposed estimator controls covariates semiparametrically if the propensity score function is correctly specified. Furthermore, it is numerically stable and relatively easy to use, compared with alternatives such as matching, regression imputation, weighting, and doubly robust estimators. The proposed estimator also has a simple valid asymptotic variance estimator that works well in small samples. The least squares estimator is extended to multiple treatments and noncontinuously distributed responses. A simulation study demonstrates that it has lower mean squared error than its competitors.

KW - Binary treatment

KW - Generalized propensity score

KW - Multiple treatments

KW - Propensity score

UR - http://www.scopus.com/inward/record.url?scp=85043283072&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85043283072&partnerID=8YFLogxK

U2 - 10.1093/biomet/asx062

DO - 10.1093/biomet/asx062

M3 - Article

AN - SCOPUS:85043283072

VL - 105

SP - 149

EP - 164

JO - Biometrika

JF - Biometrika

SN - 0006-3444

IS - 1

ER -