Balancing Covariates via Propensity Score Weighting

Fan Li, Kari Lock Morgan, Alan M. Zaslavsky

Research output: Contribution to journalArticle

45 Citations (Scopus)

Abstract

Covariate balance is crucial for unconfounded descriptive or causal comparisons. However, lack of balance is common in observational studies. This article considers weighting strategies for balancing covariates. We define a general class of weights—the balancing weights—that balance the weighted distributions of the covariates between treatment groups. These weights incorporate the propensity score to weight each group to an analyst-selected target population. This class unifies existing weighting methods, including commonly used weights such as inverse-probability weights as special cases. General large-sample results on nonparametric estimation based on these weights are derived. We further propose a new weighting scheme, the overlap weights, in which each unit’s weight is proportional to the probability of that unit being assigned to the opposite group. The overlap weights are bounded, and minimize the asymptotic variance of the weighted average treatment effect among the class of balancing weights. The overlap weights also possess a desirable small-sample exact balance property, based on which we propose a new method that achieves exact balance for means of any selected set of covariates. Two applications illustrate these methods and compare them with other approaches.

Original languageEnglish (US)
Pages (from-to)390-400
Number of pages11
JournalJournal of the American Statistical Association
Volume113
Issue number521
DOIs
StatePublished - Jan 2 2018

Fingerprint

Propensity Score
Balancing
Weighting
Covariates
Overlap
Average Treatment Effect
Propensity score
Weighted Distributions
Observational Study
Unit
Asymptotic Variance
Weighted Average
Nonparametric Estimation
Small Sample
Directly proportional
Minimise

All Science Journal Classification (ASJC) codes

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Cite this

@article{2b588630ce5f416dacdf9f94727b3b31,
title = "Balancing Covariates via Propensity Score Weighting",
abstract = "Covariate balance is crucial for unconfounded descriptive or causal comparisons. However, lack of balance is common in observational studies. This article considers weighting strategies for balancing covariates. We define a general class of weights—the balancing weights—that balance the weighted distributions of the covariates between treatment groups. These weights incorporate the propensity score to weight each group to an analyst-selected target population. This class unifies existing weighting methods, including commonly used weights such as inverse-probability weights as special cases. General large-sample results on nonparametric estimation based on these weights are derived. We further propose a new weighting scheme, the overlap weights, in which each unit’s weight is proportional to the probability of that unit being assigned to the opposite group. The overlap weights are bounded, and minimize the asymptotic variance of the weighted average treatment effect among the class of balancing weights. The overlap weights also possess a desirable small-sample exact balance property, based on which we propose a new method that achieves exact balance for means of any selected set of covariates. Two applications illustrate these methods and compare them with other approaches.",
author = "Fan Li and Morgan, {Kari Lock} and Zaslavsky, {Alan M.}",
year = "2018",
month = "1",
day = "2",
doi = "10.1080/01621459.2016.1260466",
language = "English (US)",
volume = "113",
pages = "390--400",
journal = "Journal of the American Statistical Association",
issn = "0162-1459",
publisher = "Taylor and Francis Ltd.",
number = "521",

}

Balancing Covariates via Propensity Score Weighting. / Li, Fan; Morgan, Kari Lock; Zaslavsky, Alan M.

In: Journal of the American Statistical Association, Vol. 113, No. 521, 02.01.2018, p. 390-400.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Balancing Covariates via Propensity Score Weighting

AU - Li, Fan

AU - Morgan, Kari Lock

AU - Zaslavsky, Alan M.

PY - 2018/1/2

Y1 - 2018/1/2

N2 - Covariate balance is crucial for unconfounded descriptive or causal comparisons. However, lack of balance is common in observational studies. This article considers weighting strategies for balancing covariates. We define a general class of weights—the balancing weights—that balance the weighted distributions of the covariates between treatment groups. These weights incorporate the propensity score to weight each group to an analyst-selected target population. This class unifies existing weighting methods, including commonly used weights such as inverse-probability weights as special cases. General large-sample results on nonparametric estimation based on these weights are derived. We further propose a new weighting scheme, the overlap weights, in which each unit’s weight is proportional to the probability of that unit being assigned to the opposite group. The overlap weights are bounded, and minimize the asymptotic variance of the weighted average treatment effect among the class of balancing weights. The overlap weights also possess a desirable small-sample exact balance property, based on which we propose a new method that achieves exact balance for means of any selected set of covariates. Two applications illustrate these methods and compare them with other approaches.

AB - Covariate balance is crucial for unconfounded descriptive or causal comparisons. However, lack of balance is common in observational studies. This article considers weighting strategies for balancing covariates. We define a general class of weights—the balancing weights—that balance the weighted distributions of the covariates between treatment groups. These weights incorporate the propensity score to weight each group to an analyst-selected target population. This class unifies existing weighting methods, including commonly used weights such as inverse-probability weights as special cases. General large-sample results on nonparametric estimation based on these weights are derived. We further propose a new weighting scheme, the overlap weights, in which each unit’s weight is proportional to the probability of that unit being assigned to the opposite group. The overlap weights are bounded, and minimize the asymptotic variance of the weighted average treatment effect among the class of balancing weights. The overlap weights also possess a desirable small-sample exact balance property, based on which we propose a new method that achieves exact balance for means of any selected set of covariates. Two applications illustrate these methods and compare them with other approaches.

UR - http://www.scopus.com/inward/record.url?scp=85033679181&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85033679181&partnerID=8YFLogxK

U2 - 10.1080/01621459.2016.1260466

DO - 10.1080/01621459.2016.1260466

M3 - Article

AN - SCOPUS:85033679181

VL - 113

SP - 390

EP - 400

JO - Journal of the American Statistical Association

JF - Journal of the American Statistical Association

SN - 0162-1459

IS - 521

ER -