An Introduction to the Augmented Inverse Propensity Weighted Estimator
The DOI matches its Crossref record; 33 of the 33 references it cites were found in a registry or an index.
Abstract
In this paper, we discuss an estimator for average treatment effects (ATEs) known as the augmented inverse propensity weighted (AIPW) estimator. This estimator has attractive theoretical properties and only requires practitioners to do two things they are already comfortable with: (1) specify a binary regression model for the propensity score, and (2) specify a regression model for the outcome variable. Perhaps the most interesting property of this estimator is its so-called “double robustness.” Put simply, the estimator remains consistent for the ATE if either the propensity score model or the outcome regression is misspecified but the other is properly specified. After explaining the AIPW estimator, we conduct a Monte Carlo experiment that compares the finite sample performance of the AIPW estimator to three common competitors: a regression estimator, an inverse propensity weighted (IPW) estimator, and a propensity score matching estimator. The Monte Carlo results show that the AIPW estimator has comparable or lower mean square error than the competing estimators when the propensity score and outcome models are both properly specified and, when one of the models is misspecified, the AIPW estimator is superior.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“These doubly robust approaches provide consistent estimates even if one of the models—the propensity score model or the outcome regression model—is misspecified.”
Citations
Cites
The central role of the propensity score in observational studies for causal effects
ROSENBAUM, P. R., & RUBIN, D. B. (1983) · Biometrika · doi:10.1093/biomet/70.1.41 · Cited by 31,241 · cited 2×
“Furthermore, we assume that potential outcomes are defined as in Rosenbaum and Rubin (1983) such that Yi(1) is the outcome that wewould observe if unit i had received treatment and Yi(0) is the outcome that we would observe if unit i had received treatment.”
BackgroundInfluentialCausality: Models, Reasoning, and Inference
Pearl, J. (2009) · doi:10.1017/cbo9780511803161 · Cited by 11,653
CAUSALITY: MODELS, REASONING AND INFERENCE
Hitchcock, C. (2001) · The Philosophical Review · doi:10.1215/00318108-110-4-639 · Cited by 4,741
Identification of Causal Effects Using Instrumental Variables
Angrist, J. D., Imbens, G. W., & Rubin, D. B. (1996) · Journal of the American Statistical Association · doi:10.1080/01621459.1996.10476902 · Cited by 4,230
Matching as Nonparametric Preprocessing for Reducing Model Dependence in Parametric Causal Inference
Ho, D. E., Imai, K., King, G., & Stuart, E. A. (2007) · Political Analysis · doi:10.1093/pan/mpl013 · Cited by 4,205 · cited 1×
“2 for a proof of this result and see Ho et al. (2007) for a related but distinct estimator that has this double robustness property.”
BackgroundHastie, T., & Tibshirani, R. (1986) · Statistical Science · doi:10.1214/ss/1177013604 · Cited by 3,409
Nonparametric Estimation of Average Treatment Effects Under Exogeneity: A Review
Imbens, G. W. (2004) · Review of Economics and Statistics · doi:10.1162/003465304323023651 · Cited by 2,994
Causality: models, reasoning, and inference
(2000) · Choice Reviews Online · doi:10.5860/choice.38-1494a · Cited by 2,873
Estimation of Regression Coefficients When Some Regressors are not Always Observed
Robins, J. M., Rotnitzky, A., & Zhao, L. P. (1994) · Journal of the American Statistical Association · doi:10.1080/01621459.1994.10476818 · Cited by 2,823 · cited 2×
“See Rubin (2006) for a book length treatment on matching, or Diamond and Sekhon (2005) and Ho et al.”
BackgroundRobins, J. (1986) · Mathematical Modelling · doi:10.1016/0270-0255(86)90088-6 · Cited by 2,451 · cited 2×
“Given the stated assumptions of this paper, it has been shown that such a model can be used to identify the ATE through a subclassification adjustment (Cochran 1968), the g-functional (Robins 1986), or the backdoor adjustment (Pearl 1995, 2000).”
BackgroundMethodInfluentialCausal diagrams for empirical research
PEARL, J. (1995) · Biometrika · doi:10.1093/biomet/82.4.669 · Cited by 2,355 · cited 6×
“See Pearl (1995, 2000) andMorgan andWinship (2007) for an approach to reasoning about this problem based on structural causal models.”
BackgroundMethodInfluentialLunceford, J. K., & Davidian, M. (2004) · Statistics in Medicine · doi:10.1002/sim.1903 · Cited by 1,674
(2007) · Cited by 1,428
Counterfactuals and Causal Inference: Methods and Principles for Social Research
Morgan, S. L., & Winship, C. (2007) · doi:10.1017/cbo9780511804564 · Cited by 1,146
Diamond, A., & Sekhon, J. S. (2013) · Review of Economics and Statistics · doi:10.1162/rest_a_00318 · Cited by 1,102
The Effectiveness of Adjustment by Subclassification in Removing Bias in Observational Studies
Cochran, W. G. (1968) · Biometrics · doi:10.2307/2528036 · Cited by 961 · cited 1×
“Given the stated assumptions of this paper, it has been shown that such a model can be used to identify the ATE through a subclassification adjustment (Cochran 1968), the g-functional (Robins 1986), or the backdoor adjustment (Pearl 1995, 2000).”
MethodSemiparametric Theory and Missing Data
(2006) · Springer Series in Statistics · doi:10.1007/0-387-37345-4 · Cited by 827 · cited 4×
“If the propensity scores were known, then this estimator will be unbiased for the ATE (Tsiatis 2006).”
BackgroundMethodInfluentialKang, J. D. Y., & Schafer, J. L. (2007) · Statistical Science · doi:10.1214/07-sts227 · Cited by 797
Matched Sampling for Causal Effects
Rubin, D. B. (2006) · Cambridge University Press eBooks · doi:10.1017/cbo9780511810725 · Cited by 676
The Dangers of Extreme Counterfactuals
King, G., & Zeng, L. (2006) · Political Analysis · doi:10.1093/pan/mpj004 · Cited by 540 · cited 1×
“Advance Access publication December 14, 2009 Political Analysis (2010) 18:36–56 doi:10.1093/pan/mpp036 An Introduction to the Augmented Inverse Propensity”
BackgroundInfluentialNew evidence on the finite sample properties of propensity score reweighting and matching estimators
Busso, M., DiNardo, J., & McCrary, J. (2014) · Review of Economics and Statistics · doi:10.1162/rest_a_00431 · Cited by 428
Semiparametric Theory and Missing Data
Fokianos, K. (2007) · Technometrics · doi:10.1198/tech.2007.s488 · Cited by 418
Robins, J. M., Sued, M., Lei-Gomez, Q., Rotnitzky, A. G., Ciencias, F., & Naturales, E. (2010) · Cited by 346
Robins, J., Sued, M., Lei-Gomez, Q., & Rotnitzky, A. (2007) · Statistical Science · doi:10.1214/07-sts227d · Cited by 313 · cited 3×
“The key advantage of the AIPW estimator (and other more advanced doublerobust estimators [Robins et al. 2007]) in such a situation is that it will continue to perform well as long as either the propensity score model or the outcome regressions are properly specified.”
BackgroundMethodInference for imputation estimators
Robins, J. (2000) · Biometrika · doi:10.1093/biomet/87.1.113 · Cited by 276 · cited 1×
“See Rubin (2006) for a book-length treatment on matching or Diamond and Sekhon (2005) and Ho et al.”
New Evidence on the Finite Sample Properties of Propensity Score Matching and Reweighting EstimatorsPreprint
Busso, M., DiNardo, J., & McCrary, J. (2009) · SSRN Electronic Journal · doi:10.2139/ssrn.1351162 · Cited by 143
Adjusting for Nonignorable Drop-Out Using Semiparametric Nonresponse Models: Rejoinder
Scharfstein, D. O., Rotnitzky, A., & Robins, J. M. (1999) · Journal of the American Statistical Association · doi:10.2307/2669930 · Cited by 80 · cited 4×
“For Permissions, please email: journals.permissions@oxfordjournals.org 36 consistent for the ATE whenever (1) the propensity score model is correctly specified or (2) the outcome regression is correctly specified (Scharfstein, Rotnitzky, and Robins 1999).”
BackgroundResultInfluentialComment: Understanding OR, PS and DR
Tan, Z. (2007) · Statistical Science · doi:10.1214/07-sts227a · Cited by 72
Tsiatis, A. A., & Davidian, M. (2007) · Statistical Science · doi:10.1214/07-sts227b · Cited by 63
Ridgeway, G. K., McCaffrey, D. F., Statistician, M. I. S., & The, H. O. (2013) · Cited by 50
Ridgeway, G., & McCaffrey, D. F. (2007) · Statistical Science · doi:10.1214/07-sts227c · Cited by 42
Kang, J., & Schafer, J. L. (2007) · Statistical Science · doi:10.1214/07-sts227rej · Cited by 22
Ohlsson, E., & Johansson, B. J. E. (2010) · EAA series · doi:10.1007/978-3-642-10791-7_5 · Cited by 5
Research works printed but not linked: 4
- Busso, 2009a, Finite sample properties of semiparametric estimators of average treatment effects
- Sekhon, 2009, Multivariate and propensity score matching with balance optimization
- 2007, R: A language and environment for statistical computing
- Robins James M. 1999. Robust estimation in sequentially ignorable missing data and causal inference models. Proceedings of the American Statistical Association Section on Bayesian Statistical Science 6–10.
Cited by
The Time for Austerity: Estimating the Average Treatment Effect of Fiscal Policy
Jordà, Ò., & Taylor, A. M. (2016) · The Economic Journal · doi:10.1111/ecoj.12332 · Cited by 548
“…into the broad class of “doubly robust” estimators of which Robins, Rotnitzky, and Zhao ( 1994 ) is perhaps the earliest reference (see also Robins 1999 ; Scharfstein, Rotnitzky, and Robins 1999 ; Hirano, Imbens, and Ridder 2003 ; Lunceford and Davidian 2004 ; Imbens 2004 ; Glynn and Quinn 2010 ).”
InfluentialThe great mortgaging: housing finance, crises and business cycles
Jordà, Ò., Schularick, M., & Taylor, A. M. (2016) · Economic Policy · doi:10.1093/epolic/eiv017 · Cited by 521
The performance of estimators based on the propensity score
Huber, M., Lechner, M., & Wunsch, C. (2013) · Journal of Econometrics · doi:10.1016/j.jeconom.2012.11.006 · Cited by 311
“31 Robins, Rotnitzky and Zhao (1994) and Robins and Rotnitzky (1995) show that DR is semi-parametrically efficient if both model components are correctly specified (see also the discussions in Robins, Mark and Newey, 1992, Scharfstein, Rotnitzky and Robins, 1999, Hirano and Imbens, 2001, Lunceford and Davidian, 2004,…”
BackgroundThe causal interpretation of estimated associations in regression models
Keele, L., Stevenson, R. T., & Elwert, F. (2020) · Political Science Research and Methods · doi:10.1017/psrm.2019.31 · Cited by 226
“See Glynn and Quinn (2010) for an introduction to such methods in political science.”
BackgroundDoes Collaboration Make Any Difference? Linking Collaborative Governance to Environmental Outcomes
Scott, T. (2015) · Journal of Policy Analysis and Management · doi:10.1002/pam.21836 · Cited by 224
“I specify the ÂT EAI PW estimator and describe the technical details of this approach, in particular the analytical advantages of the AIPW estimator relative to the IPW estimator, in Appendix A.3 The ÂT EAI PW estimator only removes selection bias if it suitably accounts for the factors that motivate selection into…”
Influential
Metadata
- Authors
- Glynn, A. N., & Quinn, K. M.
- Affiliations
- Harvard University (US)
- University of California, Berkeley (US)
- Corresponding author
- Glynn, Adam N.
- Venue
- Political Analysis
- Publisher
- Cambridge University Press (CUP)
- Published
- 2010
- Volume
- 18
- Issue
- 1
- Pages
- 36-56
- ISSN
- 1047-1987 (ISSN-L)
- 1476-4989
- OpenAlex
- W2066443416
- Cited by
- 374
- References
- 33
- Citations by year
- 2026: 29 · 2025: 57 · 2024: 58 · 2023: 44 · 2022: 36 · 2021: 45 · 2020: 25 · 2019: 18 · 2018: 12 · 2017: 9 · 2016: 15 · 2015: 6 · 2014: 6 · 2013: 9 · 2012: 2
- FWCI
- 1.11
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- Political Analysis (publishedVersion)
- RePEc: Research Papers in Economics (submittedVersion)
- Topics
- Physical Sciences › Mathematics › Statistics and Probability › Advanced Causal Inference Techniques
- Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods and Bayesian Inference
- Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods in Clinical Trials
- Keywords
- average treatment effect, double robustness, propensity score, inverse propensity weighting, propensity score matching, model misspecification, finite sample properties, causal inference
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.