Covariate Balancing Propensity Score

Journal of the Royal Statistical Society. Series B: Statistical Methodology - Tập 76 Số 1 - Trang 243-263 - 2014
Kosuke Imai1, Marc Ratkovic1
1Princeton University USA

Tóm tắt

SummaryThe propensity score plays a central role in a variety of causal inference settings. In particular, matching and weighting methods based on the estimated propensity score have become increasingly common in the analysis of observational data. Despite their popularity and theoretical appeal, the main practical difficulty of these methods is that the propensity score must be estimated. Researchers have found that slight misspecification of the propensity score model can result in substantial bias of estimated treatment effects. We introduce covariate balancing propensity score (CBPS) methodology, which models treatment assignment while optimizing the covariate balance. The CBPS exploits the dual characteristics of the propensity score as a covariate balancing score and the conditional probability of treatment assignment. The estimation of the CBPS is done within the generalized method-of-moments or empirical likelihood framework. We find that the CBPS dramatically improves the poor empirical performance of propensity score matching and weighting methods reported in the literature. We also show that the CBPS can be extended to other important settings, including the estimation of the generalized propensity score for non-binary treatments and the generalization of experimental estimates to a target population. Open source software is available for implementing the methods proposed.

Từ khóa


Tài liệu tham khảo

Abadie, 2006, Large sample properties of matching estimators for average treatment effects, Econometrica, 74, 235, 10.1111/j.1468-0262.2006.00655.x

Abadie, 2008, On the failure of the bootstrap for matching estimators, Econometrica, 76, 1537, 10.3982/ECTA6474

Abadie, 2011, Bias-corrected matching estimators for average treatment effects, J. Bus. Econ. Statist., 29, 1, 10.1198/jbes.2009.07333

Angrist, 2010, ExtrapoLATE-ing: external validity and overidentification in the LATE framework, 10.3386/w16566

Camillo, 2011, A multivariate strategy to measure and test global imbalance, Exprt Syst. Applic., 38, 3451, 10.1016/j.eswa.2010.08.132

Chaudhuri, 2008, Generalized linear models incorporating population level information: an empirical-likelihood-based approach, J. R. Statist. Soc. B, 70, 311, 10.1111/j.1467-9868.2007.00637.x

Cole, 2010, Generalizing evidence from randomized clinical trials to target populations: the ACTG 320 trial, Am. J. Epidem., 172, 107, 10.1093/aje/kwq084

Dehejia, 1999, Causal effects in nonexperimental studies: reevaluating the evaluation of training programs, J. Am. Statist. Ass., 94, 1053, 10.1080/01621459.1999.10473858

Deming, 1940, On a least squares adjustment of a sampled frequency when the expected marginal tables are known, Ann. Math. Statist., 11, 427, 10.1214/aoms/1177731829

Diamond, 2012, Genetic matching for estimating causal effects: a new method of achieving balance in observational studies, Rev. Econ. Statist.

Freedman, 2008, Weighting regressions by propensity scores, Evaln Rev., 32, 392, 10.1177/0193841X08317586

Graham, 2012, Inverse probability tilting for moment condition models with missing data, Rev. Econ. Stud., 79, 1053, 10.1093/restud/rdr047

Hahn, 1998, On the role of the propensity score in efficient semiparametric estimation of average treatment effects, Econometrica, 66, 315, 10.2307/2998560

Hainmueller, 2008, Synthetic matching for causal effects: a multivariate reweighting method to produce balanced samples in observational studies

Hainmueller, 2012, Entropy balancing for causal effects: multivariate reweighting method to produce balanced samples in observational studies, Polit. Anal., 20, 25, 10.1093/pan/mpr025

Hansen, 2004, Full matching in an observational study of coaching for the SAT, J. Am. Statist. Ass., 99, 609, 10.1198/016214504000000647

Hansen, 1982, Large sample properties of generalized method of moments estimators, Econometrica, 50, 1029, 10.2307/1912775

Hansen, 1996, Finite-sample properties of some alternative GMM estimators, J. Bus. Econ. Statist., 14, 262, 10.1080/07350015.1996.10524656

Hayashi, 2000, Econometrics

Heckman, 1998, Matching as an econometric evaluation estimator, Rev. Econ. Stud., 65, 261, 10.1111/1467-937X.00044

Hellerstein, 1999, Imposing moment restrictions from auxiliary data by weighting, Rev. Econ. Statist., 81, 1, 10.1162/003465399557860

Hirano, 2003, Efficient estimation of average treatment effects using the estimated propensity score, Econometrica, 71, 1307, 10.1111/1468-0262.00451

Ho, 2007, Matching as nonparametric preprocessing for reducing model dependence in parametric causal inference, Polit. Anal., 15, 199, 10.1093/pan/mpl013

Horvitz, 1952, A generalization of sampling without replacement from a finite universe, J. Am. Statist. Ass., 47, 663, 10.1080/01621459.1952.10483446

Iacus, 2011, Multivariate matching methods that are monotonic imbalance bounding, J. Am. Statist. Ass., 106, 345, 10.1198/jasa.2011.tm09599

Imai, 2004, Causal inference with general treatment regimes: generalizing the propensity score, J. Am. Statist. Ass., 99, 854, 10.1198/016214504000001187

Imai, 2008, Misunderstandings between experimentalists and observationalists about causal inference, J. R. Statist. Soc. A, 171, 481, 10.1111/j.1467-985X.2007.00527.x

Imai, 2013, Estimating treatment effect heterogeneity in randomized program evaluation, Ann. Appl. Statist., 7, 443, 10.1214/12-AOAS593

Imbens, 2000, The role of the propensity score in estimating dose-response functions, Biometrika, 87, 706, 10.1093/biomet/87.3.706

Imbens, 2004, Nonparametric estimation of average treatment effects under exogeneity: a review, Rev. Econ. Statist., 86, 4, 10.1162/003465304323023651

Kang, 2007, Demystifying double robustness: a comparison of alternative strategies for estimating a population mean from incomplete data (with discussions), Statist. Sci., 22, 523

La Londe, 1986, Evaluating the econometric evaluations of training programs with experimental data, Am. Econ. Rev., 76, 604

Little, 1991, Models for contingency tables with known margins when target and sampled populations differ, J. Am. Statist. Ass., 86, 87, 10.1080/01621459.1991.10475007

Lunceford, 2004, Stratification and weighting via the propensity score in estimation of causal treatment effects: a comparative study, Statist. Med., 23, 2937, 10.1002/sim.1903

McCaffrey, 2004, Propensity score estimation with boosted regression for evaluating causal effects in observational studies, Psychol. Meth., 9, 403, 10.1037/1082-989X.9.4.403

Nevo, 2003, Using weights to adjust for sample selection when auxiliary information is available, J. Bus. Econ. Statist., 21, 43, 10.1198/073500102288618748

Newey, 1994, Handbook of Econometrics, 2111

Oh, 1983, Incomplete Data in Sample Surveys, vol. II, Theory and Annotated Bibliography

Owen, 2001, Empirical Likelihood

Qin, 1994, Empirical likelihood and general estimating equations, Ann. Statist., 22, 300, 10.1214/aos/1176325370

Ratkovic, 2012, Achieving optimal covariate balance under general treatment regimes

Ratkovic, 2012, Cbps: R package for covariate balancing propensity score

Ridgeway, 2007, Demystifying double robustness: a comparison of alternative strategies for estimating a population mean from incomplete data, Statist. Sci., 22, 540, 10.1214/07-STS227C

Robins, 2000, Marginal structural models and causal inference in epidemiology, Epidemiology, 11, 550, 10.1097/00001648-200009000-00011

Robins, 1994, Estimation of regression coefficients when some regressors are not always observed, J. Am. Statist. Ass., 89, 846, 10.1080/01621459.1994.10476818

Robins, 1995, Analysis of semiparametric regression models for repeated outcomes in the presence of missing data, J. Am. Statist. Ass., 90, 106, 10.1080/01621459.1995.10476493

Robins, 2007, Performance of double-robust estimators when ‘inverse probability’ weights are highly variable, Statist. Sci., 22, 544, 10.1214/07-STS227D

Rosenbaum, 1987, Model-based direct adjustment, J. Am. Statist. Ass., 82, 387, 10.1080/01621459.1987.10478441

Rosenbaum, 1989, Optimal matching for observational studies, J. Am. Statist. Ass., 84, 1024, 10.1080/01621459.1989.10478868

Rosenbaum, 1991, A characterization of optimal designs for observational studies, J. R. Statist. Soc. B, 53, 597, 10.1111/j.2517-6161.1991.tb01848.x

Rosenbaum, 1983, The central role of the propensity score in observational studies for causal effects, Biometrika, 70, 41, 10.1093/biomet/70.1.41

Rosenbaum, 1984, Reducing bias in observational studies using subclassification on the propensity score, J. Am. Statist. Ass., 79, 516, 10.1080/01621459.1984.10478078

Rosenbaum, 1985, Constructing a control group using multivariate matched sampling methods that incorporate the propensity score, Am. Statistn, 39, 33, 10.1080/00031305.1985.10479383

Rubin, 2007, The design versus the analysis of observational studies for causal effects: parallels with the design of randomized trials, Statist. Med., 26, 20, 10.1002/sim.2739

Smith, 2005, Does matching overcome LaLonde's critique of nonexperimental estimators?, J. Econmetr., 125, 305, 10.1016/j.jeconom.2004.04.011

Stuart, 2010, Matching methods for causal inference: a review and a look forward, Statist. Sci., 25, 1, 10.1214/09-STS313

Stuart, 2011, The use of propensity scores to assess the generalizability of results from randomized trials, J. R. Statist. Soc. A, 174, 369, 10.1111/j.1467-985X.2010.00673.x

Tan, 2010, Bounded, efficient and doubly robust estimation with inverse weighting, Biometrika, 97, 661, 10.1093/biomet/asq035

Thông tin
Thông tin xuất bản

Journal of the Royal Statistical Society. Series B: Statistical Methodology

Tập 76 Số 1

243-263

Thông tin tác giả