The Korean Journal of Applied Statistics (응용통계연구)

The Korean Statistical Society (한국통계학회)
권호 표지
DOI QR코드

DOI QR Code

Overview of estimating the average treatment effect using dimension reduction methods

차원축소 방법을 이용한 평균처리효과 추정에 대한 개요

  • Mijeong Kim (Department of Statistics, Ewha Womans University)
  • 김미정 (이화여자대학교 통계학과)
  • Received : 2023.02.24
  • Accepted : 2023.05.08
  • Published : 2023.08.31
Download PDF
⟨ Previous Next ⟩

Abstract

In causal analysis of high dimensional data, it is important to reduce the dimension of covariates and transform them appropriately to control confounders that affect treatment and potential outcomes. The augmented inverse probability weighting (AIPW) method is mainly used for estimation of average treatment effect (ATE). AIPW estimator can be obtained by using estimated propensity score and outcome model. ATE estimator can be inconsistent or have large asymptotic variance when using estimated propensity score and outcome model obtained by parametric methods that includes all covariates, especially for high dimensional data. For this reason, an ATE estimation using an appropriate dimension reduction method and semiparametric model for high dimensional data is attracting attention. Semiparametric method or sparse sufficient dimensionality reduction method can be uesd for dimension reduction for the estimation of propensity score and outcome model. Recently, another method has been proposed that does not use propensity score and outcome regression. After reducing dimension of covariates, ATE estimation can be performed using matching. Among the studies on ATE estimation methods for high dimensional data, four recently proposed studies will be introduced, and how to interpret the estimated ATE will be discussed.

고차원 데이터의 인과 추론에서 고차원 공변량의 차원을 축소하고 적절히 변형하여 처리와 잠재 결과에 영향을 줄 수 있는 교란을 통제하는 것은 중요한 문제이다. 평균 처리 효과(average treatment effect; ATE) 추정에 있어서, 성향점수와 결과 모형 추정을 이용한 확장된 역확률 가중치 방법이 주로 사용된다. 고차원 데이터의 분석시 모든 공변량을 포함한 모수 모형을 이용하여 성향 점수와 결과 모형 추정을 할 경우, ATE 추정량이 일치성을 갖지 않거나 추정량의 분산이 큰 값을 가질 수 있다. 이런 이유로 고차원 데이터에 대한 적절한 차원 축소 방법과 준모수 모형을 이용한 ATE 방법이 주목 받고 있다. 이와 관련된 연구로는 차원 축소부분에 준모수 모형과 희소 충분 차원 축소 방법을 활용한 연구가 있다. 최근에는 성향점수와 결과 모형을 추정하지 않고, 차원 축소 후 매칭을 활용한 ATE 추정 방법도 제시되었다. 고차원 데이터의 ATE 추정 방법연구 중 최근에 제시된 네 가지 연구에 대해 소개하고, 추정치 해석시 유의할 점에 대하여 논하기로 한다.

Keywords

Acknowledgement

이 논문은 연구재단 연구 과제 NRF-2020R1F1A1A01074157에 의하여 수행되었음.

References

  1. Abadie A and Imbens GW (2006). Large sample properties of matching estimators for average treatment effects, Econometrica, 74, 235-267.  https://doi.org/10.1111/j.1468-0262.2006.00655.x
  2. Aronszajn N (1950). Theory of reproducing kernels, Transactions of the American Mathematical Society, 68, 337-404.  https://doi.org/10.1090/S0002-9947-1950-0051437-7
  3. Carpenter JR, Kenward MG, and Vansteelandt S (2006). A comparison of multiple imputation and doubly robust estimation for analyses with missing data, Journal of the Royal Statistical Society: Series A (Statistics in Society), 169, 571-584.  https://doi.org/10.1111/j.1467-985X.2006.00407.x
  4. Cheng D, Li J, Liu L, Le TD, Liu J, and Yu K (2022). Sufficient dimension reduction for average causal effect estimation, Data Mining and Knowledge Discovery, 36, 1174-1196.  https://doi.org/10.1007/s10618-022-00832-5
  5. Cook RD (1996). Graphics for regressions with a binary response, Journal of the American Statistical Association, 91, 983-992. https://doi.org/10.1080/01621459.1996.10476968
  6. Cook RD (2009). Regression Graphics: Ideas for Studying Regressions through Graphics, John Wiley & Sons, New York. 
  7. Cook RD and Weisberg S (1991). Sliced inverse regression for dimension reduction: Comment, Journal of the American Statistical Association, 86, 328-332.  https://doi.org/10.1080/01621459.1991.10475036
  8. De Luna X, Waernbaum I, and Richardson TS (2011). Covariate selection for the nonparametric estimation of an average treatment effect, Biometrika, 98, 861-875.  https://doi.org/10.1093/biomet/asr041
  9. Dong Y and Li B (2010). Dimension reduction for non-elliptically distributed predictors: Second-order methods, Biometrika, 97, 279-294.  https://doi.org/10.1093/biomet/asq016
  10. Fukumizu K, Bach FR, and Jordan MI (2004). Dimensionality reduction for supervised learning with reproducing kernel Hilbert spaces, Journal of Machine Learning Research, 5, 73-99.  https://doi.org/10.21236/ADA446572
  11. Ghasempour M and de Luna X (2021). SDRcausal: An R package for causal inference based on sufficient dimension reduction, Available from: arXiv preprint arXiv:2105.02499 
  12. Ghosh T, Ma Y, and De Luna X (2021). Sufficient dimension reduction for feasible and robust estimation of average causal effect, Statistica Sinica, 31, 821-842.  https://doi.org/10.5705/ss.202018.0416
  13. Glymour M, Pearl J, and Jewell NP (2016). Causal Inference in Statistics: A Primer, John Wiley & Sons, United Kingdom. 
  14. Glynn AN and Quinn KM (2010). An introduction to the augmented inverse propensity weighted estimator, Political Analysis, 18, 36-56.  https://doi.org/10.1093/pan/mpp036
  15. Hahn J (1998). On the role of the propensity score in efficient semiparametric estimation of average treatment effects, Econometrica, 66, 315-331.  https://doi.org/10.2307/2998560
  16. Kang JD and Schafer JL (2007). Demystifying double robust-ness: A comparison of alternative strategies for estimating a population mean from incomplete data, Statistical Science, 22, 523-539.  https://doi.org/10.1214/07-STS227
  17. Li B and Dong Y (2009). Dimension reduction for nonelliptically distributed predictors, The Annals of Statistics, 37, 1272-1298.  https://doi.org/10.1214/08-AOS598
  18. Li B and Wang S (2007). On directional regression for dimension reduction, Journal of the American Statistical Association, 102, 997-1008.  https://doi.org/10.1198/016214507000000536
  19. Li KC (1991). Sliced inverse regression for dimension reduction, Journal of the American Statistical Association, 86, 316-327.  https://doi.org/10.1080/01621459.1991.10475035
  20. Liu J, Ma Y, and Wang L (2018). An alternative robust estimator of average treatment effect in causal inference, Biometrics, 74, 910-923.  https://doi.org/10.1111/biom.12859
  21. Ma Y and Zhu L (2012). A semiparametric approach to dimension reduction, Journal of the American Statistical Association, 107, 168-179.  https://doi.org/10.1080/01621459.2011.646925
  22. Ma Y and Zhu L (2014). On estimation efficiency of the central mean subspace, Journal of the Royal Statistical Society: Series B: Statistical Methodology, 76, 885-901.  https://doi.org/10.1111/rssb.12044
  23. Ma S, Zhu L, Zhang Z, Tsai CL, and Carroll RJ (2019). A robust and efficient approach to causal inference based on sparse sufficient dimension reduction, Annals of Statistics, 47, 1505-1535.  https://doi.org/10.1214/18-AOS1722
  24. Mukherjee B and Chatterjee N (2008). Exploiting gene-environment independence for analysis of case-control studies: An empirical Bayes-type shrinkage estimator to trade-off between bias and efficiency, Biometrics, 64, 685-694.  https://doi.org/10.1111/j.1541-0420.2007.00953.x
  25. Robins JM, Rotnitzky A, and Zhao LP (1994). Estimation of regression coefficients when some regressors are not always observed, Journal of the American Statistical Association, 89, 846-866.  https://doi.org/10.1080/01621459.1994.10476818
  26. Robins JM, Rotnitzky A, and Zhao LP (1995). Analysis of semiparametric regression models for repeated outcomes in the presence of missing data, Journal of the American Statistical Association, 90, 106-121.  https://doi.org/10.1080/01621459.1995.10476493
  27. Rubin DB (1973). Matching to remove bias in observational studies, Biometrics, 29, 159-183.  https://doi.org/10.2307/2529684
  28. Vansteelandt S, Bekaert M, and Claeskens G (2012). On model selection and model misspecification in causal inference, Statistical Methods in Medical Research, 21, 7-30.  https://doi.org/10.1177/0962280210387717
  29. Wager S and Athey S (2018). Estimation and inference of heterogeneous treatment effects using random forests, Journal of the American Statistical Association, 113, 1228-1242.  https://doi.org/10.1080/01621459.2017.1319839
  30. Ye Z and Weiss RE (2003). Using the bootstrap to select one of a new class of dimension reduction methods, Journal of the American Statistical Association, 98, 968-979.  https://doi.org/10.1198/016214503000000927
  31. Yuan M and Lin Y (2006). Model selection and estimation in regression with grouped variables, Journal of the Royal Statistical Society: Series B (Statistical Methodology), 68, 49-67.  https://doi.org/10.1111/j.1467-9868.2005.00532.x