DOI QR코드

DOI QR Code

Covariate selection criteria for controlling confounding bias in a causal study

인과연구에서 중첩편향을 제거하기 위한 공변량선택기준

  • Thepepomma, Seethad (Department of Statistics and Actuarial Science, Soongsil University) ;
  • Kim, Ji-Hyun (Department of Statistics and Actuarial Science, Soongsil University)
  • ;
  • 김지현 (숭실대학교 정보통계보험수리학과)
  • Received : 2016.04.27
  • Accepted : 2016.06.10
  • Published : 2016.08.31

Abstract

It is important to control confounding bias when estimating the causal effect of treatment in an observational study. We illustrated that the covariate selection in the causal inference is different from the variable selection in the ANCOVA model. We then investigated the three criteria of covariate selection for controlling confounding bias, which can be used when we have inadequate information to draw a complete causal graph. VanderWeele and Shpitser (2011) proposed one of them and claimed it was better than the other two. We show by example that their criterion also has limitations and some disadvantages. There is no clear winner; however, their criterion is better (if some correction is made on its condition) than the other two because it can remove the confounding bias.

관측 자료를 이용한 인과연구에서 관심 있는 처리변수의 효과가 다른 공변량의 효과와 중첩되지 않도록 조건화할 공변량을 선택하는 것이 중요하다. 인과연구에서의 공변량선택 문제는 공분산분석 모형에서의 변수선택 문제와 다르다는 것을 예를 들어 설명하였다. 그리고 모든 변수들 사이의 인과관계를 파악하지 않고도 적용할 수 있는 실용적인 공변량선택기준에 대해 살펴보았다. VanderWeele과 Shpitser (2011)가 새로운 기준을 제안하면서 새로운 기준이 다른 두 기준보다 나은 성능을 보인다고 주장하였는데, 이 기준에도 한계와 단점이 있음을 예증하였다. 새로운 기준이 완전한 기준은 아니지만 조건을 조금 수정하면 다른 두 기준과 달리 중첩을 제거할 수 있다는 점에서 좀 더 나은 기준이라고 할 수 있다.

Keywords

References

  1. Greenland, S., Pearl, J., and Robins, J. M. (1999). Causal diagrams for epidemiologic research, Epidemiology, 10, 37-48. https://doi.org/10.1097/00001648-199901000-00008
  2. LaLonde, R. J. (1986). Evaluating the econometric evaluations of training programs with experimental data, American Economic Review, 76, 604-620.
  3. Pearl, J. (1993). Comment: graphical models, causality, and intervention, Statistical Science, 8, 266-269. https://doi.org/10.1214/ss/1177010894
  4. Pearl, J. (1995). Causal diagrams for empirical research, Biometrika, 82, 669-688. https://doi.org/10.1093/biomet/82.4.669
  5. Pearl, J. (2009). Causality: Models, Reasoning, and Inference, Cambridge University Press, Cambridge.
  6. Rubin, D. B. (1974). Estimating causal effects of treatments in randomized and nonrandomized studies, Journal of Educational Psychology, 66, 688-701. https://doi.org/10.1037/h0037350
  7. Rubin, D. B. (1990). Formal modes of statistical inference for causal effects, Journal of Statistical Planning and Inference, 25, 279-292. https://doi.org/10.1016/0378-3758(90)90077-8
  8. Rubin, D. B. (2009). Author's reply (to Pearl's, Arvid's and Sjolander's letters to the editor), Statistics in Medicine, 28, 1420-1423. https://doi.org/10.1002/sim.3565
  9. VanderWeele, T. J. and Shpitser, I. (2011). A new criterion for confounder selection, Biometrics, 67, 1406-1413. https://doi.org/10.1111/j.1541-0420.2011.01619.x