The Korean Journal of Applied Statistics (응용통계연구)

The Korean Statistical Society (한국통계학회)
권호 표지
DOI QR코드

DOI QR Code

Comparison of GEE Estimators Using Imputation Methods

대체방법별 GEE추정량 비교

  • Published : 2003.09.01
Download PDF
⟨ Previous Next ⟩

Abstract

We consider the missing covariates problem in generalized estimating equations(GEE) model. If the covariate is partially missing, GEE can not be calculated. In this paper, we study the performance of 7 imputation methods to handle missing covariates in GEE models, and the properties of GEE estimators are investigated after missing covariates are imputed for ordinal data of repeated measurements. The 7 imputation methods include i) Naive Deletion ii) Sample Average Imputation iii) Row Average Imputation iv) Cross-wave Regression Imputation v) Carry-over Imputation vi) Bayesian Bootstrap vii) Approximate Bayesian Bootstrap. A Monte-Carlo simulation is used to compare the performance of these methods. For the missing mechanism generating the missing data, we assume ignorable nonresponse. Furthermore, we generate missing covariates with or without considering wave nonresp onse patterns.

본 연구에서는 범주형 반복측정자료의 일반화추정방정식(GEE)모형에서 결측이 발생할 경우 결측값 대체(imputation)방법들에 대한 성능을 비교하고자 한다. 설명변수 X가 부분적으로 결측을 갖는 경우 GEE추정량을 계산할 수 없다. 본 논문에서는 시점에 따라 값이 변하는 설명변수에 결측이 있는 경우 GEE모형에서 결측값을 추정하는 7가지의 대체방법을 다루며, 실제자료와 모의실험을 통하여 대체방법별 GEE추정량의 성질을 연구한다. 대체방법별 GEE추정량의 성능을 비교하기 위해 우리는 반응변수가 범주형인 반복측정모형에서 완전자료의 GEE추정량과 완전자료에서 결측을 생성하여 결측값에 각 대체방법을 적용하여 대체한 후 구한 GEE추정량을 비교한다. 대체방법으로는 (1) 단순삭제 (2) 표본 평균대체 (3) 행 평균대체 (4) 횡 시점 회귀대체 (5) 이월대체 (6) 베이지안 붓스트랩 (7) 근사적 베이지안 붓스트랩에 대해서 살펴본다. 결측과정(missing mechanism)은 무시할 수 있는 무응답(ignorable nonresponse)을 가정하며, 결측 발생에 대해서는 원자료의 시점 무응답 패턴(wave nonresponse pattern)을 고려하여 발생시키거나 또는 시점 무응답 패턴을 고려하지 않고 단순임의추출로 결측을 발생시키는 방법을 각각 고려한다.

Keywords

References

  1. 응용통계연구 v.15 no.2 범주형 반복측정자료를 위한 일반화 추정방정식의 소표본 특성 김동욱;김재직 https://doi.org/10.5351/KJAS.2002.15.2.297
  2. Journal of the Royal Statistical Society B v.57 Regression models for longitudinal binary responses with informative drop-outs Fitzmaurice,G.M.;Molenberghs,G.;Lipsitz,S.R.
  3. The American Statistician v.49 no.2 Generating multivariate categorical variates using the iterative proportional fitting algorithm Gange,S.J. https://doi.org/10.2307/2684626
  4. Biometrics v.50 An application of maximum likelihood and generalized estimating equations to the analysis of ordinal data from a longitudinal study with cases missing at random Kenward,M.G.;Lesaffre,E.;Molenberghs,G. https://doi.org/10.2307/2533434
  5. Statistics in Medicine v.7 Missing data in longitudinal studies Laird,N. https://doi.org/10.1002/sim.4780070131
  6. Panel Survey Treatment of wave nonresponse in panel surveys Lepkowski,J.M.;D.Kasprzyk(ed.)
  7. Biometrika v.73 Longitudinal data analysis using generalized linear models Liang,K.Y.;Zeger,S.L. https://doi.org/10.1093/biomet/73.1.13
  8. Statistics in Medicine v.13 Analysis of repeated categorical data using generalized estimating equations Lipsitz,S.R.;Kim,K.;Zhao,L. https://doi.org/10.1002/sim.4780131106
  9. Statistical Analysis with Missing Data Little,R.J.A.;Rubin,D.B.
  10. Journal of the American Statistical Association v.92 The generalized estimating equation approach when data are not missing completely at random Paik,M.C. https://doi.org/10.2307/2965402
  11. Journal of the Korean Statistical Society v.31 A combined method compensating for wave nonresponse Park,J.
  12. Journal of the American Statistical Association v.90 Analysis of semiparametric regression models for repeated outcomes in the presence of missing data Robins,J.M.;Rotnitzky,A.;Zhao,L.P. https://doi.org/10.2307/2291134
  13. Biometrika v.63 Inference and missing data Rubin,D.B. https://doi.org/10.1093/biomet/63.3.581
  14. Multiple Imputation for Nonresponse in Surveys Rubin,D.B.
  15. Journal of the American Statistical Association v.81 Multiple imputation for interval estimation from simple random samples with ignorable nonresponse Rubin,D.B.;Schenker,N. https://doi.org/10.2307/2289225
  16. Biometrika v.61 Quasi-likelihood functions, generalized linear models, and the Gauss-Newton method Wedderburn,R.W.M.
  17. Biometrics v.53 Generalized estimating equation model for binary outcomes with missing covariates Xie,F.;Paik,M.C. https://doi.org/10.2307/2533511
  18. Biometrics v.53 Multiple imputation methods for the missing covariates in generalized estimating equation Xie,F.;Paik,M.C. https://doi.org/10.2307/2533521