Application of NORM to the Multiple Imputation for Multivariate Missing Data

  • 발행 : 2002.10.31

초록

The statistical analysis of incomplete data sometimes requires handling of incomplete observations. Towards this end, each case with some missing values generally should be deleted, namely, resulting in only use of non-missing cases. EM algorithm(Dempster et al., 1977) which involves prediction and estimation steps is a general method among others. In this article, we use the free software NORM developed for multiple imputation, which uses DA(Data Augmentation) algorithm in its imputation, and evaluate its efficiency through a numerical example.

키워드

참고문헌

  1. Journal of the Royal Statistical Society Series v.B39 Maximum-likehood estimation from incomplete data via the EM algorithm (with discussion) Dempster, A. P.;Laird, N. M.;Rubin, D. B.
  2. Journal of the American Statistical Association v.78 no.426 Missing Data, Imputation, and Bootstrap Efron, B.
  3. Applied Multivariate Statistical Analysis(3rd ed.) Johnson, R. A.;Wichern, D. W.
  4. Statistical Analysis with Missing Data Liitle, R. J. A.;Rubin, D. B.
  5. Proc. 6th Berkely Symposium on Math. Statist and Prob. v.1 A mssing information principle: Theory and application Orchard, T.;Woodbury, M. A.
  6. Multiple Imputation for Nonresponse in Surveys Rubin, D. B.
  7. Analysis of Incomplete Multivariate Data Schafer, J. L.
  8. Multivariate Behavioral Research v.33 Multiple imputation for multivariate missing-data problems: a data analyst and perspective Schafer, J. L.;Olsen, M. K.
  9. Statistical Methods in Medical Research v.8 no.1 Multiple imputation: a primer Schafer, J. L.
  10. Journal of the American Statistical Association v.82 The calculation of posterior distributions by data augmentation (with discussion) Tanner, M. A.;Wong, W. H.
  11. 한국통계학회 2001년 춘계 학술발표회 논문집 결측자료의 대치에 관한 방법적 비교 김현정;문승호;신재경