Imputation for missing data and variance estimation when auxiliary information is incomplete

Article type: Research Article

Authors: Liu, Li^a | Tu, Yujuan^a | Li, Yingfu^b | Zou, Guohua^{a; *}

Affiliations: [a] Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing 100080, P.R. China | [b] Department of Mathematical Sciences, University of Houston-Clear Lake, Houston, TX 77058-1098, USA

Correspondence: [*] Corresponding author. E-mail: [email protected]

Abstract: In this paper, a new imputation is developed for the treatment of missing data, which removes the bias of usual ratio imputation. The Jackknife variance estimators are provided and they are proved to be approximately design-unbiased under uniform response. Our method is compared with the existing method of Sitter and Rao [9] through simulation studies. The simulation studies show that the new method for missing data does reduce the bias without increasing the variance significantly.

Keywords: Nonresponse, missing data, ratio imputation, Jackknife variance estimator, incomplete auxiliary variable

DOI: 10.3233/MAS-2005-1204

Journal: Model Assisted Statistics and Applications, vol. 1, no. 2, pp. 83-94, 2006