Robust estimation of variance in the presence of nearest neighbour imputation

Abstract

The problem of estimating the variance of an estimator of the population total when missing values have been filled using a Nearest Neighbour (NN) imputation method is considered. The estimator is developed assuming a more general model than those considered in earlier studies. In an empirical study involving two artificial populations, the proposed estimator is found to perform better or as well as other two estimators in the current use.


African Journal of Science and Technology Vol.4(2) 2003: 5-11