Predictive Estimation of Finite Population Variance

Abstract

Adopting predictive approach, estimators are proposed for population variance [Formula: see text] using different predictors for mean and variance of unobserved units in the population. Asymptotic expressions for bias and mean square error of these new estimators are obtained and compared with those of some known estimators of population variance. Predictive character of some known estimators is also examined . An empirical study demonstrated superiority over others of one of the proposed estimators, which uses regression estimators as predictors for mean and variance of unobserved units. Also, one of the other proposed estimators which does not utilize any auxiliary information has been found to be more efficient than the traditional unbiased estimator [Formula: see text].