Improved predictive estimators for finite population mean using Searls technique

Abstract

When the prediction approach recommended by Basu (1971) is used to estimate the finite population mean. It is detected that utilizing conventional mean per unit estimator, ratio estimator, and regression estimator used as a predictor for the mean of unobserved units in the population, yields the same corresponding conventional estimators of the mean of the total population. But, in the case of product estimator, and the ratio and product exponential estimators [Bahl and Tuteja (1991)], the resulting predictive estimators of the mean of the total population is different from the conventional product estimator, and the ratio and product exponential estimators, have already been shown. In this article improved prediction type estimators are developed corresponding to conventional estimators using the Searls (1964) technique. The prediction type estimators are improved in such a way that even their mean square errors are improved at this time, and has overcome the drawback of same MSE of the predictive estimator in previous predictive estimation. The efficiency comparison clearly illustrates our superiority of the proposed estimators. A numerical study is also provided using real data to examine the performnace of the proposed estimator.