GENERALIZED DUAL ESTIMATORS FOR ESTIMATING MEAN USING SUB-SAMPLING THE NON-RESPONDENTS

In this paper an improved class of generalized dual estimators is proposed for estimating the population mean considering the prevalence of non-response in two different cases regarding the availability of population mean of the supportive variable. Expressions for the bias and the mean square error of the advised generalized class of dual estimators in each case are derived along with the optimum conditions which make the proposed generalized estimator more efficient than some of the existing estimators. Empirical studies have also been provided to establish the advantage of the advised generalized class of dual estimators. This paper illustrates that the proposed estimators perform better in each case than the existing estimators.