A class of separate regression type estimator under stratified random sampling.

Abstract

In this paper, a class of separate regression type estimators using the auxiliary information on strata means and strata variances is proposed under stratified random sampling. The expressions of its bias and mean square error under are obtained. Further the expression of minimum mean square error under the optimum value of the characterizing scalar is also given. An optimum allocation with the proposed class is obtained and its efficiency is compared with that of Neyman optimum allocation. Finally, a comparative study is made with that of separate ratio estimator, separate Singh (2003) product estimator, separate linear regression estimator and the usual stratified sample mean. Lastly, it is shown that the proposed allocation is always more efficient in the sense of having smaller mean square error than Neyman allocation.