Some Contributions in Two-Stage Sampling Using Auxiliary Information

A ratio estimator tR and a regression estimator tRG, as alternatives to the classical ratio estimator and classical regression estimator respectively, have been developed.A comparison between chain-type and predictive-type estimators shows that the chain regression estimator is unconditionally more efficient than the predictive regression estimator. On the other hand, the chain ratio and product estimators are conditionally more efficient than their respective predictive estimators. The chain regression method of estimation is then employed to estimate the population mean on the current occasion when the two-stage sampling with SRSWOR is adopted on the current and a previous occasion. For simplicity, we consider the partial matching among FSUs only although the matching problems among SSUs and both FSUs and SSUs can be easily tackled. we consider a general class of estimators of Y. This class is very flexible in the sense of being reduced to a class of separate variety of estimators when we do not consider z and to a combined variety of estimators when we do not consider x. The design-based properties of the class of estimators have also been studied.