PONDERACIONES OPTIMAS: OTRA MIRADA A LA MINIMIZACION DEL ERROR EN LA ESTRATIFICACION

Abstract

The determination of optimal strata weights for minimizing the variance is analysed. It appears as an adequate option for deriving smaller variance´s values than with the classic approach. The use of a superpopulation model enhances to obtain proxys to the unknown parameters needed for calculating the minimum variance estimates. Some examples are worked out for illustrating the behavior of the proposal.