SOME SHRUNKEN PREDICTORS IN STRATIFIED POPULATION

摘要

Bouza (1990) and Bouza & Allende (1991) proposed shrunken predictors and analyzed their behavior in stratified sampling under super-population models xi(0,1:1). Robustness of the shrunken predictors measured in terms of bias and mean squared error (MSE) was further studied by Bouza (1994) which depends on stratified balanced sample and difference between the common slope (beta) and uncommon slopes ((beta)_h) to strata. In this paper, we proposed a new set of predictors under same super population model xi(0,1:1). The proposed predictors are compared with that of due to Bouza (1994) and they have been found more efficient under certain conditions. The robustness of the proposed predictor is also studied on the same lines as discussed by Bouza (1994), and it has been found that robustness of the proposed predictor depends upon the difference between beta and (beta)_h. Simulation experiments also indicated superiority of the proposed predictor over the predictors due to Bouza.