Calibrating Design Weights in Stratified Sampling Using Auxiliary Information on Coefficient of Variation

Calibration approach is widely used survey sampling that incorporates auxiliary information to increase the precision of survey estimates. In this manuscript, we propose two new calibration estimators of population mean in stratified sampling, using the known auxiliary information on mean and coefficient of variation. The calibration estimators were written in the form of generalized linear regression to derive their estimated variance and to construct the confidence interval of population mean. A numerical example is presented to illustrate the application and computational details of the proposed calibration estimators. Moreover, a simulation study is carried out to compare the performance of the proposed calibration estimators. The study reveals that the proposed estimator I is more efficient than some common calibration estimators in stratified sampling. It was also found from this study that calibration constraints on weight and/or variance does not improve the estimate.