Estimation in Finite Population Under a Calibration Model

In this paper, we address the estimation of a finite population mean under a calibration model. We describe a general class of regression estimators that yield the standard regression estimator and the alternative classical estimator as special cases. Formulas are derived for the asymptotic bias and variance of the general regression estimator. Simulations were carried out to compare three special estimators on the basis of bias, relative efficiency, and robustness to nonnormality of the model error. It is shown that the standard regression estimator is the most efficient and robust of all three. Also, estimation of the bias and variance of this regression estimator is examined, and several variance estimators are compared.