Linear Estimation Under Superpopulation Models: Somo Results on Robustness

The present article examines the lineal estimation of a finite population mean, under a regression superpopulation model with unknown parameters and correlated residuals. Asymptotic design unbiasedness and weak robustness are desirable properties of the estimators when the model is misspecified. In this sense, two procedures of choosing among weakly robust and asymptotically desing unbiased linear estimators, are suggested.