Abstract: | A large class of estimators is considered for the mean of a finite population using information on an auxiliary variable. It is shown that members of this class of estimators are asymptotically no more efficient than the linear regression estimator. |