A Class of Robust Sampling Designs for Large-Scale Surveys

This is the authors' abstract. We don't add key points for this paper.

A combined estimation and selection scheme is proposed for use in the context of large-scale enterprise and establishment surveys. The estimator is asymptotically design unbiased and subject to this constraint has minimum expected variance under a superpopulation model. The main advantage of this scheme over current practice is the removal of the need for size stratification. The relation of both the current and the proposed schemes to the Royall and Hereon (1973a,b) stratified sampling schemes is also considered.