Simultaneously model-unbiased, design-unbiased estimation

This paper proposes a class of inferential procedures (incorporating both d esign and estimation elements) that yield estimates of means that are simul taneously model unbiased and design unbiased. Classical regression procedur es yield conditionally unbiased estimators for the mean (conditioning on th e model and choice of observation points). In contrast, design-based method s yield estimators that are unconditionally unbiased no matter what the for m of the underlying model. Variance properties of the proposed class are ex amined, and applications to bioavailability, water quality from mine run-of f, and finite population regression estimation are considered. The proposed procedures perform well, especially in the typical case where a model is o nly approximately correct.