A model-based approach to estimation for small subgroups of a population.

Abstract Sample surveys rarely provide enough data to permit accurate estimation for small subgroups of the population by using the standard methods based on the selection probabilities. In this case a possible procedure is to base the required estimate on data from the subgroup concerned, supplemented by that from similar subgroups. Such an approach makes implicit use of the population structure. This article attempts to model that structure explicitly and obtains estimates within a predictive framework. The effect of model misspecification on the bias of the estimators is also examined.