Model-Design Approach to Small Area Estimation

Abstract:  Small area estimation has been a hot topic in survey methodology in recent years. Usually the traditional design setup is replaced by model assumptions, and sampling weights are no longer used in estimation procedures. While in some situations it is justified, in general it may not be appropriate. In this presentation, we will combine both model and design setups. First, in a simplistic model, theoretical background for synthetic and composite estimators will be presented. Then BLUE and BLUP's for small areas under more general linear model with small area specific random effects will be discussed. In particular, the issues of non-existence of BLUE and BLUP's, of optimal strategies, and of empirical versions of BLUE and BLUP's will be addressed.

Refreshments and socializing: 3:30 - 4:00 PM