SURV798B

Prerequisite: STAT420/SURV420 or equvialent; or permission of instructor. Model-based small-area estimation has portance oer the past two decades. Students will learn the state-of-the-art model-based small-area estimation methods (e.g., empirical best prediction, empirical Bayes andhierarchical Bayes, etc.) and the associated important issues regarding measures of uncertainty, model selection, model diagnostics, design-consistency, etc. The bootstrap, jackknife, and delta methods will be discussed in details in the context of measuring uncertainty of EB/EBP. In order to explain certain concepts, it will be necessary to gothrough a few derivations. Data analyses using several real life examples will be presented. Application of SAS and BUGS in certain small-area data analyses will be shown. The course includes practical exercises. It runs concurrently with the University of Michigan course.