ABSTRACT
When statistical information is demanded at local level, often the survey sample sizes in the local areas are not large enough to produce reliable area-specific or direct estimates and then small area estimation techniques are required. Most of these techniques are based on models that establish some relation across all areas between the target variable and some available auxiliary variables, and this relation allows us to “borrow strength” from all areas to find estimates in a particular area. So far model-based small area estimation has largely focused on means or totals, using either area level models or unit level models. Empirical best linear unbiased prediction (EBLUP), empirical Bayes or empirical best (EB) and hierarchical Bayes (HB) methods have been extensively used for point estimation and for measuring the variability of the estimators. The book by Rao (2003) gives an extensive account of those methods. Several review papers have also appeared after the publication of Rao’s book that cover newer developments in small area estimation, see e.g. Jiang and Lahiri (2006) and Datta (2009).
