Abstract:
The importance of small area estimation in survey sampling is increasing, due to
the growing demand for reliable small area estimation from both public and
private sectors. In this paper, we address the important issue of using statistical
modeling techniques to compute more reliable small area estimates. The main aim
is to assess the use of a flexible methodology for small area estimation. We
formulate a new flexible small area model by incorporating a tuning (index)
parameter into the standard area-level (Fay-Herriot) model. We achieve this using
a combination of two methods namely, empirical Bayes (EB) approach and
hierarchical Bayes (HB) approach. Our results suggest that the proposed model can
be seen as advancement over the standard Fay-Herriot model. The novelty here is
that we have developed a flexible way to handle random effects in small area
estimation. The Implementation of the proposed model is only mildly more
difficult than the Fay-Herriot model. We have obtained results for both EB
approach and the HB approach. Compared with the corresponding HB procedure,
the EB approach saves a tremendous computing time and is very simple to
implement.
