Online Program Home
  My Program

Abstract Details

Activity Number: 567 - Recent Advances in Small Area Estimation with Applications and Evaluation of the Estimates
Type: Topic Contributed
Date/Time: Wednesday, August 2, 2017 : 2:00 PM to 3:50 PM
Sponsor: Government Statistics Section
Abstract #324460
Title: Robust Hierarchical Bayes Small Area Estimation
Author(s): Gauri Datta* and Adrijo Chakraborty and Abhyuday Mandal
Companies: University of Georgia and US Census Bureau and NORC at the University of Chicago and University of Georgia
Keywords: American Community Survey ; Exponential power distribution ; Fay-Herriot model ; Nested error regression ; Outliers ; Posterior propriety
Abstract:

In this talk, we review existing robust Bayesian methods to predict small area means in the presence of outliers. To account for the effects of outliers in prediction of small area means, some researchers proposed a scale-mixture of normal distributions to model error distributions; some others proposed a two-component mixture with respect to the scale parameter of the normal distribution. As an alternative we consider an exponential power model for the error distribution. We complete our model specification by using a class of noninformative priors. We explore the propriety of the resulting posterior densities. We illustrate and compare different methods to estimate poverty rates of school-children in the U.S. counties based on data from the American Community Survey. We conduct a simulation study to investigate effectiveness of various solutions.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2017 program

 
 
Copyright © American Statistical Association