Interface 2003
Abstract

Parametric Bootstrap Confidence Intervals in Small Area Estimation Problems
Snigdhansu Chatterjee, (U. Minnesota), chatterjee@stat.umn.edu, and
P. Lahiri, (U. Maryland), plahiri@survey.umd.edu

Abstract

We propose a parametric bootstrap method to construct confidence interval for a small area mean in the Fay-Herriot model. The method is accurate for coverage probability upto O(n^{-3/2}), a property not established for competing methods. Simulations show among competing techniques it also has the shortest length.


Take me back to the main conference page.