225 – Recent Developments on Learning Methods from Manifold Data and Related Theory
Parametric Bootstrap Confidence Intervals for Survey-weighted Small-area Proportions
Benmei Liu
National Cancer Institute
Mamadou Diallo
Westat
In this paper, we apply the recently developed parametric bootstrap method in constructing confidence intervals for the well-known Fay-Herriot model in estimating survey-weighted small area proportions. Through design-based simulation studies from a real finite population and extensive purely model-based simulation studies, we examine the coverage properties of the parametric bootstrap confidence intervals for the Fay-Herriot model. We also compare them with those obtained from other competing methods for the same model.