Two-fold subarea models for small area estimation: Some theory and model selection

• Activity Number: 631 - Small Area Estimation: “Producing Estimates for Small Areas from Sampled Data”
• Type: Invited
• Date/Time: Thursday, August 1, 2019 : 10:30 AM to 12:20 PM
• Sponsor: Survey Research Methods Section
• Author(s): J. N. K. Rao*
• Companies: Carleton University
• Keywords: empirical best prediction; importance sampling; information criterion; small area estimaton; subarea models; variable selection

Abstract

We study some inference problems under two-fold subarea models with random effects, which generalize the celebrated Fay-Herriot area model. We develop the theory of empirical best prediction for a subarea parameter under unmatched two-fold subarea models. In addition, we propose an effective variable selection method under matched two-fold subarea models, based on transformation. Simulation results are provided to illustrate the strengths of the proposed methods.