|
Activity Number:
|
492
|
|
Type:
|
Invited
|
|
Date/Time:
|
Wednesday, August 5, 2009 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #302795 |
|
Title:
|
Fence Method for Prediction Problems and Its Application in Small Area Estimation
|
|
Author(s):
|
Jiming Jiang*+ and Thuan Nguyen and J. Sunil Rao
|
|
Companies:
|
University of California, Davis and Oregon Health & Science University and Case Western Reserve University
|
|
Address:
|
Department of Statistics, Davis, CA, 95616,
|
|
Keywords:
|
Fence method ; Model selection ; Prediction ; Small area estimation
|
|
Abstract:
|
Fence method (Jiang et al. 2008) is a recently developed model selection strategy that is particularly suitable for nonconventional problems such as mixed model selection. It has been assumed that there exists a true model among the candidate models. We extend the fence method so that such a requirement is no longer needed. Furthermore, we develop two predictive measures of lack-of-fit for fence methods that are suitable for prediction problems. The method is applied to small area estimation problems. Methodology developments are supported by simulation studies and real data analyses.
|
