|
Activity Number:
|
70
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, July 29, 2007 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #308500 |
|
Title:
|
On Comparison of Different MSE Estimators in Small-Area Estimation
|
|
Author(s):
|
En-Tzu Tang*+ and Jiming Jiang
|
|
Companies:
|
University of California, Davis and University of California, Davis
|
|
Address:
|
424 Russell Park Apt 6, Davis, CA, 95616,
|
|
Keywords:
|
empirical best predictors ; mean squared error ; nested error regression ; two stage estimator
|
|
Abstract:
|
Nested-error regression models are widely used for analysis of clustered data, such as in small-area estimation. These analyses usually focus on prediction of mixed effects such as small area means. An important measure of the performance of the prediction is mean squared prediction error. In this talk, I compare several different methods for estimating the mean squared predictor error by Monte Carlo simulations. I consider situations with both normal and non-normal random effects and errors. The methods being compared include the naive, Prasad-Rao, jackknife and bootstrap methods
|
