|
Activity Number:
|
65
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, August 6, 2006 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #305829 |
|
Title:
|
Model-Based Sampling Designs for Optimum Estimation
|
|
Author(s):
|
Sun Woong Kim*+ and Steven G. Heeringa and Peter W. Solenberger
|
|
Companies:
|
Dongguk University and University of Michigan and University of Michigan
|
|
Address:
|
JungGu Pil Dong 3 Ga 26, Seoul, 100-715, South Korea
|
|
Keywords:
|
superpopulation model ; probability sampling ; optimization problem
|
|
Abstract:
|
Concerned more with bias than variance, many survey samplers prefer design-based inferences over model-based approaches. There have been many studies of methods for reducing variances of estimates through probability sampling. In designing samples, we may have a belief about or an experience with a population, specified as a superpopulation model. In this paper we show how to determine the optimum sampling plan to minimize sampling variance under a variety of model assumptions. These optimal sampling plans will be based on inclusion probability proportional to size (IPPS) sampling and will satisfy certain desirable properties with respect to variance estimation. The presence of the intercept term in a superpopulation model is an interesting issue in model-based design. We will show that these approaches depend on the form of the variance being considered as well as model assumptions.
|
