|
Activity Number:
|
153
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, July 30, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #309399 |
|
Title:
|
Bayesian Weight-Smoothing Models in Clustered or Cross-Classed Sample Designs
|
|
Author(s):
|
Xiaobi Huang*+ and Michael R. Elliott
|
|
Companies:
|
University of Michigan and University of Michigan
|
|
Address:
|
2013 Medford Road, Ann Arbor, MI, 48104,
|
|
Keywords:
|
Sampling weights ; clustered sample design ; cross-classed sample designs ; weight trimming ; weight smoothing
|
|
Abstract:
|
Highly disproportional sample designs have large weights, which will introduce undesirable variability in statistical estimates. Weight trimming fixes a cutpoint weight and sets larger weights to this cutpoint value while adjusts weights below this value to maintain the untrimmed weight sum, reducing variability at the cost of introducing some bias. Previous work developed Bayesian "weight smoothing" models to produce general model-based weight trimming estimators of population statistics, but has been limited to the context of stratified and post-stratified sample designs. This presentation extends the Bayesian "weight smoothing" methodology to a more general class of complex sample design that include single or multi-stage cluster samples and/or strata that "cross" the weight strata. The methods are applied to linear and generalized linear regression models.
|
