|
Activity Number:
|
501
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Thursday, August 2, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #308453 |
|
Title:
|
Bayesian Penalized Spline Model-Based Estimation of the Finite Population Distribution Function for Unequal Probability Samples
|
|
Author(s):
|
Qixuan Chen*+ and Michael R. Elliott and Roderick J. Little
|
|
Companies:
|
University of Michigan and University of Michigan and University of Michigan
|
|
Address:
|
1984 Traver Rd, Apt 204, Ann Arbor, MI, 48105,
|
|
Keywords:
|
penalized spline regression ; Gibbs sampling ; probit model ; distribution function ; unequal probability samples
|
|
Abstract:
|
This paper develops Bayesian estimation of finite population distribution function for unequal-probability-of-selection samples. The methods allow the probabilities of inclusion to be directly incorporated into the estimation of distribution functions, using an ordinal probit regression or a sequence of binary probit regressions on the penalized spline of the inclusion probabilities. The posterior distribution of the distribution function is then obtained using Gibbs sampling. The proposed methods are compared with the Horvitz-Thompson estimator by simulation, and illustrated using the University of Michigan Dioxin Exposure Study (UMDES) data.
|