This is the program for the 2010 Joint Statistical Meetings in Vancouver, British Columbia.

Abstract Details

Activity Number: 143
Type: Contributed
Date/Time: Monday, August 2, 2010 : 8:30 AM to 10:20 AM
Sponsor: Section on Survey Research Methods
Abstract - #307393
Title: Modeling Log-Linear Restricted Conditional Probabilities for Prediction in Surveys
Author(s): Yves Thibaudeau*+ and Eric Slud and Alfred Gottschalck
Companies: U.S. Census Bureau and U.S. Census Bureau and U.S. Census Bureau
Address: , , ,
Keywords: Log-Linear Model ; Conditional Probability ; Post-Stratification
Abstract:

It is shown how to model conditional probabilities subject to log-linear constraints and construct estimators that are more efficient than the Horvitz-Thompson weighted sum estimator for estimating population sizes in the context of a poststratified survey with unknown stratum sizes. Our approach rests on computing an ``hybrid" predictor, similar to that of Pfeffermann et al. (1998), and the expansion of a specific parameterization for conditional probabilities restricted by log-linear constraints, as proposed by Thibaudeau (2003). This parameterization facilitates the computation of MLE's and makes it possible to apply the method of Laplace for estimating variances of MLE's. The method of Laplace is strictly model-based. To assess its accuracy we compare it with balanced repeated replication, a design-based method.


The address information is for the authors that have a + after their name.
Authors who are presenting talks have a * after their name.

Back to the full JSM 2010 program




2010 JSM Online Program Home

For information, contact jsm@amstat.org or phone (888) 231-3473.

If you have questions about the Continuing Education program, please contact the Education Department.