The analysis of truncated data when an unobserved latent structure is assumed is considered. A Bayesian analysis of truncated data in the presence of covariates is presented. A method is derived for the analysis of truncated data when the response for the unobserved members is known. The Gibbs sampler is used to approximate the required posteriors. The method is applied to some simulated data and to a real data set.
Download Complete Document: trunc_regress [PDF: 779 KB]
Type: Research and Analysis Report
Sub Type: Grant
Author(s): O’Neill & Barry
Topics: Methodology, Statistics
Publication Date: 01/01/94