|
Activity Number:
|
332
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #309793 |
|
Title:
|
Bayesian Modeling of Multivariate Binary Response with a Change-Point for Applications to Adolescent Offender Registry
|
|
Author(s):
|
Dipankar Bandyopadhyay*+ and Debajyoti Sinha and Stuart R. Lipsitz and Elizabeth Letourneau
|
|
Companies:
|
Medical University of South Carolina and Medical University of South and Brigham and Women's Hospital and Medical University of South Carolina
|
|
Address:
|
135 Cannon Street Suite 303, Charleston, SC, 29425,
|
|
Keywords:
|
binary ; bridge ; changepoint ; logit ; multivariate
|
|
Abstract:
|
We develop a model for multivariate binary data using random effects logistic regression model incorporating a change-point. To preserve marginal logit structure (integrating out random effects) for interpretable co-variate effects, we model the random effects using a bridge density and compare it with normal random effects that doesn't preserve the logit link marginally. We analyze the data from South Carolina adolescent sex repeat-offenders to investigate the effects of different factors and presence of any change-point following enactment of a new set of mandatory sentencing laws (occurring between 1992--1996) regarding the solicitors' decision and judges' adjudication. A discrete prior on the change-point is elucidated to reflect the expert's opinion on possible year of change in judges/solicitors opinions and posteriori conclusion on changepoint occurrence between 1992--1996.
|
