|
Activity Number:
|
138
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Monday, July 30, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #309725 |
|
Title:
|
Expert Opinion, Informative Priors, and Sensitivity Analysis for Longitudinal Binary Data with Informative Dropout
|
|
Author(s):
|
Joo Yeon Lee*+ and Joseph W. Hogan
|
|
Companies:
|
Food and Drug Administration and Brown University
|
|
Address:
|
1733 Anderson RD, Falls Church, VA, 22043,
|
|
Keywords:
|
Informative dropout ; Pattern mixture model ; Sensitivity analysis ; Prior elicitation ; Informative priors
|
|
Abstract:
|
Elicitation of expert opinions is a crucial role in fully Bayesian way of sensitivity analysis. In this paper we use a pattern mixture model, composed of transition model within pattern, for repeated binary data with nonignorable dropout and introduce parameter identification scheme which allows the analyst to explore the effects of possibly nonignorable dropout. We show how to elicit prior distributions that reflect beliefs about the distribution of missing responses using experts opinions and how different prior beliefs can affect study conclusion. Methods are illustrated using data from the OASIS study, a longitudinal clinical trial of a motivational intervention for smoking cessation in smokers participating in outpatient alcohol treatment.
|
