|
Activity Number:
|
132
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 3, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #305023 |
|
Title:
|
Bayesian Method for Misclassified Multinomial Data with Prior Information and External Data
|
|
Author(s):
|
Yi Qian*+ and Deukwoo Kwon and Jeesun Jung
|
|
Companies:
|
Amgen, Inc. and National Cancer Institute and Indiana University School of Medicine
|
|
Address:
|
One Amgen Center Drive, Thousand Oaks, CA, 91320-1799,
|
|
Keywords:
|
Bayesian inference ; misclassified multinomial data ; misclassification
|
|
Abstract:
|
Many epidemiological studied are based on surveys or questionnaires. Some key variables are recorded incorrectly since respondent misunderstands questions or is reluctant to give honest answer. In medical data, some disease is easily misclassified with other type of disease. With either gold standard or double sampling scheme, we may estimate misclassification probabilities. In this presentation we examine how estimates of true proportions are biased when we ignore misclassification. We also show improvement by using prior information and/or external data.
|