|
Activity Number:
|
100
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Monday, August 3, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #305116 |
|
Title:
|
A Generalized Skewed Model for Binary Response Data
|
|
Author(s):
|
Xia Wang*+ and Dipak K. Dey
|
|
Companies:
|
University of Connecticut and University of Connecticut
|
|
Address:
|
215 Glenbrook Rd. U-4120, Storrs, CT, 06269-4120,
|
|
Keywords:
|
Complementary log-log link ; Generalized extreme value distribution ; Latent variable ; Markov chain Monte Carlo ; Posterior distribution ; Skewness
|
|
Abstract:
|
A critical issue involved in modeling binary response data is the appropriate choice of the link function. In this paper we introduce a new flexible skewed link function for modeling such data based on the generalized extreme value (GEV) distribution. The proposed GEV links provide much more flexible and improved skewed link regression models than the existing skewed links, especially when there exists extreme difference between the number of 0's and 1's for which the response curve would be extremely skewed. We first explore theoretical properties of the proposed links and the propriety of posterior distributions under various proper and improper priors. The flexibility of the proposed model is then illustrated by extensive simulation as well as through a billing data set from a Fortune 100 company. The Deviance Information Criterion is used for guiding the choice of link functions.
|
