|
Activity Number:
|
504
- Model/Variable Selection
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 2, 2017 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section
|
|
Abstract #323361
|
|
|
Title:
|
A Score Test for Over-Dispersion in Marginalized Zero-Inflated Poisson Regression Models
|
|
Author(s):
|
Gul Inan* and John Preisser and Kalyan Das
|
|
Companies:
|
Middle East Technical University and University of North Carolina-Chapel Hill and University of Calcutta
|
|
Keywords:
|
count data ;
excess zeros ;
marginal models ;
over-dispersion ;
score test
|
|
Abstract:
|
Count data with excess zeros are frequently encountered in a wide variety of disciplines, from environmental and horticultural studies to transportation safety and insurance claim studies. Marginalized zero-inflated Poisson (MZIP) and marginalized zero-inflated negative binomial (MZINB) regression models have been recently proposed for analysis of such data with direct population-based inferences on the marginal mean count that includes "excess zeros". This study proposes a score test for testing a MZIP model against a MZINB model to investigate whether the zero-inflated count data can be better represented via MZIP or MZINB due to possible over-dispersion. The sampling distribution and empirical power of the proposed score test are investigated via a Monte Carlo simulation study and the procedure is illustrated by a horticultural data set.
|
Authors who are presenting talks have a * after their name.
