|
Activity Number:
|
110
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 3, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #303275 |
|
Title:
|
Modeling Count Data and Confidence Interval for the Over-dispersion Parameter, with Applications to Biological Data
|
|
Author(s):
|
Krishna K. Saha and Debaraj Sen*+ and Roger Bilisoly
|
|
Companies:
|
Central Connecticut State University and Concordia University and Central Connecticut State University
|
|
Address:
|
Department of Mathematics and Statistics, Montreal, PQ, H3G 1M8, Canada
|
|
Keywords:
|
Asymptotic variance ; Biological data ; Confidence interval ; Negative binomial model ; Over-dispersed parameter
|
|
Abstract:
|
In this article, we discuss the modeling of count data occurring in biological applications. We then derive asymptotic procedures for the construction of confidence limits for the over-dispersion parameter of count data when there is no likelihood available. We also obtain closed-form asymptotic variance formulae for the estimator of the over-dispersion parameter. We finally conduct a simulation study to compare these with a procedure using the maximum likelihood estimator based on the negative binomial model, in terms of the coverage. It appears that confidence interval based on the method of moments or the double extended quasi-likelihood of Lee and Nelder ({\it Biometrika} 2001, 88, 987-1006) is better for smaller deviations from the Poisson assumption and larger sample sizes. An example using biological data illustrates these procedures.
|
