Title: Overdispersion of Regression Models
Speaker: Feng Jin (Case Western Reserve University)
Advisor: Danhong Song (Lecturer, Case Western Reserve University)
Abstract: Overdispersion in Poisson models occurs when the response variance is greater than the mean.
It may cause standard errors of the estimates to be deflated or underestimated. In this presentation, the data set contains information on smoking behavior and other variables for a random sample of single adults from the United States. In order to evaluate how average number of cigarettes smoked per day is influenced, I establish poisson regression, negative binomial regression and zero- inflated regression. These regressions allow us to see whether an overdispersion is apparent or real. In addition, by comparing three different regression estimates through BIC, Likelihood ratio test and Deviance goodness-of-fit test, I will provide the best prediction on smoking behavior.