Disadvantages of negbin regression Does anybody know if there're common known disadvantages of a negbin regression? In my opinion it seems to fit every problem pretty good (measured with the estimated dispersionparameter). So why not always use it?
 A: There is another nice paper that discusses the differences between quasi-poisson and negative binomial models for overdispersed count data and their effects on prediction results which can be dramatically different:
ver Hoef, J.M., & Boveng, P.L. (2007). Quasi-Poisson versus negative binomial regression: how should we model overdispersed count data. Ecology, 88, 2766-2772.
the essence from the abstract is 
"The variance of a quasi-Poisson model is a linear function of the mean while the variance of a negative binomial model is a quadratic function of the mean. These variance relationships affect the weights in the iteratively weighted least-squares algorithm of fitting models to data. Because the variance is a function of the mean, large and small counts get weighted differently in quasi-Poisson and negative binomial regression."
A: In addition to the paper linked by Andy, which can pretty much be summarized as "using the negative binomial distribution can fix your problem, but don't assume it does", you are estimating another parameter when you switch from Poisson to Negative Binomial, and you'll pay for the protection of that extra parameter through a decrease in precision. How much that matters to you likely depends on your question, but as with any parameter, there's no reason to estimate it if it's not needed.
