09-26-2013 04:52 PM
I wonder how to find the estimated dispersion parameter of the negative binomial distribution when I use Proc Genmod with a repeated statement. The dispersion parameter is provided in the Analysis of the parameters estimate when the repeated statement is not used. However with the repeated statement this information is not provided anymore. Is it because the analysis with the repeated statement is based on a pseudo-likelyhood instead of the maximum likelihood? Is there any way to retrieve the dispersion parameter when using the repeated statement?