Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | "Marsh, Kimberly A" <k.marsh07@imperial.ac.uk> |
To | "'statalist@hsphsun2.harvard.edu'" <statalist@hsphsun2.harvard.edu> |
Subject | st: prevalence rate ratios and the robust poisson glm option - model fitting |
Date | Mon, 28 Mar 2011 09:28:36 +0100 |
Hi - I have a cross-sectional survey with an dependent variable (1-disease 0-no disease) and many socio-demographic and risk variables. The model does not converge using a log-binomial glm in stata. In the robust poisson glm model, it converges without a problem. My question is: how do I know if the model is a good fit for the data? I have seen fitmacro, but the information on interpreting this is skant. I'm not sure if I can also look at whether the deviance divided by the DF is less than 1. The literature that I've read thus far assumes that the model converges in log-binomial and poisson, so they mostly compare the bias or differences in the adjusted PRRs between the two methods. Thank you. Kimberly Marsh * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/