01-24-2013 10:36 AM
I would like to do a multilevel regression analysis using a Bayesian approach. Could any body clarify what the difference/pros/cons in using GENMOD or MCMC? Does GLIMMIX have a BAYES option as GENMOD? thank you!
01-24-2013 01:44 PM
When you say multilevel, do you mean hierarchically nested data only or you also need to model on longitudinal data?
There is a SAS training course on the subject "Multilevel modeling of hierarchical and longitudinal data using SAS". There should be a more cost-effective web class avaiable.
01-25-2013 03:38 AM
Hello Jason, thank you for your message. I am not interested in longitudinal analysis, at the moment, I simply have 2 levels of nested data (Hospital and patients within). If you have a link for the mentioned web class, I would appreciate it.
Thank you and regards.
01-25-2013 09:22 AM
GLIMMIX does not have the same sort of Bayesian analysis as found in GENMOD. Take a look at Example 55.8 Nonlinear Poisson Regression Multilevel Random-Effects Model in the PROC MCMC documentation. That should provide an excellent starting point (well, assuming you have a background in MCMC methods).