01-09-2014 10:07 AM
Dear forum members,
I would like to know, whether there is a way to fit multivariate beta-binomial regression models in sas.
This question arose while attempting to evalaute rehabilitation programmes targeting workers with an elevated risk of unemployment.
We would like to model months of employment in a given year following rehabilitative intervention. However, the variable months of employment shows a bimodal u-shaped distribution with most of the insurees being in a job either for 12 or 0 months during a given year. As a result, even carefully adjusted OLS models feature more or less non-normal distributed residuals (see below). I could only find papers dealing with attempts to model similar distributed data, which illustrated the application of multivariate beta-binomial regression for stata (http://onlinelibrary.wiley.com/doi/10.1111/1475-6773.12055/abstract;jsessionid=89EFC73879C5478B54AB7...) and some software named Egret for Windows (http://onlinelibrary.wiley.com/doi/10.1002/sim.2612/abstract). Does anyone know a way to compute multivariate beta-binomial regression using SAS?
Any answer would be eminently helpful. Thanx in advance.