10-11-2014 10:14 PM
Hello all. I am working on a model that uses a wide variety of categorical predictors vs. a continuous dependent variable. Unfortunately, my usable n ends up being 28. I am having difficulty getting significance with my modeling, and I was wondering if there were some model selection techniques that were useful for this situation. I am assuming a gamma distribution for my Y with a log link function. A number of my categorical variables have 5 or more levels, so I know I can't include to many of them. Any help would be appreciated!