Hi,

I am working on an abstract. I want to build model for continuous outcome variable (not normally distributed), with categorical predictor and several covariates (continues and categorical). Since the residual not pass the normality assumption, I would like to try npregress.

However, I did not find any source of the assumption checking and evaluating goodness of fit information for npregress. How can I know how well is npregress fits the observation, only using R^2?

Hope someone could get suggestions.



Thanks,
Jessica