mixed censored model

[gperry]

Hi. I'm looking to use a model with a random classification term and another fixed or random classification effect, using a model that natively or descriptively censors at zero (the distributional field is left-skewed). QLIM handles censoring and classification effects but not random effects, while NLMIXED handles censored models with random effects but not class variables. Does anyone have a rough idea or model for a system that covers censored dependent variables with random classification effects?

[SteveDenham]

NLMIXED will handle class variables, you just might have to preprocess your dataset with PROC GLMMOD to get all of the dummy variables needed to fit your model.

Steve Denham