My first latent categorical var (LCV) has two categories. The second LCV has three categories, similar to your example on page 182 of your new user's guide. It would be interesting to regress one of the new latent class patterns on covariates. Is that possible? For example pattern 6=c1#2/c2#3. If it is possible, where could I find the code?
bmuthen posted on Friday, July 16, 2004 - 10:23 am
I think you mean that you want to regress a combination class - c1#2, c2#3 - on x's. This would be thinking in terms of a single c variable (with many classes) where this class relative to the last class is regressed on x's. I don't think you can do that with 2 c's, but have to revert to a single c. Still, you can have the benefit of 2-c modeling visavis your outcomes in line with the trick carried out in paper #86, Latent Variable Mixture Modeling, see the example on page 11-12.