Hi, I've estimated a latent profile analysis in Stata 17.0 with three continuous dependent variables and four latent classes. I've also used gsem to run a multinomial logit predicting the classes, with two predictors (control1 and 2). The code structure is:

Code:
gsem (dv1 dv2 dv3 <-,)(C <- control1 control2), lclass(C 4) vce(cluster pid)
I would like to change the base category for the multinomial logit, but the usual factor variable notation doesn't work. I'm able to run the following, but there's no different in output, ie. base category is still class 1:

Code:
gsem (dv1 dv2 dv3 <-,)(b4.C <- control1 control2), lclass(C 4) vce(cluster pid)
Wonder if anyone can help? Thank you!