I have a fairly complicated mediation model that I would like to test, and I am wondering if it is practicially possible to test all parameters simultaneously in MPlus. I have run a preliminary LCA and determined that the 3-class solution was the best-fitting. Herein lies my issue, this nominal variable will be the mediator in an SEM model. This variable will be predicted by multiple covariates and then predict my distal outcomes. Is it possible to test these relationships in the model given the mediator is general categorical? Related to this, can the 3-STEP APPROACH be used as well?