Hi Drs. Muthen et al., I am running a CFA using MRL with 14 5-level indicators and N>700. I'd like to compare a revised model with 12 of those indicators to the original 14-item model. The MPlus output advises me not to use chi-square to compare models, and the DIFFTEST function does not work with the MRL estimator. Can you suggest a strategy to compare these to models?
I don't see a way you can get a statistical test of different in model fit for that situation; you need to have the same DVs (factor indicators). You could perhaps look at how factor determinacy differs across the models.