Jan Keil posted on Thursday, February 05, 2015 - 8:25 am
Dear Dr. Muthen, I have gathered behavioral data of one underlying concept within three different conditions (lets call them "c", "u" and "b"). Now I want to investigate the mean-differences from "c" to "u" to "b" and I was thinking, that I could actually treat it as a "quasi-longitudinal"-design and apply a "latent-change-score" model. But since measurment-invariance is a neccessary condition to perform this analysis, and since I don't find invariance in the data (not surprisingly, since conditions fairly differ) I was wondering, if I needed to apply a different model. Or do you have other suggestions how to deal with that issue?