Hello Dr. Muthen, I ran into a question when reading your 2006 paper, Item response mixture modeling: Application to tobacco dependence criteria. In Figure 6, item probability derived from a factor mixture model (IRT mixture model) is plotted against Tobacco Dependence Criterion. I am wondering if this is conditional item probability or not. If so, is there a way to make Mplus plot this probability rather than the default, estimated probability? I couldn't find a way to obtain conditional item probability directly from Mplus and am wondering if additional calculation has to be made using other softwares or not.
What is plotted in Figure 6 is the item probability conditional on class but integrated over the factor. So it is marginal in the sense of not being conditioned on the factor, but conditional with respect to the class. For these plots, Mplus always conditions on the class. Because the integration is done via numerical integration it would be hard to do by hand.