Tim Huijts posted on Friday, August 26, 2016 - 10:17 am
I would like to estimate a multigroup model (or multilevel model) with the same latent concept for each group (or higher order unit), but with different observed variables for each group.
For example, suppose that my latent concept would be 'Pain complaints', and my grouping variable would be 'Gender'. Some observed variables of the latent concept would apply to both men and women (e.g. 'pain in joints'). However, some observed variables would be specific for one of the groups only, and not measured for the other (e.g. 'menstrual pains', only measured for women and not for men).
My questions are: (1) is it possible to fit a model like this in MPlus, and (2) would it be possible to separate the part of the concept that applies to all groups (e.g. pain in joints) in the model from the part of the concept that only applies to one group (e.g. by fixing pathways between latent and observed variables in the model)?