Hey guys,

I am conducting a measurement invariance test with categorical data in Mplus. The items are measured on a 5-point-scale. I am working with the script by Svetina et al. (Article Multiple-Group Invariance with Categorical Outcomes Using Up...

Now, when I am trying to do the configural invariance testing, Mplus gives me this error:

"Group 2 does not contain all values of categorical variable" and lists three variables (or one when testing for other invariance aspects).

I use listwise deletion, so it is indeed possible that some of the rarely chosen categories are not included in some groups. However, obviously, this ruins the measurement invariance test, which I need.

Does anyone know how to handle this?

I already tried freeing the number of categories (by using (*) when defining the categorical data), but this does not work with WLSMV-estimation, which I need for the categorical data (I tried switching to MLR, which caused new problems).

Do I need to impute missing data? If yes, how? Or is there any other way?

All ideas are highly appreciated! (I feel a little lost right now)

Thanks!

Saskia