I would like to ask for advice regarding the following question: I would like to compute a mixed effects regression with longitudinal data (2 time points) with multiple imputation data. This is the syntax to set the data:

mi set wide
mi register imputed anx_1 anx_2 Neuro_1
mi impute mvn anx_1 anx_2 Neuro_1 , add(10) rseed (53421)

mi reshape anx_ , i(id) j(time)


The question is whether values of anx_ change differently over time and whether this association is independent from a constant variable measured at the first time point (Neuro_1). I specified the model as follows:

mi estimate: xtmixed anx_ ib1.time##ib1.Group || id:

Now a ran the model adjusting for Neuro_1:

mi estimate: xtmixed anx_ ib1.time##ib1.Group Neuro_1 || id:

While the main effects of time and Group differ, the values for the interactions time#Group are completely identical in both models, so it appears that adjusting does not work for this part.

Can anybody give advice were I am wrong?