I want to turn off IIV for only some variables. That is, some variables have both fixed- and random-effects, but some variables have only fixed-effect without random-effect.
In fact, I tried to control it in ‘Covariance model’ by putting ‘0’ in the diagonals of these variables without IIV. However, every time I have gotten an error message that it is not positive definite.
Hope my question is clear to all and somebody can give me a solution.
this error appears when estimation have some not invertible matrices in the middle of the algorithm. It can be caused, among other reasons, by under or over parameterization. Of course, it depends on the project itself so, right now, the only way to know for sure the exact reason is to send us a project with the structural model and the data so we can debug it.