I used Mplus version 5.21 to estimate a growth curve model. Upon returning to the data later, I updated my version of Mplus (to 7.11), and now the code no longer runs. I am getting non-positive definite errors on identical data an identical models (I'm running more than one). I've tried looking through the version updates to see what might be causing the problem, but there have been enough changes that it is not clear. Any suggestions on where to start to try to eliminate this issue?
The variables that are now problematic are ones that must be included in the model to test the question of interest, and are on the same scale as all the other variables in the model, so removing them is not an option.