I want to examine a SEM with categorical dependant variables using a pooled dataset of approx. 1,500 individuals surveyed over two time points. To adjust for their non-independance, am I missing something by simply computing robust standard errors with the commands TYPE=COMPLEX, CLUSTER = id and ESTIMATOR=MLR?
Thank you for your time, Thierry
P.S. This forum is amazing! Your work is truly appreciated.