Mai Sherif posted on Wednesday, November 12, 2014 - 6:23 am
I am running a SEM where I have 4 binary indicators measuring an underlying continuous latent variable at each of 5 waves. The same indicators are used to measure the latent variable at each wave, where measurement invariance is applied. The time-dependent latent variables are connected via an AR1 structure.
I used both WLSMV and Bayes estimators to run this model but I found that there are discrepancies in the results. I wonder what is causing these discrepancies?