Chferen posted on Wednesday, January 15, 2014 - 6:20 pm
Thank you, Dr. Muthern!
Yes, I read the file you sent to me. But it looks in the growth curve model, the covariance of the error term is a diagonal matrix rather than AR1.
Do you have the details about how to estimate the correlation in a Autoregressive Latent Trajectory models? Can Mplus handle that?
It looks there is no closed estimation formula for it. Under ML framework and using EM algorithm, how can I estimate the correlation by using fisher score method or some other optimal methods in the SEM with a latent variable as dependent variable?