I am trying to run a growth model of how fast food intake increases across life events in early adulthood, e.g. leaving parental home, having a child. I have been reading the paper of Curran, Muthen, Harford 1997 http://www.statmodel.com/bmuthen/articles/Article_084.pdf Is this still how you would advise to set up this kind of model? Should the latent variables have covariances? Thank you.
Thank you. I was also interested in whether the additional latent variables for each time point should covary with i and s. In the 2nd diagram in the paper, there are no covariences shown, including between i and s and between these and the additional latent variables.