|
Message/Author |
|
|
Dear Drs Muthen, I am running a parallel process latent growth model and want to test the effects of continuous (time-invariant) moderators. I have seen in some papers people conducting multiple-group analysis using continuous moderators as dichotomous with median splits. Is this more advantageous compared to using XWITH command? Also when I use XWITH command, how do I compare the fit of models after adding the interaction, as I don't get chi-square in TYPE=RANDOM? Should I use AIC? |
|
|
XWITH can be used when you have at least one latent variable involved in the interaction. It may be better so as to not dichotomize the moderator and lose information. With XWITH you can compare models by loglikelihood differences, providing LRT chi-square. BIC and AIC are also useful. See also the FAQ on our website: Latent variable interactions. |
|
Back to top |
|
|