LMS moderated correlation PreviousNext
Mplus Discussion > Structural Equation Modeling >
 Johanna Hartung posted on Tuesday, May 10, 2016 - 2:56 am
Hello everyone!
I am doing an approach similar to latent moderated Structural Equation modeling (LMS; Klein & Moosbrugger, 2000). The model consists of two latent correlated variables. There are two continuous moderators (both z-standardized). The loadings are moderated (using XWITH of the latent variable and the moderator). Now I want the correlation/covariance between the two latent variables to be moderated as well.
f1 WITH f2;
f1m | m XWITH f1;
f1e | e XWITH f1;
f2 ON f1m f1e;

My problem is that the moderation coefficients depend on which factor interacts with the moderators. Theoretically this should not be the case. And does not happen if only one of the moderators is used.
Do you have any ideas why this is the case? Is this type of analysis possible with XWITH?
Thank you in advance!
 Bengt O. Muthen posted on Friday, May 13, 2016 - 1:55 pm
I don't know how to get a correlation to be moderated by a continuous moderator unless you use the Constraint= option and Model Constraint as in UG 5.23.
 Johanna Hartung posted on Monday, June 06, 2016 - 5:09 am
Thank you very much for your response!
Unfortunately, the Constraint option cannot be used in conjunction with TYPE = RANDOM, which is used with XWITH, according to the feedback from Mplus.
Is there another possibility to use both simultaneously?
I only found the statement in UG p.611, that XWITH needs TYPE = RANDOM.
 Bengt O. Muthen posted on Monday, June 06, 2016 - 5:46 am
No, you can't use both at the same time.
Back to top
Add Your Message Here
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Options: Enable HTML code in message
Automatically activate URLs in message