Mediation over time PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 Donna posted on Thursday, February 07, 2013 - 3:37 am
Hi Linda and Bengt,

Using SEM, I would like to build a mediation model where the mediation variable represents change over time across two time points

the model has:

x measured at time 1
m1 measured at time 2
m2 measured at time 3
y measured at time 4

Can I build the m difference as (m2 - m1) or should I use latent curve analysis?

Is there an example in the manual that covers a similar model?

Grateful for your advice.
 Linda K. Muthen posted on Thursday, February 07, 2013 - 9:49 am
I think you should use one of the following models not a difference or latent curve analysis:

x -> m1 -> m2 -> y

or


x -> m1 -> y
x -> m2 -> y
 Donna posted on Thursday, February 07, 2013 - 2:26 pm
Thank you, is there an example in the manual on how to build the top model ?

Thanks again
 Linda K. Muthen posted on Thursday, February 07, 2013 - 2:30 pm
It would be

y ON m2;
m2 ON m1;
m1 ON x;
 Donna posted on Friday, February 08, 2013 - 8:37 am
Thanks Linda
 Donna posted on Monday, February 11, 2013 - 7:20 am
Hi Linda,

one more question. I would now like to see if my mediation model (with 1 mediator) is moderated by a 4th variable, which was measured at the same time as the mediator...

How do I include this moderator in the mediation model ?

Thank you
 Linda K. Muthen posted on Monday, February 11, 2013 - 10:00 am
For categorical moderator, you can use multiple group analysis or include and interaction term. For a continuous modeator, you would include an interaction term, for example,

y ON m z mz;
m ON x;
 Donna posted on Tuesday, February 12, 2013 - 2:09 am
Thanks again for such quick and helpful reply.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: