

Moderated mediation with 2 DVs 

Message/Author 

YK Leung posted on Monday, March 12, 2018  6:03 am



Hi there, I am running a moderated mediation model with 2 DVs. Specifically, the moderator(z) should moderate the path between the IV (x) and mediator (m). I would like to know... 1. if the script with the "MOD" below is correct Variable: names are x z m y1 y2 xz; Usevariables are x z m y1 y2 xz; MISSING = ALL (999); Define: xz = x*z; Analysis: Estimator = ML; Bootstrap = 5000; Model: y1 ON x m; y2 ON x m; m ON x z xz; Model Indirect: y1 MOD m z(5,6,1) xz x; Output: Cinterval(Bootstrap); 2. if I could test both DVs in the same model (since indicating 2 MODs is not allowed in Mplus) Thank you very much! Best, YKLeung 


1. Yes, this is the expression in the middle of page 765 of the V8 UG. It assumes a binary X. 2. Not possible using MOD. But you can do it if you use Model Constraint to express the effects instead of getting them automatically using MOD. 

YK Leung posted on Wednesday, March 14, 2018  4:19 am



Thanks a lot for your reply Prof. Muthen. I have some followup questions. 1. Does it mean the MOD command cant be used with continuous X? 2. Could you suggest some materials for moderated mediation using model constraint in Mplus? Many thanks Yk Leung 


1. You can use MOD with a continuous X. Just add 2 numbers in parenthesis for X, e.g. X(1 1) which would mean that the effect is considered for a change in X from 1 to 1. This would be relevant, for instance, if X has mean 0 and sd 1. 2. See e.g. chapter 8 of our book Regression and Mediation Analysis using Mplus, as well as my paper on our website: Muthén, B. (2011). Applications of causally defined direct and indirect effects in mediation analysis using SEM in Mplus. Click here to download the paper. Click here to view the Technical appendix that goes with this paper and click here for the Mplus input appendix. Click here to view Mplus inputs, data, and outputs used in this paper. Note, however, that having 2 DVs analyzed together or separate gives the same results unless you have specific model restrictions. 

Back to top 

