Moderated multiple mediation PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 jill smith posted on Sunday, January 24, 2010 - 11:03 pm
I wish to analyze a model in which an experimentally manipulated, dichotomous IV is linked to a (1=lo,9=hi) Likert DV through three parallel (1=lo,9=hi) Likert Mediators (M1, M2, M3), where the indirect path thru M1 is predicted to be stronger (vs paths thru M2 and M3) for experimentally manipulated, dichotomous moderator condition W=1 and indirect paths thru both M2 and M3 are predicted to be stronger (vs the path thru M1) for moderator condition W=0.
Can Mplus do this and, if so, what syntax would be needed? I wish to determine the relative strength of the indirect paths thru M1, M2, and M3, for moderator conditions W=1 versus W=0. Can the demo version do this?
I would very much appreciate any advice.
 Linda K. Muthen posted on Monday, January 25, 2010 - 10:28 am
Mplus can do this. See the path analysis examples in Chapter 3 to get started. MODEL TEST can be used to obtain the Wald test. Whether the demo works depends on the number of variables you have in the model.
 Daiwon Lee posted on Tuesday, May 04, 2010 - 6:42 pm
Hello!

I have a question regarding moderated mediation.
I need to use XWITH option to used interactions between continuous latent variables which require "type = random" command.
However, I also need to use "model indirect" command to see indirect effect which is now available with "random" command.
So when I use
"ANALYSIS: Type is random complex ;"
I get an error message that
"MODEL INDIRECT is not available for TYPE=RANDOM"
What should I do?

Thank you.
 Linda K. Muthen posted on Wednesday, May 05, 2010 - 9:53 am
What are the variables involved in the indirect effect?
 Ken Sakurai posted on Thursday, August 05, 2010 - 6:45 am
Hello Drs. Muthen

I want to run a moderated mediation analysis. The model is specified as follow: A->B->C, D moderates the link between B->C. All variables are continuous and observed.

One twist is that I have 2 control variables. Can Mplus6 model the model above plus control variables?

If so, I want to use bootstrap tests to 1) obtain indirect effect value given the moderator values of -1SD, Mean, and +1SD and 2) produce conditional indirect effects for a range of a moderator variable.

I was wondering if you could provide with me with an example of Mplus command for a moderated mediation. I prefer to use Mplus6 over Preacher et al's SPSS Macro command.

Thank you
 Linda K. Muthen posted on Thursday, August 05, 2010 - 10:24 am
Yes, you can include additional covariates as control variables.

I do not know how you would obtain the bootstrap tests that you describe.

See Example 5.13.
 Meghan Miller posted on Saturday, June 18, 2011 - 12:05 pm
I have a model in which both A and B->C->D, E, F. The paths between A/B->C and C-> D/E/F are moderated by a dichotomous variable (all other variables are continuous). I also have another model in which variables D, E, F are dichotomous. I am trying to determine how to assess indirect effects and obtain bootstrapped standard errors, as in example 3.16, while including this moderator variable. Any information would be very helpful.

Thank you.
 Linda K. Muthen posted on Sunday, June 19, 2011 - 9:29 am
You should use the WLSMV estimator with MODEL INDIRECT and the BOOTSTRAP option. This can be extended to multiple group analysis.
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: