Multilevel mediation model and grouping PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
 tiksu11 posted on Tuesday, July 02, 2013 - 3:59 am

I have a mediation model that has as an independent variable the implementation of intervention divided into 3 groups (control, low level and high level of implementation). The aim is to see whether these different groups have different influence on mediators and dependent variable.
I have used grouping command to check differences between these groups (for example checking intraclass correlations), but I'm not able to use grouping command in my overall multilevel mediation modeling. I tried to search information if I can use grouping command in this case, but did not find. Should I use another command? Or how is it possible to do with Mplus? Thank you for your help.
 Linda K. Muthen posted on Tuesday, July 02, 2013 - 12:04 pm
I think you need to use the KNOWNCLASS option and TYPE=MIXTURE in this case. When all classes are known, it is the same as multiple group analysis. See Example 7.21 in the user's guide.
 tiksu11 posted on Tuesday, July 09, 2013 - 2:30 am

thanks for your reply. I have tried to do the analysis with knownclass option and type=mixture, but I don't know how to add the measuring of total, indirect and direct effects in the model.
Using the model indirect, I cannot use X variable of my model, because I have determined it in the knownclass option. (the X is the divided into three categories and I defined it as a knownclass variable).
Should I use model constraint? Is there a good example how to conduct mediation with model constraint? I tried to search it, but did not have any luck with it. Thanks once again!
 Linda K. Muthen posted on Tuesday, July 09, 2013 - 8:40 am
If you want to create indirect effects that include the grouping variable, you need to include it in the analysis and not use it as a grouping variable. For three categories, create two dummy variables using the DEFINE command.

Following is an example of how to create a indirect effect that is a product using MODEL CONSTRAINT:

y ON m (p1);
m ON x (p2);

NEW (ind);
ind = p1*p2;
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