Multilevel CFA? PreviousNext
Mplus Discussion > Confirmatory Factor Analysis >
 Shira Mor posted on Wednesday, December 26, 2012 - 4:07 pm

I am interested in running a multilevel CFA, but not very familiar with Mplus but HLM.

I have a two level dataset where peer raters (level 1 variables) are nested within target students (level 2). The datasets are two separate datasets that contain the target student ID as the matching variable. Can you please let me know whether I need to transform these two datasets to run a multilevel CFA in Mplus?

Many thanks and happy holidays,
 Linda K. Muthen posted on Wednesday, December 26, 2012 - 4:11 pm
All data must be in one data set. Cluster level variables take on the same value for all cluster members. See Example 9.7 in he user's guide and the data for that example.
 Shira Mor posted on Wednesday, December 26, 2012 - 4:53 pm
Ok, thanks. I am still not clear what to do if I want to have cluster level continuous variables (student level self-reports)as part of the CFA with level 1 variables (peer ratings of students on cooperation).
 Linda K. Muthen posted on Wednesday, December 26, 2012 - 8:50 pm
If you have a cluster level continuous variable, each member of the cluster will get the same value for that variable, so if it is y, the data will be:

cluster y

1 3
1 3
1 3
1 3
2 5
2 5
2 5
2 5
 Hassan posted on Friday, May 13, 2016 - 11:26 am
Dear Dr. Muthen,

I'm running a multilevel CFA to check the validity of my scale. The scale has three factors and each factor has 6 items.

When I wanted to check the L2 (or between) part of my model, I aggregated each item for each class. Therefore, my model sounds like this:


MSTw BY SMst1 SMst2 SMst3 SMst4 SMst5 SMst6; !latent variable Mastery (within)

PERAPPw BY SPerApp1 SPerApp2 SPerApp3 SPerApp4 SPerApp5 SPerApp6; !latent variable performance approach goals (within)

PERAVOw BY SPerAvo1 SPerAvo2 SPerAvo3 SPerAvo4 SPerAvo5 SPerAvo6; !latent variable performance avoidance goals (within)


MSTb BY Mst1m Mst2m Mst3m Mst4m Mst5m Mst6m; !latent variable Mastery (between)

PERAPPb BY PerApp1m PerApp2m PerApp3m PerApp4m PerApp5m PerApp6m; !latent variable performance approach goals (between)

PERAVOb BY PerAvo1m PerAvo2m PerAvo3m PerAvo4m PerAvo5m PerAvo6m; !latent variable performance avoidance goals (between)

I wonder my approach is correct?

 Bengt O. Muthen posted on Friday, May 13, 2016 - 1:45 pm
You don't have to aggregate into mean scores. You can instead use the Mplus latent variable decomposition - see UG examples of 2-level FA.
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