MCFM and structural model specification PreviousNext
Mplus Discussion > Confirmatory Factor Analysis >
 Patchara Popaitoon posted on Sunday, August 21, 2011 - 2:15 am
Dear Linda,

I have tested a leadership contruct applying MCFM and found the construct is statistically well valid at the group level of analysis. (I followed the processes described in Dyer et al.'s (2005) paper).

I would like to know how to specify this construct in the structural model, which is aimed at identifying the relationship at the group level.

My model is quite simple:

Group's Perceptions of Leader --> Group's Behaviours --> Group's Financial Performance

Do I need to use the whole specification of MCFM for leadership construct in the model? Or given the validity of the construct I got from MCFM test, I am allowed to aggregate the leadership construct to the unit level? Please advsie a more appropriate method.

Many thanks.
 Linda K. Muthen posted on Monday, August 22, 2011 - 3:10 pm
Which level are your variables measured on? It sounds like they are measured on level 2. If this is the case, I don't know what you mean by aggregating to level 1.
 Patchara Popaitoon posted on Tuesday, August 23, 2011 - 5:13 am
Dear Linda,

You are right, this is an analysis at the unit level (level 2).

I think my question is how to specify the leadership latent construct at the unit level.

1) I should specify the measurement model of leadership construct as a multilevel CFM in the structural model.

2) I should aggregate these observed variables, which are measured at the individual level (level 1), to the unit level. Then, I specify the leadership latent variable based on these aggregated scores.

If 2) is the case, am I correct that the results from MCFM for the leadership construct is used only to justify the validity of the construct. That is, although the leadership variables are measured at the individual level, they can be applied as the variables at unit level.

 Patchara Popaitoon posted on Tuesday, August 23, 2011 - 6:57 am
Dear Linda,

I guess I know what you meant in the previous correspondence. Thanks.

 Bengt O. Muthen posted on Tuesday, August 23, 2011 - 8:18 am
I would go with 1) in your message above.
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