Local identification PreviousNext
Mplus Discussion > Structural Equation Modeling >
 J_F posted on Monday, June 22, 2015 - 7:10 am
I'm dealing with a sem model in which each latent factor has two manifest indicators. The full model is identfied and fits well with the data. However, having only two indicators for each latent construct does not allow to determine the "local" fit of each latent factor. Do you have any advice to deal with this problem?
 Linda K. Muthen posted on Monday, June 22, 2015 - 7:55 am
This is the problem with having only two indicators. If you want to test local fit, you need a minimum of four indicators for each factor.
 J_F posted on Thursday, June 25, 2015 - 1:31 am
Thank you. I wondered whether it is possible to obtain a reasonable global fit if the latent constructs don't fit with the data locally. If not, does a well fitting overall model imply that each single factor's fit is good (or at least not poor), too?
 Linda K. Muthen posted on Thursday, June 25, 2015 - 6:47 am
Good overall fit does not necessarily imply good fit of all model parts.
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