I am currently performing a confirmatory factor analysis in Mplus. I wonder, Is it posssible to add correlated errors between items (e.g. item 1 to item 3) when we treated likert type items as categorical? I did not see any article like this. If not why?, Is it possible but not preferable?
Note: I used Mplus wlsmv estimation and treated items as categorical.
The treatment of residuals with categorical indicators is tricky in Mplus.
I think you should ask your question at www.statmodel.com.
Also look at the answers to related questions here:
http://www.statmodel.com/discussion/messages/23/6984.html?1464881709
Also look at this discussion thread:
http://www.statmodel.com/discussion/messages/11/475.html?1478132030
Dear Ivanouw,
Thanks for your informative answer. Now, I am checking the discussions in the Forum.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Structural Equation Modeling