Interpretation of Model Fit Statistics PreviousNext
Mplus Discussion > Structural Equation Modeling >
 Alissa Banyasz posted on Monday, June 02, 2014 - 1:41 pm
Dear Drs. Muthen,

While every path in my model except one is significant, it does not meet standard fit criteria, (even after trimming the one insignificant path; i.e., X2 (14)= 187.46; CFI=.84; RMSEA=.169; SRMR=.128). Therefore, I was wondering:

-To what degree can I interpret a model that does not meet standard fit criteria (i.e., is it appropriate to interpret the significant individual paths even if the fit for the overall model is poor)?

-How can a model in which all the paths are significant still have a poor fit?

Thank you so much in advance!
 Linda K. Muthen posted on Monday, June 02, 2014 - 5:32 pm
Significance compares a parameter to zero. Model fit compares the observed and model estimated variance/covariance matrices.

A model with poor fit should not be interpreted.
 Alvin  posted on Tuesday, June 03, 2014 - 10:15 pm
Hi Dr Muthen, I tested one-factor CFA of a subscale of a measure and I repeated this procedure for the other four sub-scales. I realize the n required for multigroup CFA (given the small sample and complex five-factor model) so decided to do it on each subscale. In addition, I tested CFA of each subscale for each group separately. I noticed however in one subscale that the chi-square was 0 with 0 df (with 9 parameters) .. there are 3 items on this subscale. Why?
 Linda K. Muthen posted on Wednesday, June 04, 2014 - 6:13 am
A factor with three indicators is just identified. Fit cannot be assessed.
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