Fit indices non-normal longitudinal data PreviousNext
Mplus Discussion > Growth Modeling of Longitudinal Data >
Message/Author
 Ling Ning posted on Tuesday, April 04, 2017 - 10:16 am
Dear Prof. Muthen,


I am estimating a piecewise growth curve model and try to use fit indices to identify model misfit but the literature suggested that I can't use the CFI and TLI directly from Mplus output because the default null models used in Mplus is not adequate (Hancock, G. R., & Mueller, R. O. , 2010). Instead I need to estimate an adequate null model and then used the values to calculate CFI and TLI. But the question is my data is not normal and I used estimator = MLR, in which case, can I directly used the chi-square value generated from the MLR estimator and plug them into the formula to calculate CFI and TLI? I understand it is ok to use SB chi-square from MLM estimator to do the calculation but not sure if it is the same with the MLR estimator.

Thanks.

Ling
 Ling Ning posted on Tuesday, April 04, 2017 - 1:35 pm
The default null models used in Mplus is not adequate for any growth curve model(Hancock, G. R., & Mueller, R. O. , 2010, pp.196)

Ling
 Bengt O. Muthen posted on Tuesday, April 04, 2017 - 6:59 pm
Please scan the page 196 you refer to and email to me so I can see where you are coming from.
 Bengt O. Muthen posted on Wednesday, April 05, 2017 - 11:53 am
Yes, you can use the MLR chi-2 from any null model you like and plug it into the CFI formula. You can check that this is what we do for MLR by setting up a regular null model.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: