Model fit indices in Multilevel SEM PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 Tony Reed posted on Monday, August 01, 2011 - 4:08 am
Dear Dr. Muthen,

I am still learning multilevel CFA and multilevel SEM. Yet there is one thing I read from papers that confused me with what the Mplus manuals presented.

Specifically, I read about papers applying multilevel SEM, in which they reported CFI, TLI, RMSEA, and SRMR. But, in the Mplus manual sample codes and results (e.g., 9.10 or 9.8), none of these were reported.... So, is it possible to request such model fit indices through multilevel SEM or not?

Also, I saw researchers report RMSEA/SRMR while specifying it's for between or within level. Is that correct?

Thanks!
 Linda K. Muthen posted on Monday, August 01, 2011 - 10:39 am
You can obtain chi-square and related fit statistics with multilevel models if you use the weighted least squares estimator. These are given for both levels.

When TYPE=RANDOM is used, only maximum likelihood estimation is available. Chi-square and related fit statistics are not available in this cases.
 Tony Reed posted on Monday, August 01, 2011 - 10:33 pm
thank you for your response!! it makes sense!!
 Jinxin ZHU posted on Friday, February 05, 2016 - 2:21 am
Dear Prof. Muthen,

Do Mplus report confidential interval of RMSEA in multilevel analysis?

If not, would you please attach a reference so that I can cite in my manuscript?

Thank you very much.
 Bengt O. Muthen posted on Friday, February 05, 2016 - 5:56 pm
I don't think we do - check the output.
 Jinxin ZHU posted on Friday, February 05, 2016 - 6:05 pm
Dear Prof. Muthen,

Thank you so much for your prompt reply.

The confidence interval of RMSEA was not reported in the output.

Is there any technical issue or will it be added in later version?

Thank you very much!
 Bengt O. Muthen posted on Friday, February 05, 2016 - 6:06 pm
I am not sure at this point.
 Jinxin ZHU posted on Friday, February 05, 2016 - 6:12 pm
Dear Prof. Muthen,

Thank you so much for your prompt reply.
 Ben  posted on Saturday, September 24, 2016 - 7:22 am
Dear Prof. Muthen,
Thanks for the great product and service. I am new to Mplus and bought it for one purpose, to run Multilevel (2 level) SEM. Most of my IVs are latent variables (categorical) and I have two interactions between latent variables also as IVs. I have 4 DVs that are all categorical. I am using MLR as estimator.I run the program with no issue except RMSEA is nowhere to be found. Back in 1999, you mentioned that it would be added for multilevel modeling. I have the latest version. Is there anyway to obtain RMSEA. Thanks in advance.
 Linda K. Muthen posted on Saturday, September 24, 2016 - 2:16 pm
If you have categorical factor indicators and are using maximum likelihood estimation, chi-square and related fit statistics are not available. These are available only when means, variances, and covariances are sufficient statistics for model estimation.
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: