Model fit output for continuous/binar... PreviousNext
Mplus Discussion > Structural Equation Modeling >
 John posted on Wednesday, May 04, 2016 - 1:17 pm
Please excuse my ignorance given how new I am to Mplus.

I am aiming to assess model fit output for a model with 5 continuous variables and 1 binary variable (so I am using WLSMV estimation). I have 3 related questions:

1- Is the chi-square value useful when I get this message: "*The chi-square value for MLM, MLMV, MLR, ULSMV, WLSM and WLSMV cannot be used for chi-square difference testing in the regular way. MLM, MLR and WLSM chi-square difference testing is described on the Mplus website. MLMV, WLSMV, and ULSMV difference testing is done using the DIFFTEST option."?

2- Would I only get WRMR output when both continuous AND binary variables are in the model?

3- Relatedly, would I only get SRMR output when only continuous variables are in my model?

Many thanks.
 Bengt O. Muthen posted on Wednesday, May 04, 2016 - 6:35 pm
1. That is a standard message and does not reflect on the fit of your model.

2. We no longer recommend relying on WRMR.

3. The output will tell. The CFI index is usually a reasonable fit measure.
 John posted on Friday, May 06, 2016 - 6:52 am
Thanks for your reply.

What exactly does the "Chi-Square Test of Model Fit for the Baseline Model" tell you?
 Linda K. Muthen posted on Friday, May 06, 2016 - 9:17 am
This is used to compute CFI and TLI. It is a model of free means and variances for all observed variables in the model and covariances among the exogenous and endogenous variables.
 Charlotte posted on Tuesday, May 31, 2016 - 9:49 pm

Following up on Dr. Muthen's response on 5/4/16, "2. We no longer recommend relying on WRMR."

1) Why are you no longer recommending relying on WRMR?
2) My understanding from previous postings was that WRMR was a statistic that was computed if a model included binary items, and that it replaced SRMR. Provided that my understanding has been correct, is there a way for us to compute SRMR in Mplus given that WRMR is no longer deemed reliable?

Thank you!
 Linda K. Muthen posted on Wednesday, June 01, 2016 - 6:10 am
WRMR is an experimental fit statistic. It always has been. It is not a replacement for SRMR. If SRMR is available, it is given automatically.
 Charlotte posted on Friday, June 03, 2016 - 1:46 pm
Thank you!

Following up now on the response to the first post in this thread, I have a question about the following: "1. That is a standard message and does not reflect on the fit of your model."

Does this mean that the values for the chi-square and df in the chi-square test of model fit can be interpreted as is? Or, does a different test need to be run to get an appropriate chi-square test of model fit?

Thank you!
 Bengt O. Muthen posted on Friday, June 03, 2016 - 5:47 pm
The chi-square test can be used as is, but chi-square difference testing when comparing models need a special approach shown on our website.
 Charlotte posted on Tuesday, June 07, 2016 - 10:28 pm
Thank you!
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