MLM vs ML under normality conditions ... PreviousNext
Mplus Discussion > Confirmatory Factor Analysis >
 Stephanie E. A. Mendez posted on Wednesday, March 09, 2016 - 11:28 am
Hi Bengt & Linda,

I was wondering if MLM and ML are supposed to provide equivalent fit statistics in context of normality (e.g., Kurtosis/Skewness less than absolute values of 2 and 7 respectively) when conducting a CFA.

I found that in my CFA model (variables normally distributed) that with the inclusion of the scaling correction, RMSEA and CFI/TLI fit statistics were approx. .1 better for the MLM estimation relative to ML.

I am curious if this has been observed before, and if you have any suggestions related to use of MLM in normality contexts (e.g., if MLM provides better fit statistics, why don't more people use it?)

Thank you!
 Bengt O. Muthen posted on Wednesday, March 09, 2016 - 11:58 am
I don't think there is any theory for MLM testing doing better than ML for normal data. Small differences can occur in any given sample - their properties are asymptotic.
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