lavTestLRT using Yuan-Bentler correction

[Amélie Godefroidt]

Dear, 

I am fitting a SEM model using the MLR estimator in Lavaan. Hence, I always report the Robust fit indices in my paper. However, the standard anova or LavTestLRT function still uses the ML estimated Chi-square values. Why? And how can I change this?

Thanks in advance! Amélie 

[Terrence Jorgensen]

the standard anova or LavTestLRT function still uses the ML estimated Chi-square values. Why? And how can I change this?

That is what it is supposed to do.  A scaled chi-squared-difference test is not the difference between two robust chi-squared values.  Like the scaling of the individual models' chi-squareds, the (naïve) chi-squared difference needs to be scaled.  You can see some details here:

https://www.statmodel.com/chidiff.shtml

Terrence D. Jorgensen

Postdoctoral Researcher, Methods and Statistics

Research Institute for Child Development and Education, the University of Amsterdam

UvA web page: http://www.uva.nl/profile/t.d.jorgensen