Simulation Study for Cutoffs of Fit Criteria for WLSMV Estimator possible in Lavaan?

[Isa]

Dear Lavaan Group

 

I am doing SEM with mixed type data including categorical (ordinal and binary)variables but also interval variables. Unfortunately, there are no Fit Criteria for Good Model Fit in such a case (so far I used the well-known criteria for continuous data with ML estimator as orientation). According to https://www.statmodel.com/download/Yudissertation.pdf I would need a Simulation Study to evaluate the Criteria for Good Model Fit since I am using WLSMV estimator and mixed type data. Is this possible in Lavaan? If yes, how would this work? If no, what would be other alternatives to just using the “wrong” criteria?

 

Does anyone have experiences in this topic?

 

Best regards,

isabel

[Alex Schoemann]

Hi Isabel,

I'd recommend checking out the simsem package (simsem.org). It is designed to aid in Monte Carlo simulations with lavaan. 

Alex

[Terrence Jorgensen]

I'd recommend checking out the simsem package (simsem.org).

Specifically, vignette 21 should be helpful for finding a reasonable criterion.

https://github.com/simsem/simsem/wiki/Vignette

Vignettes 24-28 show how to do model comparison with Monte Carlo, if that is also something you will need.

Terrence D. Jorgensen

Assistant Professor, Methods and Statistics

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

http://www.uva.nl/profile/t.d.jorgensen