Hi, if I remember correctly you can find information on how to calculate them separately for different levels in this book: J.J. Hox (2010). Multilevel analysis. Techniques and applications. 2nd Edition. New York: Routledge. Can´t check it myself because I have lended my copy of the book to a colleague.

Hi, you can also see Ryu and West (2009).

Ryu, E., & West, S. G. (2009). Level-specific evaluation of model fit in multilevel structural equation modeling. Structural Equation Modeling, 16(4), 583-601.

and there is current a discussion on "Multilevel SEM fit indices" on SEMNET (see posts in march 2015 on https://listserv.ua.edu/archives/semnet.html).

Christoph

Thanks John and Kristoff,

I will see the references and I will sign up...

Best regards!

Thiago.

Hi Thiago,

how many predictors do you have?

If you have more than one, it is safer to test them stepwise and also to test them at first as fixed effect and in second step as random effect. If you use R or SAS, you can test each predictor and its kind of effect (fixed or random) separately by chi-square. Actually, that does say more than mere fit indices.

Good luck!

Christin

P. S. in case, your using SAS, the reference is:

Littell R. C., Milliken G. A., Stroup W. W., Wolfinger R. D., & Schabenberger, O. (2006). SAS for mixed models (2nd ed.). Cary, NC: SAS Institute, Inc.

Hi Christin,

I use MPLUS.

Even so, I'll look the reference you sent.

You can always go and look for Barbara Byrne's work. http://www.bookdepository.com/Structural-Equation-Modeling-with-Mplus-Barbara-Byrne/9781848728394