How to get the "goodness of fit" measure/s between two/or multiple models of mediation?

Buster

Hi everyone!

I'm conducting a mediation analysis (with 3 predictor, 1 mediator and 2 outcome variables), and I'm interested in getting the measure/s of "the goodness of fit" so I could compare models.

My professor mentioned that I could compare them by checking the "chi square for nested models" (If I understood him correctly), but I can't seem find that option in mediation analysis. Since I'm quite inexperienced, I'm interested about:

1. How can I obtain the relevant measures?
2. Which measures should be included and consulted as "the goodness of fit" parameters in that analysis/procedure, and;
3. How do I report them?

Any information and suggestion would be highly appreciated.

Thanks in advance!

EJ

I've asked around, sorry about the delay.

Buster

Thank you so much EJ!

EJ

Thanks for the reminder, I've sent another email. We will shortly be hiring new people with expertise in SEM, so then we should be able to respond to these issues faster.

Cheers,

E.J.

evankesteren

In order to get these metrics, you need to copy the lavaan code (check the "lavaan syntax" under options) to the structural equation modeling analysis. There, you can get all the fit metrics you need. In this blog there's more information about the SEM analysis:

https://jasp-stats.org/2018/07/03/how-to-perform-structural-equation-modeling-in-jasp/

Here's a guideline I found on what to report https://cscu.cornell.edu/news/Handouts/SEM_fit.pdf

Buster

Thanks Evankesteren, this is very helpful!