What's the best way to decide whether to analyse nested or non-nested CFA models?

by Lisa Murphy   Last Updated February 13, 2018 13:19 PM

I have two measures. One is a global measure of a construct, the other is a domain specific measure of the construct. I'm trying to show that they a different, but related.

So, I've conducted CFA to test a 1-factor model (all items from both scales) and a two factor model (half of the items from the global measure, the other half from the domain specific measure), as well as a 4-factor model (the global measure is sometimes reported to contain 2 factors itself, so I tested a 4 factor model with 2 subscales from the global measure and 2 subscales from the domain specific measure).

I am unsure whether to examine their fit indices visually and let that inform my hypothesis, or to nest the models and compare them that way. Any guidance would be much appreciated!

Related Questions

how to compare nested measurement models

Updated October 10, 2017 15:19 PM

Comparing CFA models (AIC)

Updated May 04, 2017 22:19 PM