Compare non-latent variables across groups

#1
Hi everyone,

Is there a way to compare non-latent variables across groups? Can measurement invariance be applied to non-latent variables?

Thank you so much,
Orchid
 

Lazar

Phineas Packard
#2
Invariance is a test of the similarity of parameters between 2 or more groups. So the question is what parameters are you interested in?
 
#3
Invariance is a test of the similarity of parameters between 2 or more groups. So the question is what parameters are you interested in?
Hi Lazar,

Thank you so much for your response. I have been looking into metric, scalar, and strict invariance. However, the main issue is that my variables are measured, not latent (as most examples of MI use). I have tried using all four variables as manifest variables to create a latent variable. I obtained metric invariance, but not scalar or strong. If I use two latent variables (each made up of two of the variables), the model does not converge. Also, I am not sure how to ascertain whether the relations between variables are also measurement invariant.

Thank you,
Orchid
 

Lazar

Phineas Packard
#4
So your aim should be to apply your underlying theory of measurement to your data. Creating a latent variable only makes sense if your theory implies that your four variables of interest are indicators of a latent factor. So I think the best thing for you to do is provide more detail on what you are trying to measure, what you theory is, etc. Invariance is not always appropriate and may not be here.