Hi everyone,
I would like to know whether both cultural groups interpreted my questionnaire in a similar way. To do so, I conducted a measurement invariance analysis on the whole model of latent variables. However, I have seen that some authors report MI for their constructs separately and they do not report configural, metric, scalar and strict invariance for the whole model. What is the rule here?
Thank you.
Robin
Hello Robin,
in most cases, tests for measurement invariance are conducted in a CFA model but there is no reason not to do it in the full SEM--to the contrary, as you can add the tests for structural invariance to the list of tests.
All the best,
--Holger
Hi Holger,
Thank you very much for your answer. I guess I will report both methods in my papers; tests for measurement invariance on each construct separately (i.e., CFA) and on the full SEM model including predictions between variables.
Best,
Robin
Hi Robin,
when I said that you can do it in an CFA, I did not mean to do it separately for each factor. You can do that of course but the test of the baseline model is rather weak and thus uninformative regarding discriminant validity. Further, if the baseline model is misspecified and you don't catch that by testing them separately, you will get problems when going to the integrated SEM.
Best,
--Holger
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Structural Equation Modeling