Hello,

I've recently had a paper using SEM models reviewed. In the model I had four latent variables each with >2 indicators. For each indicator I used sub-scales from the same questionnaire (which is what my reviewers have questioned).

I did this because I didn't want to use the total questionnaire score due to issues with model identification and I didn't want to over complicate things by using each individual item. I appreciate that sub-scales from different measures assessing the same construct are usually used, but I wanted to check if it was viable to do what I had done?

Would there be any relevant reading to suggest that this might be a suitable approach?

Cheers.