Use of an observed (measured) variable in a structural equation model?


Is it possible to include an observed (measured) variable in a structural equation model (SEM)?

In different words: Must one only use latent variables in SEM?

I have several latent variables, but one variable does not have multiple measures. Nevertheless, if it is technically possible, I would like to include it in the model. As simple as this question might be, I have had a difficult time locating a precise answer.

Thank you for any help that you can provide.

Best wishes,


Phineas Packard
Yes you can have manifest varaibles. This is not an issue provided you make it clear that it was manifest when you write up your results and let the read what implications this has for bias.

If you really wanted you could create a single item latent variable were you constrain the loading and residual term to some sensible values.
Thank you very much for the reply. I have a bit more confidence going forward now, and will look into the second comment.