On the use of structural equation models in experimental designs: Two extensions