Yes, both models can be built and analysed with PLS-SEM. The small model is a formative measurement model. The big model includes several (i.c. 3) formative measurement models and is a so-called Hierarchical Component Model (HCM) of Type 4 (i.e. Formative-Formative). The Higher Order Component (HOC) however should include indicators too.

Thank you so much for your prompt response.

My problem is that same, in the larger model I wanted the 4th latent variable to be built based on the other three, what you are telling me is that this is not possible, is it?

And a model like the one presented attached will be feasible? If it is classified in which category?

Hi Ana,

The larger model is possible, provided you have the scores for the fourth variable, the one which does not have any manifest variables yet. In your diagram you have to the left three latent variables build in a formative manner, which account for the measurement (outer) model in your PLS SEM, and a fourth variable, to the right, which does not derive from anything. If you already have the scores for this variable (namely, if this variable was already measured), you can fit the model easily.

Also the model presented in the last document you attached is possible: it is, in fact, the measurement model from the larger model you presented before.

I hope this is of help.

Elena