sem( ) with observed binary outcome interpretation

[Justine Maltais-Proulx]

Hello all,

I am a novice in SEM and I am trying to build a model in a similar fashion to this simplified diagram

with the "Q13.binaire" being my "main" dependant variable, if you will. However, in my data base, this variable is an observed binary outcome. I am using the ordered argument (ordered = "Q13.binaire") in my syntax.

I have made a lot of research, but am still unsure how to proceed with SEM in this regard. Can a binary outcome variable be employed with CB-SEM at all, or is PLS-SEM the only way forward?

If I can indeed use lavaan and CB-SEM for this model, how do I interpret the results? Can the parameter estimates be used as is or is there an additional operation that needs to be made to use those numbers? I would love to read up on how to proceed in this matter, but don't have any luck with finding literature on CB-SEM with a binary outcome. I would greatly appreciate any references that could be shared with me.

Thank you very much!