I read an article "Investigating multilevel mediation with fully or partially nested data" saying that:
"a more general limitation of MLM is the inability to model outcomes above the lowest level of the data hierarchy (Krull & MacKinnon, 2001). That is, the effects of individual-level variables on group level variables, such as in 1-1-2 or 1-2-2 designs, cannot be assessed using MLM. Limitations of the previous methods can be overcome by extending MLM to include aspects of structural equation modeling (SEM) to yield multilevel structural equation modeling (MSEM)"
However, as my understanding and reading of your discussions, this is NO longer a limitation in Mplus. We can test 1-1-2 mediation model, without MSEM. Am I right?