Dear R Experts, I have a query concerning SEMs in a multilevel context. I have an unbalanced panel where children are nested in families, which in turn are nested in districts. The problem is the following: the outcome variable of interest is measured at the child level but the (explanatory) latent variable is at the family level. Most the variables are discrete and I am able to fit the model using the sem() and hetcor() packages ignoring the hierarchical structure of the data. But, how do I incorporate the fact that the outcome variable is at the child level and the latent variable at the family level? Any help would be greatly appreciated. Thanks a lot already Henry -- http://portal.gmx.net/de/go/dsl02