Multilevel CFA with categorical variables

There are a number of examples of twin and family data in our documentation and the 'workshops' tab. The major difficulty for your problem is the inclusion of ordinal data: OpenMx currently has a hard limit of 20 ordinal variables, and any 'wide-format' method for 12 variables will require more ordinal variables than we can support. Multilevel SEM is something that we're working on, but unfortunately we're not there yet. The same goes for weighted least-squares optimization, which should be free of that ordinal variable limit. Hopefully we'll be of more help in the future, and the code examples throughout the site can help you specify some version of the model in either OpenMx or another program.