Dear all,

I am trying to make simple model with Latent construct predicted by Ordinal endogenous observed variables.

a) I believe that I cannot use SEM command with ordinal data. Is this correct? Can I use perhaps dummy variables? Any advice?

b) I can use GSEM, but GSEM cannot handle the missing data. Is there a way to overcome this? In the GSEM manual it says that sometimes missing data option can be used.

Many thanks in advance