You did not state a direct question, but selecting a proper multivariate method depends on variable scale (continuous, ordinal, nominal) and purpose (prediction, association, data reduction, etc). For instance, if one have a set of data measured in different scales (dependent variable) and wants to predict an independent variable (score?) measured in a rank scale, multivariate ordinal regression would apply.

More on this example: http://epub.wu.ac.at/5389/1/Report132_lvana.pdf