This is the preferred way of representing hierarchical data. On initialization, the Source object should create a Table object called HDATA and add it to its Tables collection. The Column objects in HDATA should match the top-level VariableInstance objects in the MDM document, if one is supplied. Otherwise, Column objects should represent the top-level variables in the underlying data source.

For each other level in the data source, the CDSC should create another Table (added to the Source object's Tables collection) with Column objects for the variables at that level. The Table objects' Parent and Children properties should reflect the relationships among the different levels.