The .mdd file does not have a DTD because the Data Model does not support direct access to the XML in an .mdd file. Access to the .mdd file should always be managed by the MDM Object Model.

However, when a connection is made to the case data with an .mdd file attached, the .mdd file defines the schema for the case data. When the connection is made without an .mdd file attached, the schema is taken from the actual case data.

A good way to illustrate this is using DM Query. When you open case data in DM Query, you can use the Columns command on the DM Query Schema menu to display the case data schema.

For example, supposing you have some data that you have collected using one of the UNICOM Intelligence products, like Scan or Interviewer Server. You can use DTS to export the data to an SPSS Statistics .sav file. You then have the choice of opening the .sav file in DM Query using the metadata defined in the project's .mdd file, using the MDSC component of the SPSS Statistics SAV DSC to read the metadata directly from the .sav file, or not using metadata at all.

When you connect without using the .mdd file, the column names correspond to the eight-character variable names used in the .sav file. When you connect with an .mdd file, the column names correspond to the longer names used for the VariableInstances in the .mdd file.