Exporting SAS Datasets to DDI 3 XML files

The Data Documentation Initiative (DDI) is an international effort to establish an XML-based standard for representing lifecycle information about social science research and similar data. DDI facilitates the automation of documentation and production systems for the delivery of social science data as well as for preservation and exchange. DDI 3 files also have the capacity to contain both metadata, and the data being described. For some SAS users there will be a need to document existing SAS datasets via system-independent DDI files. This paper describes approaches to tools that can extract metadata and or data from an existing SAS dataset and add other metadata such as title, creator, funding agency etc. to create a DDI file. Two working implementations are compared, one uses PROC SQL and DATA steps, and the other uses a user-defined ODS tagset. Other issues are discussed - such as the user interface for including that additional metadata (the who, what, when, where, why, and how); the potential need to document relationships like time series among tables; and the potential need to document referential constraints (foreign keys). While some of the techniques involve advanced features, SAS users of all skill levels may find this paper of interest.