DDI Lifecycle: Moving Forward

Motivation

DDI (Data Documentation Initiative) began as an XML archival codebook format and then branched off to cover the research data life cycle. In the process of becoming a metadata standard for the social, behavioral, and economic sciences, the specification has continued to add new coverage and functionality to respond to new user requirements. DDI is now growing beyond social science into the official statistics and medical research communities.

As data take on increasing importance in research and policy-making and as new data types are explored and combined, expectations for metadata rise as well. The DDI has an opportunity to respond to community expectations by creating a new version of the specification that can transcend traditional disciplinary barriers to document data about humans and their impact more broadly. As an example, while data collection instruments in the social sciences have traditionally been surveys, we can also view blood pressure gauges and magnetic resonance imaging (MRI) scans as new types of instruments that capture and export data. There is also a growing emphasis on data from administrative registers and various Internet sources. This leads to the need for a new way of looking at metadata that does not require added complexity but rather a smart and economical approach to metadata scoping and modeling.

Goals

It is critical that the new version of the DDI specification be based on UML data model as the canonical version. This model can then be expressed in XML Schema, RDF/OWL Ontology, relational database schema, and other languages. Such an abstract data model will make it easier to interact with other disciplines and other standards, to understand the specification, to develop and maintain it in a consistent and structured way, and to enable software development that is less dependent on specific DDI versions.

In addition, we envision that the new version will address:

• Abstraction of data capture/collection/source with “plug-ins” to handle different types of data
• New content on sampling, survey implementation, weighting, and paradata
• New content pertaining to qualitative data
• Framework for data and metadata quality
• Framework for access to data and metadata
• Process (work flow) description across the data life cycle, including support for automation and replication
• Integration with existing standards like GSBPM/GSIM, SDMX, CDISC, Triple-S
• Disclosure review and remediation
• Data management planning

Expected Results

This workshop will provide a forum for interested participants with both a substantive and technical focus to contribute to a re-envisioned model-driven DDI specification. Deliverables from the workshop will include drafts of the model and its documentation, which will then be made available for public review, and mappings to other standards.