Produced by the “Survey Metadata: Barriers and Opportunities” Meeting
June 26, 2014, London
Issued 31 October, 2014
This collaborative statement calls upon the survey design, production and archiving communities to take leadership in facilitating survey metadata exchange through adoption of shared metadata standards for questionnaire and data description.
Statement of Shared Principles for Survey Metadata Reusability and Exchange
Currently there are some obstacles in creating good structured documentation for data that complies with international gold standards for enabling use and longer-term re-use.
The journey from questionnaire design using commercial software to Computer Assisted Interviewing (CAI) delivery in the field, and to delivery of data to owners and archives is currently inefficient and error-prone. The metadata journey from owner to producer back to owner and to archive continues to use relatively old-fashioned processes which rely on manual intervention and some replication of activities. With the move to joined up and federated resource discovery and online data delivery systems, which adopt universal data description standards, it is vital that the processing time is improved and that systems are more harmonised (enter once, publish many times).
The model of delivery of questionnaire content to producers to enter into their CAI systems and then exporting back out as data and documentation would benefit from being streamlined and making use of an approved metadata standard, such as the Data Documentation Initiative (DDI), used by most international survey data archives.
Metadata specialist companies are looking at the data production cycle in a more holistic way, considering reusable elements (same questions, same response sets) in questionnaire design through to delivering useful structured metadata for archiving data in a recognised format (DDI3.x) for exchange.
In considering these challenges, it is useful to think about whose role it is to produce data documentation. Survey producing agencies play a central role in upholding the principles of data quality in surveys. If the decision-makers for such survey outputs, such as survey owners and commissioners, raise the bar to insist upon fine-grained data documentation in contracts (over and above the norm of a standard SPSS file as output), the effects will ripple across the survey data production enterprise. This kind of demand puts pressure on the producer to use tools that can deliver metadata in this format.
Currently the core tools in use, IBM SPSS Data Collector and Blaise, cannot export metadata in DDI format, or only in very limited ways. Does this mean owners/producers may be forced to choose new CAI software suppliers that do enable such delivery or does it mean that current suppliers need to rise to the challenge of creating DDI export formats from their CAI systems?
The Survey Metadata: Barriers and Opportunities meeting held on 26th June 2014 in London brought together survey owners, producers, CAI suppliers, and archives to discuss the challenges of working in a more harmonised climate that avoids the many hours of manual effort spent on creating high quality survey data documentation. The meeting agreed to support a set of principles and to define minimal metadata standard via a designated DDI profile for stakeholders to adhere to. The emphasis is on a pragmatic solution.
Recommendations for Survey Owners, Producers and CAI Vendors
Better documented data at the variable-level facilitates re-use of such metadata in and out of question banks and CAI systems, enables publishing and sharing of variable-level data in dissemination system, and improves longer-term, more sustainable archiving. Survey agencies rely on commercial software to deliver their surveys, and are thus reliant on the capability of such software to accommodate import and export of richer question-level documentation.
The following recommendations offer a set of practices that exemplify a commitment to survey metadata re-use. While there is no one-size-fits-all approach, these principles can offer a standard towards which stakeholders should strive.
Questionnaire design and CAI systems that embody a commitment to survey metadata re-use:
- Can accommodate description of variables in full context, including information such as full question text, universe and routing, showcard information, and so on
- Offer import and export of metadata in the recommended DDI3.2 Questionnaire profile in XML format (Description).
- Agree to keep exchange mechanisms up to date with future upgrades to the DDI 3.2 Questionnaire profile