Transparency in the Reporting of Quality for Integrated Data: A Review of International Standards and Guidelines

Publisher: Washington, DC: Mathematica Policy Research
Apr 27, 2018
Authors
John L. Czajka and Mathew Stange

Key Findings:

  • Only one national statistical organization—Stats NZ—has developed a quality framework explicitly designed to address integrated data.
  • Eurostat’s quality standards and guidelines, which apply to most of Europe and are perhaps the most extensive, deal with integrated data to a much more limited degree and instead focus on quality more generally.
  • Many of the quality assurance frameworks and the associated standards and guidelines reviewed in this report are associated with extensive prescriptions for quality assessments and their communication to data users in detailed quality reports. The volume and types of information requested in Eurostat quality reports bears substantial resemblance to what was included in the quality profiles prepared by a number of U.S. federal agencies in the 1990s and early 2000s but, for multiple reasons, not continued.  
  • Efforts to deal with quality aspects of administrative data are much farther along than efforts to deal with the quality of other forms of Big Data.
The research landscape for federal statistical agencies is moving to a new paradigm in which survey data are no longer the principal data type. This shift is due to growing challenges facing traditional survey research, including an increasing reluctance of people to complete surveys and deteriorating coverage of sample frames. The new paradigm is characterized by the use of administrative data and other forms of Big Data as alternatives to survey data, and, increasingly, the use of integrated data that combines data from multiple sources, such as linked survey and administrative data. This new paradigm necessitates new quality standards that address integrated data. This report reviews information on international standards and guidelines on quality reporting relative to statistical estimates that combine survey data with other types of data.
Senior Staff

John Czajka
Read More