Building Quality Linked Data Sources Can Help Develop Evidence for Policy
There is strong interest in using linked human services administrative data to inform policy and research, from the public sector and academia alike. However, the quality of both the sources of these data and the methods used to link them are not often examined with this particular use case in mind. The current trend of research and software development in this area are taking the field further away from the pursuit of transparency and rigor.
What We Did
We interviewed analysts and managers from four human services-focused integrated data systems and drew upon our own experience developing and maintaining the Integrated Database on Child and Family Programs in Illinois at Chapin Hall at the University of Chicago. We also conducted a literature review of record linkage methodologies, pulling from across disciplines.
What We Found
We identified three features of the human services record linkage use case that threaten the quality of the resulting linked data.
- The human services use case has unique characteristics that impact the fitness of particular record linkage methodologies.
- The current lack of independent standards for what constitutes a “good match” in linked data leave analysts and researchers without benchmarks.
- New record linkage methods and software are becoming more complex or are proprietary, with a cost to transparency.
What It Means
- The human services field needs an active research community dedicated to record linkage methodology to help develop best practices and advocate for transparency. Without such a community, there are few benchmarks to understand the quality of linked data sources used for policy research.
- Current practitioners of human services record linkage should emphasize methods that allow researchers to test the sensitivity of analytic results to match decisions.
Recommended Citation
Wiegand, E. R. & Goerge R. M. (2019). Recommendations for ensuring the quality of linked human services data sources. Washington, DC: Family Self-Sufficiency and Stability Research Consortium.