Skip to main content

The Promise and Peril of State Administrative Data: An Example from an Evaluation of Low-Barrier Buprenorphine Treatment for Opioid Use Disorder

Headshot of Jason Williams

Jason R Williams


Large datasets sourced from administrative data represent an enticing advantage over primary data collection in research and evaluation. Whereas 25 years ago “administrative data” meant poring over large paper files, cheap computer storage makes records for thousands of observations readily available. To the extent that data entry is primary to the work of the institution, these records may accurately capture interactions, background characteristics, and even outcomes. They may not, however, provide this information in a way that can be readily entered into a statistical model. This presentation illustrates some of the learning that ensued when state records were brought to bear to evaluate a low-barrier treatment option for opioid use disorder.

Dr. Williams is a research scientist at the Addictions, Drug & Alcohol Institute at the University of Washington School of Medicine. At ADAI, he works on tracking trends in problematic drug use, analysis of survey data, and evaluations using both primary and secondary data. He received his PhD from the Evans School of Public Policy and Governance in 2013 with assistance from CSSS and the West Coast Poverty Center.