The Committee on National Statistics (CNSTAT) of the National Academies of Sciences, Engineering, and Medicine recently issued a consensus report entitled Federal Statistics, Multiple Data Sources, and Privacy Protection: Next Steps. The report was produced by the Panel on Improving Federal Statistics for Policy and Social Science Research Using Multiple Data Sources and State-of-the-Art Estimation Methods, chaired by Robert Groves of Georgetown University. The Panel’s first report, Innovations in Federal Statistics: Combining Data Sources While Protecting Privacy, was published in January 2017, and described some of the challenges currently facing the federal statistical system’s current paradigm of heavy reliance on sample surveys and recommended a new approach of combining different kinds of federal and private data, as well as the creation of an entity to facilitate that. Federal Statistics, Multiple Data Sources, and Privacy Protection builds on the first report and examines statistical methods for combining diverse types of data, the implications relying on multiple data sources may have for IT systems, different statistical and computer science approaches to enhancing privacy protections, how to ensure the quality and utility of statistics produced using multiple data sources, and ways to implement the “new entity” that would facilitate combining data sources. The pre-publication version of the report is available on the National Academies’ website.
There is quite a bit of overlap in the areas addressed by the CNSTAT panel and those addressed by the Commission on Evidence-Based Policymaking, which released its report in September (see COSSA’s coverage the Commission)—in fact, Panel Chair Robert Groves served on the Commission as well. However, while the resulting reports from the two groups are hopefully complementary, their work was conducted independent of one another.