Frauke Kreuter (University of Maryland)
Within the U.S. Federal Statistical System there is a common notion to transform statistical agencies from relying primarily on survey data to create statistics by combining survey and administrative data. This notion is shared by program agencies and based on recommendation provided by the Commission on Evidence-Based Policymaking, new legislation is now in place that promotes the access of administrative data and the sharing of such data across program agencies. In all policy documents proper protection of privacy is mentioned as a desired goal. This presentation will introduce a set of typical administrative data, describe its structure, size, types of variables, describe applications/use cases for the data, showcase typical amounts of data cleaning and feature generation, and discuss the challenges for analysts should data be accessible only in a DP context.