Toward Protecting the Privacy of Individuals When Disseminating Data: Challenges in Disclosure Risk Assessment

Slides from Privacy & Confidentiality Committee sponsored webinar presented on February 6, 2019. Link to recording Description: We discuss the theory and background of useful approaches for determining the level of data disclosure risk prior to disseminating data. We focus on challenges experienced from several case studies and provide insights into the ways that the risks can be estimated. These challenges relate to risk estimate diagnostics and sensitivity analysis, longitudinal risk, when public use data already exists, variable selection, setting risk thresholds, impact of weights and missing values, query tools, geography, attribute disclosure, risk of synthetic data, and external data. Some discussion of differential privacy is provided in this context. Once the key elements that trigger the relatively high risks are understood, data protection approaches can be determined and processed while balancing the risk reduction with retention of data utility. Tom Krenzke, senior statistician and Associate Director in the Westat statistical group. He has led research on data disclosure limitation methods working toward solutions for several Federal agencies and Disclosure Review Boards. Mr. Krenzke is an ASA Fellow and President of the Washington Statistical Society, and serves on the ASA Privacy and Confidentiality Committee and Westat Institutional Review Board. Jianzhu Li, senior statistician at Westat. Dr. Li has 16 years of experience in survey research, including sample design, nonresponse adjustment, imputation, data confidentiality and disclosure protection, and small area estimation. She has worked on several statistical confidentiality projects over the past few years for Government agencies such as NSF, NCHS, NCES, Census Bureau, and USDA