Protecting Against Disclosure Risk in Shared Research Data Files

Instructor(s):

• JoAnne McFarland O'Rourke, SAMHDA, ICPSR
• Stephen Roehrig, Public Policy and Management, Carnegie Mellon University
• William Birdsall, University of Michigan
• David Metcalf, SAMHDA, ICPSR
• Kristine Witkowski, Research Staff, ICPSR
• Deborah Schild, SAMHDA, ICPSR

Ensuring disclosure risk protection in shared data files involves concurrent analysis of (a) key analytic data utility (b) key disclosure risk factors and (c) disclosure protection factors inherent in the data. This workshop examines how to best achieve this through disclosure analysis and the application of disclosure protection methods (statistical disclosure control).

Disclosure analysis involves the careful examination of a data file for indirect identifiers that could pose the risk of re-identification of a research participant. By examining variables containing detailed personal characteristics such as education, income, race, ethnicity, and military service or organizational characteristics such as capacity, services offered, and programs for special populations, it quickly becomes possible to begin to narrow identity. Analysts are often interested in subgroups of survey populations (e.g., pregnant women, racial minorities, and persons with health conditions) and comparisons of subsets within the data. Yet these are often the very characteristics that create disclosure risk.

Disclosure protection methods must take into account the key uses of the data and balance the trade-off between analytic utility and data protection. (See http://www.icpsr.umich.edu/ICPSR/org/publications/bulletin/2003-Q3.pdf). Using examples of ICPSR and other studies having undergone disclosure analysis, this hands-on workshop provides participants with tools for understanding and reducing disclosure risk. In addition to examples and hands-on exercises, the workshop covers:

• Background on disclosure risk analysis
• Types of disclosure risks
• Types of disclosure protection methods
• Impact of research and sample designs on disclosure risk
• Relationship between disclosure risk and the confidentiality requirements of ethical research
• Five essential steps ICPSR takes to reduce disclosure risk
• Impact of different disclosure protection methodologies on data utility
• Models of disclosure review boards (DRBs)

Dates:  July 27-29 

Fee:  Member: $550; Non-member: $550

This course is limited to 20 participants.