Sharing Student Data Across Organizational Boundaries Using Secure Multiparty Computation

• Activity Number: 212 - GOVT CSpeed 1
• Type: Contributed
• Date/Time: Tuesday, August 10, 2021 : 1:30 PM to 3:20 PM
• Sponsor: Government Statistics Section
• Author(s): Stephanie Straus* and Amy O'Hara and David Archer and Rawane Issa
• Companies: Georgetown University, Massive Data Institute and Georgetown University, Massive Data Institute and Galois and Galois
• Keywords: privacy-preserving technology; data sharing; cryptography; confidentiality

Abstract

Our research addresses barriers to interagency data sharing for informed decision making, as mandated by the Evidence Act (2018) and the Federal Data Strategy. Given interagency distrust and statutes that restrict sharing, we conducted a demonstration of cryptographic privacy-preserving technology--secure multiparty computation (MPC)--sponsored by the National Center for Education Statistics (NCES) at the Department of Education, to show that two independent parties can safely share and compute on their joined confidential data. We accurately reproduce statistics on average federal Title IV aid from the annual 2015-16 National Postsecondary Student Aid Study (NPSAS). We simulate the record linkage of the two different data sources, NPSAS and the National Student Loan Data System (NSLDS), used to create these statistics, and the subsequent joint analysis. Using virtual machines held in distinct “trust zones” to represent NPSAS and NSLDS host computers, these machines carry out MPC, in the form of Private Set Intersection (Pinkas et al, 2019) with associated computation. We explain the accuracy of our results, resource utilization, and degree of cryptographic privacy assurance.