Publications about Privacy Preserving Methods
Using large or sensitive datasets requires a balance between efficiency, accuracy, and privacy. We have produced analyses that describe privacy preserving methods in several different domains, and demonstration projects showing the strengths (and weaknesses) of specific approaches.
Privacy Preserving Technologies in Education MDI recently completed a project with support from the Bill & Melinda Gates Foundation to study privacy preserving technologies (PPTs) in the education sector. This report identifies the barriers to PPT implementation in K-12 and higher education data systems, including lack of technical capacity, institutional setbacks, cultural opposition, and legal fears. We outline recommendations for government, foundations, and associations to help overcome these barriers, to enable greater insights into learner data that can inform student success.
The TopDown Algorithm MDI summarized the approach taken by the U.S. Census Bureau for release of the Redistricting Data File and Demographic and Housing Characteristics Files. This formal privacy protection applied to the 2020 Census results differs from disclosure limitation approaches taken in past censuses.
Sharing Sensitive Department of Education Data Across Organizational Boundaries Using Secure Multiparty Computation In the spring of 2021, MDI partnered with the National Center for Education Statistics within the U.S. Department of Education (ED) to demonstrate how secure multiparty computation, a privacy preserving technique, could be deployed in a production environment. This project involved linking and computing on two sensitive, disparate ED datasets to assess average undergraduate federal student aid. For a more in-depth summary of the computer science methodology behind the project, please see the technical version of the report.
Postsecondary Data Infrastructure: What is Possible Today With the Institute for Higher Education Policy, MDI published a white paper describing the as-is data access and protection mechanisms among post-secondary institutions and outlined improvements in privacy preserving methods that could be implemented immediately.