To make evidence-based decisions, public policymakers need access to data that accurately represents the population. However, a growing concern among researchers and policymakers is that statistical data privacy methods (or statistical disclosure control) — which allow access to data while preserving participants’ privacy — often fail to explicitly consider participants’ representation such as in rural regions. Techniques such as data suppression, the addition of random noise under differential privacy, or the generation of synthetic data aim to balance the need for accurate information with privacy considerations. However, each approach involves a utility-risk tradeoff that can have equity implications for different racial groups. Without accounting for people’s representation in data, researchers risk unintentionally perpetuating harm by disproportionately affecting either the privacy risks or the utility of the information derived from the data.

This work represents my efforts at the intersection of data privacy and public policy, where I strive to address these critical issues and promote a more equitable approach to data protection and governance.

Related Reports and Public Comments

• Opportunities for the New Administration to Improve Americans’ Well-Being (December 20, 2024)
• Response to a Request for Information from the White House Office of Science and Technology Policy on the Federal Evidence Agenda on Disability Equity (July 18, 2024)
• The Nation’s Data at Risk: Meeting America’s information needs for the 21st century (July 2024)
• Public Comment on Proposed Rulemaking to Expand Tax Data Sharing with the US Census Bureau (April 30, 2024)
• Toward a 21st Century National Data Infrastructure: Managing Privacy and Confidentiality Risks with Blended Data (January 2024)
• Public Comment on Initial Proposals from the Federal Interagency Technical Working Group on Race and Ethnicity Standards for Revising OMB’s 1997 Statistical Policy Directive No. 15 (April 27, 2023)

Related Papers

• Bowen, CMK. & Snoke, J. (2023) “Do No Harm Guide: Applying Equity Awareness In Data Privacy Methods.” Urban Institute.
  • Click here to view the article on the project landing page (open-access).

Related Blogs

• A blog on Data@Urban my intern wrote and led titled, Analyzing the Privacy and Utility Trade-off fro Synthetic Datasets with Imbalanced Demographic Groups
• I wrote a blog on the Urban Wire with my colleagues titled, To Advance Racial Equity, Releasing Disaggregated Data while Protecting Privacy Will Be Key
• A blog I co-authored on the Urban Wire about the Presidential Executive Order, How the Federal Government Can Use Data to Make the Most of the Executive Order on Racial Equity
• I wrote a blog on the Urban Wire titled, Will the Census’s Data Privacy Efforts Erase Rural America?

Within the extensive field of statistical data privacy literature, both differential privacy and data synthesis, and their integration, have gained significant popularity as solutions for releasing analytically valuable data while protecting individual privacy. It’s crucial to recognize that there is no methodological “silver bullet” that applies to all data. Therefore, ongoing development and refinement of differentially private methods and data synthesis techniques remain essential.

Below, I present my work in the development and practical application of differentially private and synthetic data methods to real-world datasets.

Balancing data utility against data privacy risks is a complex task. Many individuals seeking an answer to this question expect a one-size-fits-all utility or disclosure risk metric that perfectly assesses the quality of any data or statistic released under a privacy-preserving method or technology. But, such a metric doesn’t exist.

Below, I outline my work in evaluating the efficacy of different statistical data privacy methods and offer insights into how we should approach the delicate balance between privacy and utility.

The following represents my efforts in introducing data synthesis, differential privacy, and various statistical data privacy methods and technologies to a wider audience, including the scientific and non-technical communities.