Balancing data privacy and usability in the federal statistical system.

Area of Science:

• Statistics
• Data Privacy
• Social Science Research

Background:

• Federal statistical agencies face conflicting demands: increasing data accessibility for research versus protecting subject privacy.
• Current data privacy methods may reduce data accuracy and limit research potential.
• Existing discussions often overlook the social benefits of data availability and usability when weighing privacy concerns.

Purpose of the Study:

• To advocate for a balanced benefit-cost framework for assessing data privacy methods.
• To critically evaluate synthetic data and differential privacy methods.
• To propose improvements in measuring disclosure risk and enhancing data usability for research.

Main Methods:

• Analysis of current pressures on the federal statistical system.
• Critique of synthetic data methods and differential privacy criteria.
• Proposal for a revised benefit-cost framework and risk assessment.

Main Results:

• Current privacy protection methods, including synthetic data and differential privacy, may inadequately balance privacy with data utility.
• The measure of disclosure risk in differential privacy may not align with risks individuals are concerned about.
• Widespread adoption of synthetic data and differential privacy may hinder valuable research due to unknown impacts on data usability.

Conclusions:

• A more comprehensive benefit-cost analysis is needed to guide decisions on data privacy and accessibility.
• Disclosure risk measures should focus on risks relevant to individuals.
• Further research is essential to understand the impact of new privacy methods on data usability and to develop better alternatives.