What is Differential Privacy?

Turn on any news outlet and data privacy and accuracy sit at the forefront of the conversation. No organization is too small or too large to outwit potential privacy breaches, not even the federal government.

The U.S. Census Bureau has long been a steward of data privacy to ensure that responses to surveys such as the decennial census, American Community Survey (ACS), and Current Population Survey remain confidential. Title 13 of the U.S. Code prohibits the Census Bureau from releasing any identifiable information of people who fill out the decennial census and the ACS.

Currently, the Census Bureau applies advanced statistical methods called disclosure avoidance techniques to protect the data. These methods include data swapping and noise injection to safeguard against linking the data back to any specific individual.

With evolving technology and reinterpretation of existing laws comes increased responsibility. Recognizing the need to further balance data privacy and accuracy, the Census Bureau is in the process of adopting a more stringent privacy technique called differential privacy. This change has been a hot topic in the data world.

All agree that protecting people’s privacy is of utmost concern and must remain a priority. However, the error rate resulting from the new equations and techniques has raised alarm for state agencies, data intermediaries, and community partners who rely on accurate data to address pressing community needs such as planning for schools, transportation, and social services.

The question remains—at what point does the data become so masked that it is rendered unusable for planning, programmatic, and policy purposes?

The Census Bureau sought feedback on their initial metrics and numerous partners commented with concerns, including Connecticut. Real change is happening based on this work! The Census Bureau continues to evaluate and update their metrics, and is committed to transparency throughout the process. New summary metrics are released online so that users can judge their utility for themselves.