2020 Census Data And Differential Privacy: What You Need To Know The Census Bureau must protect people's privacy when it releases demographic data from the 2020 count. Plans to change how it does that have sparked controversy over how it may affect redistricting.

For The U.S. Census, Keeping Your Data Anonymous And Useful Is A Tricky Balance

  • Download
  • <iframe src="https://www.npr.org/player/embed/993247101/1008304562" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript

SCOTT SIMON, HOST:

Federal court is expected to rule soon in a lawsuit that could cause major delays to upcoming elections. The state of Alabama is trying to stop the Census Bureau from putting in place new privacy protections for redistricting data. NPR census correspondent Hansi Lo Wang explains.

HANSI LO WANG, BYLINE: This legal fight is about a tricky balancing act the Census Bureau has to face every 10 years. How does the bureau release detailed demographic information from the national headcount in which every household is legally required to participate while also keeping this promise...

(SOUNDBITE OF ARCHIVED RECORDING)

UNIDENTIFIED PERSON: After sending your census response, your personal information is kept safe.

WANG: ...As described in these 2020 census ads?

(SOUNDBITE OF YOUTUBE VIDEO, "MAKE IT COUNT / #2020CENSUS")

ANNE FRANCIS GUZMAN: (As Mrs. Lopez) It's completely safe. They can't share your answers with anyone - no police, no immigration.

WANG: Those census answers are shared with the public 72 years after they're collected. Until then, the bureau uses them to produce a lot of statistics about people's race, ethnicity, age, sex and other characteristics. Under federal law, the bureau has to keep people anonymous in those statistics. And a few years ago, the bureau concluded the privacy protections it's used in the past are no longer strong enough.

CYNTHIA DWORK: It was only a matter of time in the digital age before somebody would break them.

WANG: This is Cynthia Dwork.

DWORK: I'm a professor of computer science at Harvard.

WANG: Who's also with the research arm of Microsoft, one of NPR's financial supporters. Dwork co-invented differential privacy. It's the mathematical concept the bureau is using to build a new privacy protection system for the census data expected out by mid-August. Dwork says the new protections are needed to keep up with advances in computing and the growing amount of data available from commercial sources. They've made it easier to trace statistics back to an individual and use them against people.

DWORK: I come from a privileged home. I never think of whether anyone's going to question my right to be residing there. But there are people who are in much more precarious situations than I am. It's to protect them that these things become so important. It's to protect the vulnerable.

ALEXIS SANTOS: What's at stake is our understanding of the portrait of the United States.

WANG: Alexis Santos is a demographer at Penn State University and one of the many users of census data who have been looking at early tests of the bureau's privacy protection system and are concerned about how it blurs the census results.

SANTOS: What we're seeing with the implementation of differential privacy, at least with the data I've analyzed, is that you have blurry patches. You have blurry areas. You have areas that do not look as they should look.

WANG: Based on preliminary analysis, Santos says the new system could make some of the new 2020 census data useless, especially information that many policymakers and researchers rely on to determine the needs of small geographic areas and minority groups within communities.

SANTOS: A lot of people who are working everyday jobs who rely on demographics, they get their information from public data. And what we are at risk here is of putting out tabulations that do not help them at all and can misguide the decisions they're making.

WANG: Alabama is arguing in its lawsuit that the bureau's privacy plan will make the data unusable for the redrawing of voting districts. The bureau has been adjusting its privacy protections, and it says the plan it recently finalized will ensure, quote, "the accuracy of data necessary for redistricting and Voting Rights Act enforcement." Still, some civil rights advocates are skeptical of the bureau's plans, including Thomas Saenz, president of the Mexican American Legal Defense and Educational Fund.

THOMAS SAENZ: At this point, it seems not at all clear that anything the bureau releases will eliminate the possibility that the Voting Rights Act and its enforcement could be adversely affected by differential privacy.

WANG: And that could potentially result in more lawsuits.

SAENZ: Yes.

WANG: For Alabama's lawsuit, if the courts ultimately block the bureau's differential privacy plans, officials estimate it could take at least half a year to develop new plans and would further push back the release of data that state and local redistricting officials have already spent months waiting for because of delays caused by the pandemic and the Trump administration.

Hansi Lo Wang, NPR News, New York.

Copyright © 2021 NPR. All rights reserved. Visit our website terms of use and permissions pages at www.npr.org for further information.

NPR transcripts are created on a rush deadline by an NPR contractor. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.