Privacy methods hamper native 2020 census data
SANTA FE, N.M — During the 2020 census, Native American leaders across the U.S. invested time and resources to make sure their members were tallied during the head count, which determines political power and federal funding.
But the detailed data sets from the 2020 census they will receive this month are more limited and less accurate than they were in the previous census — and it isn’t because the COVID-19 pandemic severely limited outreach efforts.
Rather, it’s due to new privacy methods implemented by the U.S. Census Bureau in order to protect the confidentiality of participants, one of which introduces intentional errors, or “noise,” to the data.
At stake is the availability and accuracy of data helping tribal leaders make decisions about where to locate grocery stores or schools and estimate future population growth. Census numbers determine funding for social programs, education, roads and elderly care for tribes that have been historically undercounted.
“It was never clearly articulated to them by the Census Bureau that this would be the case, that they wouldn’t receive the level of data that they received from the previous census,” New Mexico State Demographer Robert Rhatigan said. “In those tribal conversations it was never made clear that the data would not be available, or that it would be so noisy in these smaller areas.”
In fact, more than 80% of tribes in the U.S. won’t receive the full suite of detailed demographic data from the 2020 census at tribalarea levels they had in the 2010 census because of the changes, according to a report released in August by the Center for Indian Country Development, which is part of the Federal Reserve Bank of Minneapolis.
Many leaders in Indian Country are unaware they are going to get fewer tables when the detailed data sets are released Sept. 21, said Brandi Liberty, a consultant who helps tribes get federal and state grants.
“It’s going to be difficult for a lot of tribes when they need the data,” said Liberty, a member of the Iowa Tribe of Kansas and Nebraska.
The 2020 census put the American Indian and Alaska Native alone population at 3.7 million people; it was 9.6 million for those who identified as American Indian and Alaska Native in combination with another race. The Census Bureau provides detailed data for 1,200 American Indian and Alaska Native tribes and villages.
The privacy changes to the detailed census data “will harm the ability of self-governing tribes to meet the needs of their citizens,” the Federal Reserve report said.
The Census Bureau told The Associated Press that it doesn’t comment on outside reports but acknowledged the number of tables for tribes in 2020 were reduced from 2010 because of the privacy concerns.
The privacy changes arrive during heightened sensitivities about who controls data from Indian Country.
“The concept of tribal data sovereignty and just data sovereignty in general has been kind of elevated. In a sense, this is their data,” Rhatigan said. “You can say that it’s a problem for the smaller tribal communities that won’t even get the detailed age data. It’s possible that the bigger problem comes from the tribes that do receive the data. Nobody knows … how inaccurate those data are.”
That’s because of the privacy method, known as “differential privacy,” uses algorithms to create intentional errors to data by adding or subtracting people from the actual count in order to obscure the identity of any given participant in a particular area.
The Census Bureau has said the differential privacy algorithms are needed because, without them, the growth of easily available third-party data combined with modern computing could allow hackers to piece together the identities of participants in its censuses and surveys in violation of the law.
The statistical agency already has released 2020 census data used to draw political districts and determine how many congressional seats and Electoral College votes each state gets.
Differential privacy’s effect on accuracy is greatest when population totals are broken down by race, age and sex, making it harder to understand demographic changes in individual tribal areas, the Federal Reserve report said.
Also complicating the availability of detailed tribal census data are new population thresholds by the Census Bureau. The thresholds determine how much data tribes, or racial or ethnic groups, get for a particular area.
In 2010, in order to protect people’s identities, a tribe or a racial or ethnic group in any particular geography like a county needed at least 100 people to get all 71 available data tables. In 2020, “dynamic population thresholds” are being used, with the size of the tribe or racial or ethnic group in a location determining how many data tables they get.
For national or state level data, the 40% of all tribes with less than 500 people across the U.S. will receive only country or state-wide population totals, keeping them from getting the more detailed data they got in 2010. At the tribal-area level, 80% of tribes will only receive population totals instead of breakdowns of age data reported by sex, according to the Federal Reserve report.