The new census data privacy formula could affect Arizona redistribution, but opinions differ

Posted on 7/15/2021 at 12:06 pm

Jeremy doubt

Arizona mirror

A fresh attempt by the US Census Bureau to protect people’s privacy could make it difficult for the Arizona Independent Redistricting Commission to draw districts composed mostly of colored people that can withstand litigation under the Voting Rights Act.

Federal law requires the Census Bureau to protect the identity of each respondent for 72 years in each ten-year census. To meet this requirement in the face of advances in computing and other technology making it easier to discover individual identities from census data, the Office began using a system known as differential privacy for the 2020 census.

Differential data protection is a mathematical system that uses arbitrary changes to data to prevent individuals from being identified within a data set. For reassignment, this means that the Census Bureau will add “noise” to the changes and obfuscate some data to prevent people from using census data to identify people. For example, if all of the residents on a census block were white, except for a Latino, that person would be easy to identify from the census data. Differential privacy would obscure this information so that the individual could not be identified individually.

A federal appeals court last month denied an Alabama state motion to prohibit the Census Bureau from using differentiated privacy in census data. The appeals court also denied the state’s motion to force the office to bring forward the publication of the census data that states need for redistribution. The office now plans to release this data on August 16 in an archaic but usable format.

The commission on Tuesday heard competing presentations from two redistribution efforts on the impact different privacy will have on their efforts.

Thomas Bryan, the founder and CEO of GeoDemographics, a demographic and analytical consultancy, told the commission that disparate privacy issues could create disparate privacy issues for AIRC when faced with complaints about majority minority districts or counties where the majority of the People are minorities.

Differences in data protection can lead to strange anomalies at the counter block level. Bryan showed commission data from a Census Bureau demonstration project in the Alabama litigation that showed blocks in Arizona listing all residents as under 18 years of age or old, blocks with more than 15 people per household, or blocks with no population.

The biggest problem with redistribution, however, is how different privacy changes block-level population data on racial and ethnic demographics, which Bryan said disproportionately affected minorities. Of nearly 242,000 Arizona census blocks in the demonstration project, Bryan said that the difference in privacy has changed about 6,000 from mostly white to mostly non-white.

Section 2 of the Voting Rights Act calls for districts where minority voters have the opportunity to vote for candidates of their choice. States and other jurisdictions often accomplish this by creating districts in which minorities form a majority of a district’s electorate.

These population disparities are worrying, Bryan said, because there is no way of knowing exactly how or how much the population has changed in any given block.

“We’re not going to be sure what you’re saying when we create a district because we know the dates for those building blocks are wrong,” Bryan told commissioners.

Moon Duchin noted that the difference in privacy is much less of a concern from a redistribution standpoint.

Duchin, a researcher at the MGGG Redistricting Lab at Tufts University, took a broader look at the implications of different privacy and urged the AIRC to look at it in a much more reassuring way. While differential privacy changes data at the census block level, she said that when aggregated into larger units, the population statistics are accurate.

“The whole design of Differential Privacy is made up of lots of small changes to the smallest units, so when you put things together they look really good,” said Duchin.

That also applies to much smaller districts than the commission will pull, said Duchin. Arizona congressional counties will each have approximately 795,000 residents, while the legislative counties will average nearly 239,000 residents.

Duchin presented the commission with data from a number of theoretical district regulatory counties in Navajo County of 20,000 people. After adding “noise” from the difference in privacy, these districts had only a half percent error rate in the Indian population of these districts. Based on the extent of the changes the Census Bureau is making to blocks with differential privacy, Duchin said they created “almost invisible discrepancies” in the districts of Navajo County.

“The big takeaway message is that for each region we considered … we have concluded that the census data is clearly appropriate for each remap application we considered,” said Duchin, who added that their organization found no threat to the enforcement of the voting rights law due to differing privacy.

Bryan countered that the totals advertised by Duchin would not play a major role in litigation over voting rights.

In order for the plaintiffs to be able to assert themselves in contesting voting rights against districts, they must prove that it was possible to create majority-minority districts that were ultimately not drawn. That makes the exact position of minority populations important, said Bryan. In districts with very small minority majorities, he explains that different privacy levels can make the difference in whether a district is considered a majority minority.

Bryan predicted that the advent of differential privacy in the census will end the practice of drawing such districts and force mapmakers to use “a higher bar” to ensure their districts can withstand litigation.

“I understand that when you put a thousand blocks together it will look great overall. But in litigation in these cases, we cannot afford the luxury of simply saying that we want it all together. We have to go to the scalpel, look at small, individual pieces of geography and say, we want this, we want this out, ”he said.

Commission chair Erika Neuberg noted that differences in privacy could be a bigger problem for Congress than for the legislature. While the courts allow some variations in the population of a state’s legislative districts within about 5% of the average, such variations are not allowed in congressional districts that must have almost exactly the same population within a state.

“It sounds as if the smaller, closer group is more at risk from the noise,” said Neuberg.

But even if the difference in privacy poses problems for the AIRC, Commissioner David Mehl asked if there were other data that it could rely on instead.

“Is there any reason we should hesitate to use the census data, and if so, what would the alternative be?”

The only option any of the AIRC’s experts put forward was the Census Bureau’s American Community Survey, which collects more limited population data that is used for periodic estimates between census years. Duchin noted that some states have already started to refine the work with such data, while Bryan emphasized that the data are only estimates based on 10-year-old information and that there are sometimes notable differences between official data and the data of the survey.

Using the American Community Survey’s estimates for Arizona could be particularly difficult because they are so different from the official census figures. The ACS estimated that Arizona’s population had reached 7.4 million in 2020, but the census put the state’s population at less than 7.2 million. The difference cost Arizona a widely anticipated 10th congressional seat.

The variety of privacy issues may be less important to Arizona than they were in previous redistribution cycles. For decades, Arizona was subject to a provision of the Voting Rights Act known as “Preclearance,” which required the Department of Justice to approve all changes to electoral and voting laws, including new counties. But the US Supreme Court in 2013 overturned the formula of the Voting Rights Act, which established which jurisdictions are subject to preliminary screening, and overturned the requirement of Arizona and other states.

Tuesday’s commission meeting was suspended due to a power outage that hit several buildings in the State Capitol complex, including the Arizona Department of Environmental Quality building that houses the AIRC headquarters. The remainder of the meeting will take place on Monday at 8:00 a.m., followed by the regular meeting of the commission on Tuesday.

– 30 –