Questions Raised About Census Method That Estimated Data For Some Households

Questions Raised About Census Method That Estimated Data For Some Households

By Terri Jo Neff |

Questions remain in the aftermath of the U.S. Census Bureau’s release last month of the 2020 Decennial Census which put Arizona’s official population at nearly 7.16 million.

That figure, up more than 746,000 from a decade earlier, represented the eighth largest increase by number and the ninth fast growth rate in the country.  But census officials determined it was not enough to earn Arizona its 10th seat in the U.S. House of Representatives.

There was much finger pointing at the time of the announcement that Arizona’s population fell 250,000 short of estimates even though Gov. Doug Ducey’s AZ Census 2020 Taskforce reported that census enumerators and volunteers attempted to reach 99.9 percent of all households in the state despite COVID-19 lockdowns and social distancing challenges. 

But according to a theory by a former official who served in the Trump Administration’s Commerce Department, the reason some states ended up with unexpected results might have more to do with how the Census Bureau calculates the number of people who live in known households that did not fill out a census questionnaire.

Adam Korzeniewski is a Marine Corps combat veteran who specializes in fiscal and economic policy as well as national security topics. In an article published earlier this month in The American Mind, he cites the Census Bureau’s reliance on estimates as cause for concern for some states, including Arizona.

Federal law does not allow the Secretary of Commerce to rely on statistical sampling to fill in the blanks for households that do not respond to the census. But other methods are allowed, and Korzeniewski believes some of those need to be questioned.

According to the U.S. Census Bureau, demographic characteristics about the people who live in every verified residence is necessary to obtain “a complete and accurate census. Unfortunately, not all households respond to the census questionnaire. When that happens, a census taker first turns to administrative records from the IRS and other government agencies to fill in the blanks about who likely lives in that household.

If a reliable administrative record is not available and the local census taker is unable to contact the household after three visits, information can be obtained from a neighbor, landlord or building manager. This is referred to as a proxy response.

The last method the Census Bureau uses is imputation, a statistical technique officials say makes the overall dataset “more accurate than leaving the gaps blank” by using what data is known to filling in what data is not known.

“We recognize that using information from these three techniques — imputation, using administrative records, and proxies — may not always match the reality of an address’s occupancy status or the characteristics of the people who live there,” the Census Bureau says. “However, these techniques are widely used in statistics because they have been proven to be more accurate than leaving the information blank.

Public records show about 1,172,000 “people” were imputed nationwide in 2000 and 1,163,000 in 2010. The 2020 figure has not been released yet although it is expected to be higher due to COVID-19 related difficulties.

But Korzeniewski, who also served in Trump’s Treasury Department, wrote in his article that the Census Bureau used another form of imputation in 2020 which he contends is based on statistical sampling. This occurred when census officials decided to utilize a “Group Quarters Imputation” due to problems gaining access to “households” located in places like colleges and residential healthcare facilities.

The greater reliance on such imputation was not part of the 2020 Decennial testing phases nor did state census officials have any input on the decision, Korzeniewski wrote.  That could be the basis for Arizona officials to acquire state-by-state imputations records from the Census Bureau with details of the types of imputation used.

“To my knowledge, the Census does not normally produce such documentation and it takes years for the Census to publish studies on itself,” Korzeniewski wrote, adding that states would also need to ask for records pertaining to the decision-making processes around the data calculation processes in order to determine whether it has grounds to challenge the Census outcomes.

“Successfully challenging the Census results would affect appropriation and could affect apportionment,” he wrote. “The Census typically takes years to officially release information on the Decennial, making it impossible for states to seek redress if action is not taken quickly.”