2020 Census May Have Missed More Than 1.6M Residents

The 2020 census may have undercounted the U.S. population by more than 1.6 million people, drastically affecting the distribution of federal funding across the country, according to new research from the Urban Institute.

The census, researchers noted, is always an imperfect and inexact process that undercounts certain populations, particularly minorities. Their goal, then, was not to determine “whether accuracy was achieved, but its utility for specific purposes such as apportionment and allocation of federal resources,” according to the report.

Robert Santos, one of the report’s advisers and the chief methodologist for the Urban Institute, is President Biden’s nominee to head the U.S. Census Bureau.

The left-leaning think tank estimated the accuracy of the census by conducting a hypothetical full population count, then comparing the results with official census data. That “microsimulation” allowed researchers to better understand the various factors that can influence the accuracy of the census “and corresponding implications for political representation and allocation of federal resources.”

According to the results, the overall U.S. population was likely undercounted by about .5%, or roughly 1.7 million people. The degree of underrepresentation varied widely between demographics, researchers found, and across state lines. For example, Black people were undercounted by 2.45%, the highest in the country, followed by Latinx people (2.17%) and Pacific Islanders (1.52%). By contrast, “the data show that 0.39 percent of the white population was net overcounted,” the report says.

There are a number of explanations for the discrepancy, researchers noted, including the demographic shifts in the United States between the last census in 2010 and the count in 2020. Historically, the report said, “households with a non-Hispanic/Latinx white head of household have had higher percentages of overcounts and lower percentages of being missed by the census” compared to other ethnicities.

“For these reasons, we expect greater racial/ethnic diversity likely contributed to a larger net undercount for the overall U.S. population,” researchers concluded.

Children under the age of 5 were likely undercounted by 4.86%, higher than the rate in 2010, and a national decrease in homeownership contributed to an uptick in the undercount rate for renters, which nearly doubled in the last 10 years, from 1.1% in 2010 to 2.13% last year. That rate was much higher—3.36%—for households with noncitizens, the report found.

Crises Impact CountThe report, released this week, is one of a host of studies examining the 2020 census, a massive undertaking that occurred against the backdrop of multiple crises that threatened its accuracy. There was the pandemic, announced officially in the United States on March 13, 2020, “a day after the 2020 Census began mailing information to households to participate,” the report said. The spread of Covid-19 led to the deployment of an online response tool, but likely also limited participation among already hard-to-count populations due to delayed field operations and hiring.

“Indications exist that enumerating apartment dwellers became harder in 2020, in part because of the pandemic, further contributing to concerns about coverage,” researchers wrote. “Then the pandemic affected living arrangements, complicated in-person follow-up counts, and delayed post-enumeration data cleaning and other processes.”

The census also became political, the report noted. Researchers estimated that that process began as far back as 2017, when the Trump administration came into power and fundamentally shifted the nature of the conversation on immigration, including via executive orders that banned travel from Muslim-majority countries and attempted to dissolve the Deferred Action for Childhood Arrival program.

“This rhetoric contributed to unprecedented confidentiality concerns, particularly among immigrants and people of color,” the report said. 

A proposal by the Trump administration to add a citizenship question to the census “could have also heightened growing fears and suppressed household participation,” particularly among Hispanic populations and both documented and undocumented immigrants, the report said. The Supreme Court eventually blocked the addition of the question, but officials continued to discuss using administrative records to determine citizenship status, which likely led to a chilling effect among multiple populations.

“For these reasons, many experts assume that some households likely did not participate and noncitizens were missing from household rosters, even when households responded, because of fear and government distrust,” the report said.

States With Biggest Discrepancies

Geographically, Mississippi had the highest rate of undercounting (1.3%), followed by Texas (1.28%). By contrast, Minnesota’s population was likely overcounted by .76%. Even those seemingly small numbers can have big implications, researchers noted.

“Because Texas has such a large population, this means that 377,187 residents in the true population of Texas were not counted in the 2020 census,” they wrote. “Framed in different terms, this means that more than one-fifth of all people not counted in the 2020 census resided in Texas.”

The discrepancies translate to large differences in funding allocations for each state. Texas, for example, would have received more than $247 million in additional Medicaid reimbursements if every one of its residents had been counted, while Minnesota stands to gain an extra $156 million due to some of its residents being counted more than once. 

There are also implications for congressional representation, the report said—if the national population had been counted accurately, Minnesota would drop from eight to seven representatives, while New York would gain one, for a total of 27.

The Urban Institute report comes months ahead of the expected release of what’s known as a modified race file, a data analysis from the U.S. Census Bureau that reassigns people who identified only as “some other race” into Black and non-Black categories. The agency will also release results from a post-enumeration survey, a standard sampling designed to assess the accuracy of the count, sometime next year.

In 2010, the post-enumeration survey estimated that the census count was within 0.1%of the actual population. 

In a statement, the bureau reiterated its “ongoing commitment to producing data that depict an accurate portrait of America, including its underserved communities.”

“We understand firsthand how important this data is,” Ron Jarmin, the agency’s acting director, wrote on his blog on Nov. 2. “We use it ourselves to better reach hard-to-count populations during the census as part of our goal to count everyone once, only once and in the right place.”