Skip to content

Census may have missed more than 1 million, report suggests

Minorities, renters and children among those undercounted — again

A Census Bureau employee demonstrates at an outreach event for the 2020 census last year in Washington. The count was hampered by the pandemic and decisions by the Trump administration.
A Census Bureau employee demonstrates at an outreach event for the 2020 census last year in Washington. The count was hampered by the pandemic and decisions by the Trump administration. (Sarah L. Voisin/The Washington Post via Getty Images file photo)

The 2020 census may have missed more than 1.5 million people, enough to cost New York the congressional seat that went to Minnesota, according to a report released Tuesday.

Research conducted by the Urban Institute, a nonpartisan research organization, also estimated that the Census Bureau double counted white residents and missed people of color, renters and young children in its once-a-decade count, which was hampered by the coronavirus pandemic and decisions by the Trump administration.

New York would have kept its 27th District seat at Minnesota’s expense, according to the Urban Institute report, which compared official census data with a model of a hypothetical full count. In apportionment data, New York lost that seat by 89 people this year, the smallest margin in modern history.

[Census: California, Northeast, Midwest lose House seats]

The Urban Institute estimated a close to 1 percent net double count in Minnesota, contributing to the state keeping all of its House seats in reapportionment.

Census misses could also contribute to massive shifts in federal funding. The Urban Institute report said the misses could have cost Texas $247 million and Florida $88 million in Medicaid funds in fiscal 2021. On the other end of the ledger, states like Pennsylvania, Ohio, Minnesota and Michigan would have received considerably less Medicaid funding under an accurate count.

Urban Institute’s research adds to a slew of reports by demographers estimating last year’s troubled count could have missed hundreds of thousands or even millions of people. Experts said undercounts can skew congressional representation as well as the distribution of more than $1.5 trillion in federal funds guided by census results annually.

“Overall, these data show that some communities and their residents will be shortchanged for the next decade due to an incomplete count,” the report’s chief author, Diana Elliott, told reporters prior to the release of the report.

In a statement, the Census Bureau acknowledged the importance of reports examining the accuracy of the 2020 census and said it will release more information about the quality of the count next year. When the agency released the population data in April, officials said the census met the agency’s internal quality benchmarks.

Elliott wrote that the decennial census will likely never get a perfect count, but several factors contributed to last year’s undercount. She noted that people the census has missed in the past — renters, young children in minority communities and other traditionally “hard to count” groups — have grown as a proportion of the population.

Several moves by the Trump administration also politicized the census, including the failed attempt to add a citizenship question, which Elliott said contributed to heightened distrust and made some people less likely to respond. The unfolding coronavirus pandemic also scrambled the census process, delaying in-person counting efforts for months.

Elliott noted that her report found less of an undercount than some people feared, including her own organization. Before the 2020 census started, the Urban Institute published a report estimating the count could miss more than 1 percent of the population due to underfunded preparations, untested techniques and fears caused by the Trump administration’s effort to add the citizenship question.

Still, the report found the count could have had the largest miss on the true U.S. population in 30 years. The undercounts were most concentrated in minority communities and major cities, including Miami, Los Angeles and Houston. According to the report, the census may have missed more than 2 percent of the nation’s Hispanic or Latino population and almost 2.5 percent of the Black population, while double counting about 0.4 percent of the white population.

Robert Santos, a vice president at the Urban Institute and President Joe Biden’s choice to lead the Census Bureau, authored several of the organization’s reports on the census process but was not cited in Tuesday’s report.

The Census Bureau is currently conducting a post-enumeration survey, which will serve as the official measure of the quality of the decennial census. The results are expected sometime next year.

The post-enumeration survey for the 2010 census estimated the count was within 0.1 percent of the true U.S. population; in 2000, the agency found it undercounted the population by less than half a percent. In 1990, the agency overcounted the population by more than a percentage point.

Costs of an undercount

The Urban Institute report noted that outside research has shown that census results influence decisions in more than 300 federal programs.

The National Association of Latino Elected and Appointed Officials Education Fund pointed out that the undercount could also hamper efforts to draw fair electoral districts. Arturo Vargas, the group’s CEO, argued the undercount could prevent Latino communities from having representation that reflects their growth.

Between the 2010 and 2020 census, the Hispanic or Latino population grew by about 12 million people, from 50 million to 62 million. 

“Not only do we have the challenge of gerrymanders for race and partisan purposes, that seek to minimize Latino electoral potential, but it’s adding insult to injury that the numbers don’t even take into full account the Latino population,” Vargas said.

He also pointed out that the Biden administration’s efforts to aid recovery from the pandemic relies on data about the most vulnerable populations, which in many cases are the same as the hardest to count.

“Many of those programs are specifically designed to help the most vulnerable households in the country — households that are living below or at the poverty line, that are having difficulty getting access to child care, to job training,” Vargas said. “These are all the same households that probably were least counted in the census.” 

Missing children

The largest undercount in the Urban Institute model was for young children, estimating that last year’s effort missed nearly 5 percent nationwide.

That’s similar to an estimate from demographer William O’Hare, a consultant with the Count All Kids campaign, who found that the number of children missed by the 2020 census may be higher than in 2010. O’Hare’s research found the count may have missed more than 2 percent of all children, based on a Census Bureau estimate of the 2020 population.

That number increases for Hispanic children; O’Hare estimated the 2020 count missed more than 4 percent of them. O’Hare said the limitations of the census dataset prevented him from breaking down the count for younger children.

He also pointed out that the proportion of children who are Black or Hispanic, groups the census has missed in past counts, increased in the past decade.

“The fact that a big chunk of those children had a very high undercount means the number for children is going to be skewed,” O’Hare said.

Research from O’Hare and others have found that young children disproportionately live in hard-to-count households — renters, those living in poverty — contributing to census misses.

The Committee on National Statistics at the National Academies of Sciences, Engineering and Medicine plans to meet later this month and issue two detailed reports on census accuracy.

A report released last month by demographer Connie Citro with the American Statistical Association found that the census may have undercounted the Black population by between 3.2 percent and 7.2 percent, compared with missing 2.3 percent of the population in 2010.

Citro said her analysis, which was conducted independently and used a census estimate of the population in 2020, also lined up with census data showing lower self-response rates in diverse parts of the country. 

“This got me worried that the undercount for disadvantaged groups may be worse in this census than the last few censuses,” she said.

Census experts and researchers consider responses from people who fill out the form themselves more accurate than when completed by an in-person interview by a census worker or counting people through administrative records.

Later this week, the Census Bureau will hold a meeting of its National Advisory Committee on Racial, Ethnic and Other Populations. The agency plans to present updates on the quality of the count and future plans. In past meetings, panel members have raised concerns about potential undercounts among vulnerable populations in last year’s effort.