Where to look for local stories about Census undercounts

A roundup of story angles plus data you can use to investigate the places you cover

This article was developed by Irene Casado Sanchez, Sophia Chen, Kavish Harjai, Yoni Lerner, Melissa Newcomb, Camryn Pak, and Syler Peralta-Ramos as part of Stanford’s Big Local Journalism class, in coordination with Angel Kastanis of the Associated Press and Eric Sagara of Big Local News.

Experts say 81% of children 5 years old and younger across the U.S. live in neighborhoods that were likely to be undercounted in the 2020 Census, leaving those communities at risk of missing out on millions of dollars for vital early childhood education and nutrition services.

Millions of already underrepresented communities could also lose political clout as states redraw Congressional district boundaries. The Black population alone was undercounted by about 2.5 percent, about a million people, according to estimates from the Urban Institute. College towns across the U.S., where students are vital to local jobs and businesses, are challenging the Census numbers, too, saying remote education during pandemic “caused miscalculations.”

How do undercounts affect communities?

Here’s one example: Neighborhood organizers reported that in 2010, an undercount of about 7,000 people cost North Minneapolis two city council members and hundreds of millions of dollars in federal funding over the next decade.

The Census aims to correctly count the United States population to proportionately disperse federal funds and allot political representation. In reality, the Census often overcounts and undercounts the population, failing especially to account for young children, the highly mobile, racial and ethnic minorities, non-English speakers, those experiencing homelessness, low-income persons, and those living in multi-family housing. The Trump administration’s politicization of the Census and COVID-19 further exacerbated the 2020 Census miscounts.

Investigating the differences between Census data and other databases can help us understand which populations have been miscounted, and better report on inequities in funding and political representation. This review of coverage and data provides an overview to help local journalists identify on-the-ground stories of the communities missed by the 2020 Census.

Some of the story angles you might explore

Potential Latino undercount:

  • According to NBC News: “Research by demographer William O’Hare, an adviser to the organization Count All Kids, suggests that the bureau may have missed 832,000 Latino youths, an increase in the number missed in the 2010 census.”
  • According to the AP, in Arizona: “The overwhelmingly Hispanic community has grown enough over the last decade that it’s also building a new elementary school […] But the Census Bureau says Somerton actually lost 90 residents during that time, putting its official population at 14,197 people, not the 20,000 that the mayor expected.”

Potential undercount of Black Americans:

  • According to The Washington Post: “Two new analyses suggest the 2020 Census may have undercounted Black people at a significantly higher rate than usual, raising concerns about whether minority communities could lose out on fair representation and funding over the next 10 years.”

College towns challenge Census data:

  • According to Newsweek: “Some college towns are claiming they were undercounted during the 2020 census and plan to challenge its results, saying the data could result in a loss of federal money and prestige […] The municipalities blamed the pandemic in part for the undercount, saying the loss of students in March 2020 to their hometowns caused miscalculations during the census, which started around the same time.”
  • There is an exception according to Opelika-Auburn News: “Auburn is the rare college town not challenging the census […] Officials said a higher-than-projected number of residents could be an indicator that Auburn University is not the sole attractor for people to the city as population continues to grow. Final numbers exceeded the U.S. Census Bureau’s projection by almost 8,000 people.”

Local authorities and demographers challenge the Census data:

  • According to Boston.com: “Boston is joining other communities across the U.S. with large numbers of university students in planning to challenge the results of the once-a-decade head count, saying the 2020 census undercounted the city’s students as well as jail inmates and foreign-born residents.”
  • According to The Dallas Morning News: “Several Dallas council members on Wednesday [Oct. 20, 2021] said they’re concerned that despite the 2020 census results showing the city’s population increasing, there could be many who weren’t counted at all […] The city doesn’t have an estimate of how many people weren’t included in the latest tally.” There are major political implications: “Dallas is in the beginning stages of its redistricting process, which uses the census numbers to redraw the city’s 14 council districts to allow equal voter representation through evenly populated areas.”
  • An update this year from Pew Research reported about challenges from many cities and states. The bureau’s Count Question Resolution program will accept complaints through June 2023: “Many cities and states say the count wildly underestimated their residents, costing them significant federal and state money for the social services and infrastructure their areas need. […] The bureau’s Count Question Resolution program allows tribal, state and local governments to ask the Census Bureau to review their counts for errors. It already has drawn complaints from 20 places in Arkansas, Georgia, Iowa, Kansas, Mississippi, Nebraska, New York, Ohio, Tennessee, Washington and Wisconsin as well as Puerto Rico.”

News about the Urban Institute analysis:

  • The AP summarizes, saying that undercounts were not as bad as predicted: “The 2020 census missed an estimated 1.6 million people, but given hurdles posed by the pandemic and natural disasters, the undercount was smaller than expected, according to an analysis by a think tank that did computer simulations of the nation’s head count…. The analysis estimates there was a 0.5% undercount of the nation’s population during the 2020 census. If that modeled estimate holds true, it would be greater than the 0.01% undercount in the 2010 census but in the same range as the 0.49% undercount in the 2000 census.”
  • Bloomberg reports: “A new simulation conducted by the Urban Institute suggests that the worst predictions may have been too gloomy. Their model shows an overall undercount of 0.5% of the U.S. population. That would make the 2020 census less accurate on net than the 2010 count—and in some places, far less accurate. But the model does not line up with last year’s worst-case scenarios.”
  • NPR focuses on higher undercounts among people of color: “Last year’s approximately $14.2 billion census likely undercounted people of color at higher rates than those of the previous once-a-decade tally, an Urban Institute study involving simulated census results released Tuesday suggests.”
  • The Atlanta Journal-Constitution pulls out local numbers: “The 2020 census undercounted Georgia’s population by just over 1%, or 124,438 people, including 58,098 in the Atlanta-Sandy Springs-Roswell region, according to estimates released Tuesday by the Urban Institute, a left-leaning nonprofit research organization.”

Data you can use to analyze potential undercounts

We’ve collected three kinds of data into a Big Local News project called “census undercount”, which you can use to investigate the likelihood of undercounts in the areas you cover.

You can access this data with a free Big Local account. To download the CSV files, click on the download button to the left of the dataset. The Python and R scripts used to join the datasets, as well as relevant data dictionaries, can be found in the project.

Hard to count data

The “hard to count” files have data on factors that have historically resulted in census undercounts or overcounts, such as access to a computer, percentage of minority population, high income, low income, etc. From these factors, the census estimates a percentage of the population that is hard to count at the county level. You could think of this as a litmus test for information gathered from other data, pointing to where and why a miscount is occurring.

We joined the “hard to count” data with the census data using R so that `percent hard to count` can be viewed alongside census counts to answer the question of whether or not an undercount was predicted, and if so what might be the cause.

IRS data and school enrollment data

The project also contains four datasets about minority groups: “2019-2020 Asian Enrollment by County,” “2019-2020 Black Enrollment by County,” “2019-2020 Hispanic Enrollment by County” and “2019-2020 Native American Alaska Native Enrollment by County.” Each compares census data about the under-18 population with school enrollment data for that minority group.

Each dataset is organized by its final column, which shows the difference between enrollment percentage and census percentage for each minority community. This difference is shown as an absolute value, so there are no negative numbers.

The counties with the biggest percentage gaps are at the top — these counties are where we might suspect a likely undercount.

Data on residential building permits

Data on the number of residential building units can be consulted by county, just like census data. By analyzing the number of building permits over time, we can check whether a change in the number of residential units is in line with population counts from the census. When it does not match, this may be a clue to an inconsistency in the census data.

If you have questions about a story you’re working on, our free peer data review program is here to help.

Programs like these are part of the OpenNews COVID-19 community care package. If you’re using this story recipe, please let us know — we’d love to promote your work! If you’ve got a story recipe idea, we’d love to hear about it. Drop us a line at source@opennews.org.


Current page