Two recent data analyses highlight a question researchers have been asking for years: why is it so hard to get an accurate count of the number of children and is the undercount getting larger?

First, a new report from Count All Kids looks at how well—or rather, poorly—children were counted in the 2020 decennial Census, pinpointing the counties with the biggest errors. The undercount of children in decennial census data has been increasing since 1990, when kids and adults were counted with about the same level of accuracy. The undercount tends to be even worse for young children and Black and Hispanic children. In 2020, Count All Kids estimates that kids were undercounted by 2.1 percent nationally, equal to more than 1.5 million children, while adults were slightly overcounted by 0.2 percent. There was wide variation across counties: according to the report, it ranged from a 54 percent undercount to a 27 percent overcount. More than 1 in 7 counties (15 percent) were found to have especially low estimates that undercounted children by 5 percent, 500 children, or both. Nearly half of these counties were in located in just five states: Texas, North Carolina, Georgia, Oklahoma, and Mississippi.

Second, a recent fiscal analysis from Maryland’s Department of Legislative Services found that an additional 110,503 public school students were eligible for free and reduced-priced meals (FARMs) than the state had previously estimated—equal to an undercount of about 1 in 9 students. As part of an effort to update estimates of lower-income students, the state has started using income information from Medicaid enrollment records to identify more children and families who meet the school lunch eligibility thresholds.  (For the 2022-2023 school year, the eligibility limit for free meals for a family of four was 130 percent of the federal poverty line, or under $36,075 annually. The eligibility limit for reduced-price meals was 185 percent of the federal poverty line, or under $51,338 annually.) Most of the undercounted kids live in the state’s rural and suburban counties, while larger cities like Baltimore had more accurate estimates. After incorporating the Medicaid enrollment and income data, the number of students estimated to be eligible for FARMs jumped by 34 percent, and more than half of Maryland’s public school students are now estimated to qualify during the current 2022-2023 school year. However, the utility of this Medicaid data could itself become endangered in the upcoming months as states begin unwinding the continuous coverage protection that has kept millions of children from losing their Medicaid coverage during the pandemic. If children who are still income-eligible are disenrolled for procedural reasons, they could also lose access to their school meals if officials are no longer able to use their Medicaid eligibility and enrollment information to get them enrolled into FARMs in Maryland. And not just in Maryland: 39 states will use Medicaid data to help identify FARM-eligible kids next year.

At the end of the day, our research, funding formulas, and ability to identify children in need are only as good as the data we have to put into them. These two stories suggest that some children are more likely to get left out of the data for a number of reasons: where they live, low family incomes, race and ethnicity, language, limited internet access, housing instability, food insecurity, disability or special health care needs. Counting children more accurately is the first step towards understanding how, why, and which children are getting the services they need, like Medicaid and CHIP, to stay healthy and do well in school and how to best target efforts to help those who may otherwise get left behind, like children who are eligible for Medicaid and CHIP but remain unenrolled.