How the Census Undercounts Will Impact CED Work

Published for Community and Economic Development (CED) on April 19, 2022.

The U.S. Census Bureau recently released its coverage estimates for the 2020 Census. While the Census was seen as a success – at least in terms of not over- or undercounting the national population, when researchers examined census results for specific groups, the results were different. For the next decade, CED professionals will need to know how these differences affect the $1.5 trillion dollars in federal aid allocated using these numbers.

When there are undercounts, the allocation of federal aid is smaller than needed by the affected group. This translates into less census-based funding provided for Medicaid, Free and Reduced Price Lunch, Supplemental Nutrition Assistance Program and many other CED programs than service levels will likely be. As part of its Census 2020 Counts Program, a 2017 brief by the George Washington University Institute of Public Policy lists the largest 16 assistance programs impacted by Census counts. The Program’s website provides an Excel version of the entire list of 316 federal programs affected. It also has estimates of impacted funds specifically for North Carolina – although the figures are several years old, it provides some context for what we should expect in terms of the forthcoming data based on 2020 census data.

Coverage estimates refer to estimates of how much error may be in Census data. They are obtained by using two different statistical methods, each using sample counts obtained independently of the formal Census count, to double-check its accuracy. In effect, this process might be thought of as checking your measurements for building a table using different yardsticks to ensure you have a consistent and correct number before you start sawing.

Understanding under- and overcounts is much more important than building a table, of course. It is critical for planning for CED services. We should note that in terms of counts, we focus on net under- or overcounts. For example, there can be undercounts in some areas, and overcounts in others. The Census focuses on the net over- or undercount nationally.

The groups with a statistically significant undercount in 2020 included Black or African Americans, American Indian or Alaska Natives, and Hispanic or Latino populations. Children under five were also statistically significantly undercounted Whites and Non-Hispanics were statistically significantly over-counted.

Being statistically significant means the under- or overcount was large enough that it is very unlikely the under- or overcount could be the result of simple random error. In other words, there was not a clear random pattern of over- and undercounts across of the country with these groups, but instead a systematic pattern when compared to the quality control alternatively measured samples.

Historically, while not desirable, under- and overcounts are typical. But CED professionals in North Carolina should pay particular attention to the percentage undercount of Hispanics. This group was undercounted by almost 5 percent, three times greater than in the 2010 Census. The 2020 percent undercounted for Blacks was a little over 3 percent, and the percent undercount of Native Americans living on reservations was about 5.5 percent. These latter two undercounts are about the same percentages as in 2010. In addition, researchers are also concerned about the continuing undercount of children under five.

Look for state level data to be released this summer.

Topics - Local and State Government