Mark Mather
Associate Vice President, U.S. Programs
FIGURE. Risk of Undercounting Young Children and 2020 Census Self-Response Rates in the Washington, DC Area, by Census Tract (June 25, 2020)
Source: PRB analysis of data from the U.S. Census Bureau.
Just 20 counties—mostly located in California, New York, and Texas—account for 41% of all young children living in very high-risk census tracts (see Table 2). A few of these counties stand out because they have both large numbers and shares of children living in very high-risk neighborhoods, in combination with relatively low self-response rates. For example, 84% of children under age 5 in Miami-Dade County, Florida (132,235), live in neighborhoods with a very high risk of undercounting young children, and the average self-response rate in that county in late June was 59%, which is below the average across all large counties (63%). Over two-thirds of young children in Hidalgo County, Texas, live in very high-risk neighborhoods, and the mean self-response rate in that county was 46% in late June.
TABLE 2. 2020 Census Self-Response Rates in 20 Counties With the Largest Numbers of Young Children Living in Very High-Risk Census Tracts
| wdt_ID | Counties | Number of Young Children Living in Very High-Risk Tracts | Percent of Young Children Living in Very High-Risk Tracts | Average Tract-Level Response Rate (%) |
|---|---|---|---|---|
| 1 | All large counties | 4,062,432 | 24.7 | 63.0 |
| 2 | Los Angeles County, CA | 290,389 | 46.5 | 57.8 |
| 3 | Harris County, TX | 140,160 | 39.7 | 55.0 |
| 4 | Miami-Dade County, FL | 132,235 | 84.0 | 58.6 |
| 5 | Cook County, IL | 130,949 | 39.9 | 59.9 |
| 6 | Kings County, NY | 94,935 | 49.0 | 48.8 |
| 7 | Queens County, NY | 91,986 | 63.5 | 51.0 |
| 8 | Bronx County, NY | 87,466 | 82.5 | 52.9 |
| 9 | Dallas County, TX | 86,532 | 42.2 | 57.9 |
| 10 | Broward County, FL | 71,264 | 63.9 | 59.1 |
| 11 | Philadelphia County, PA | 69,185 | 64.2 | 50.9 |
| 12 | Maricopa County, AZ | 55,238 | 19.8 | 62.2 |
| 13 | Hidalgo County, TX | 54,216 | 67.7 | 46.4 |
| 14 | Orange County, CA | 52,528 | 27.8 | 71.0 |
| 15 | Wayne County, MI | 45,665 | 39.6 | 61.9 |
| 16 | San Diego County, CA | 45,340 | 21.4 | 67.5 |
| 17 | San Bernardino County, CA | 44,715 | 28.9 | 59.5 |
| 18 | El Paso County, TX | 42,160 | 64.7 | 58.8 |
| 19 | Santa Clara County, CA | 41,967 | 35.4 | 71.3 |
| 20 | Bexar County, TX | 39,492 | 28.5 | 59.7 |
| 21 | Riverside County, CA | 36,421 | 23.1 | 60.5 |
Note: Data on response rates reflect rates as of June 25, 2020. Large counties are the 689 counties that had at least 5,000 children ages 0 to 4 in the 2010 Census.
Source: PRB analysis of data from the U.S. Census Bureau.
Why does it matter if a neighborhood has a large share of households that have not responded to the census? Low self-response rates could lead to less accurate counts and fewer dollars for communities that need those funds the most. Accurate census data ensure that funding is equitably distributed for numerous programs benefitting children and families, such as the National School Lunch Program and Head Start. Census Bureau data are used to distribute more than $675 billion in federal funds to states and local communities for health, education, housing, and infrastructure programs each year.
Response Rates Are Lower in Neighborhoods With Higher Concentrations of Black, American Indian, and Latinx Children
Self-response rates were lowest in neighborhoods with high concentrations of racial and ethnic minorities in the young child population. The mean self-response rate for all tracts where Blacks make up the majority of young children was 51%, compared with 64% for tracts with a majority of non-Hispanic White children. The average self-response rate was just 21% in tracts with a majority of American Indian/Alaska Native children, which probably reflects the delayed start of the Census Bureau’s Update Leave operation in many rural areas. The average self-response rate for tracts with a majority of Latinx children was also relatively low, at 54%. The mean self-response rate was 62% in neighborhoods with a majority of Asian American children under age 5—similar to the average rate for neighborhoods with a majority of non-Hispanic White children.
These low response rates in communities of color are important because historically, certain racial and ethnic groups have faced a higher risk of being missed in the decennial census. Results from the 2010 Census show that among children under age 5, the net undercount rate was 7.5% for Latinx children and 6.3% for children classified as Black alone or in combination with one or more other races. The net undercount rate for all children under age 5 was 4.6%—higher than any other age group.
Identifying Neighborhoods to Target for Outreach
PRB has developed a series of maps and databases, which are being updated on a weekly basis, to help improve targeting of communities where children are most likely to be missed in the census. These resources highlight census tracts with a very high risk of undercounting young children and low 2020 Census self-response rates.
Interactive maps
Users can zoom in and out of these maps to view patterns in their states and local areas and can click on a census tract to view the tract FIPS code, undercount risk category, 2020 Census self-response rate, and estimated number of children under age 5 in 2014-2018.
The maps are divided into 11 separate files, covering all 50 states and the District of Columbia.
Users can zoom in and out of these maps to view patterns in their states and local areas and can click on a census tract to view the tract FIPS code, undercount risk category, 2020 Census self-response rate, and estimated number of children under age 5 in 2014-2018.
The maps are divided into 11 separate files, covering all 50 states and the District of Columbia.
- AK, AZ, HI, ID, MT, NV, OR, UT, WA.
- TX.
- NJ, NY, PA.
- IN, KY, OH, TN, WV.
- IL, MI, MN, WI.
- FL.
- DC, DE, MD, NC, SC, VA.
- CT, ME, MA, NH, RI, VT.
- CA.
- AR, CO, IA, OK, KS, MO, NE, NM, ND, SD, WY.
- AL, GA, LA, MS.
Each database includes data on the risk of undercounting young children, the latest 2020 Census self-response rates, weekly change in response rates, key predictors of child undercount, and the racial/ethnic composition of the young child population.
- State-County FIPS code.
- State FIPS code.
- State abbreviation.
- State name.
- County FIPS code.
- County name.
- Tract code.
- Undercount of young children risk category.
- 2020 Census self-response rate.
- Weekly change in 2020 Census self-response rate.
- Number of children ages 0 to 4.
- Total population.
- Percent of population that are young children ages 0 to 4.
- Population ages 0 to 4 that is Black Alone.
- Population ages 0 to 4 that is American Indian and Alaska Native Alone.
- Population ages 0 to 4 that is Asian Alone.
- Population ages 0 to 4 that is Native Hawaiian/Other Pacific Islander Alone.
- Population ages 0 to 4 that is Two or More Races.
- Population ages 0 to 4 that is White Alone, Not Hispanic.
- Population ages 0 to 4 that is Hispanic or Latino.
- Percent of children ages 0 to 4 in families with incomes below 100% of poverty.
- Percent of adults ages 18 to 34 with less than a high school diploma.
- Percent of children ages 0 to 17 living in a female-headed household.
- Percent of children ages 0 to 5 living with grandparents.
- Percent of households that are limited-English speaking.
- Percent of children ages 0 to 5 living in immigrant families.
- Percent of population living in renter-occupied housing units.
About These Estimates
The Census Bureau calculates household self-response rates for geographic areas that receive their census invitations in the mail, as well as households in Update Leave areas that receive their census invitation and paper form when a census taker drops off a package of materials at their residence.
Net undercounts represent a balance between two groups. One group is people omitted from the Census. The second group is erroneous enumerations (mostly people counted twice) and whole-person imputations.
The estimated risk of undercount for young children is based on PRB’s analysis of American Community Survey estimates and the U.S. Census Bureau’s Revised 2018 Experimental Demographic Analysis Estimates for young children. Data are based on 2020 Census tract boundaries.
While 2020 Census self-response rates are available for 2020 Census tracts, PRB’s original database on the undercount of children is based on 2010 Census tract boundaries. PRB matched 2010 Census tracts to 2020 Census tracts using a crosswalk file provided by the Census Bureau.
For a detailed description of the methods and data sources used to predict child undercount risk, please refer to William P. O’Hare, Linda A. Jacobsen, Mark Mather, and Alicia Van Orman’s report, Predicting Tract-level Net Undercount Risk for Young Children.
Acknowledgement
This research was funded by The Annie E. Casey Foundation, Inc., and we thank them for their support. The findings and conclusions presented in this report are those of the authors alone and do not necessarily reflect the opinions of the Foundation.
We also thank Dr. William P. O’Hare for all his work on the undercount of children in the census and for providing expert guidance to PRB staff on this project.
If you have any questions, please contact Mark Mather at PRB.
