Potential Problems in Measuring Racial Change Among Young Children Using Data from the American Community Survey

Dr. Bill O'Hare

The data from the 2020 Census show the diversity explosion in the U.S. is continuing (Frey 2015).  As the racial and ethnic diversity of the U.S. population has increased, collecting, tabulating, and reporting data by race has become more complex (Prewitt 2013).

There is no doubt the that the diversity explosion we have seen in the U.S. over the past few decades is more advanced in the youngest population. The percent of each five-year age group who are persons of color (anyone not Non-Hispanic White Alone) is shown in Figure 1.  This figure clearly shows that persons of color are much more apparent in younger cohorts than in older ones.  The percentage of young children (ages 0 to 4) who are persons of color (52 percent) is more than twice as high as any cohort older than 70 years old.

Since racial and ethnic[1] diversity is more advanced for young children, changes in how race and ethnicity are measured have a bigger impact on this age group. This article focuses on the potential problems in tracking racial changes using the American Community Survey (ACS).  The race and Hispanic Origin questions used in the 2021 American Community Survey are shown below.

[1] Following federal government usage, the term ethnicity is used here to denote Hispanic Origin status.

The ACS has become one of the most widely used data sources for tracking characteristics and well-being of young children (Gutierrez 2022; O’Hare 2022).  However, measuring racial change using the ACS data has recently become problematic.  The new method the Census Bureau used in the 2020 Census to classify people by race and ethnicity is now being used in the ACS (U.S. Census Bureau 2021a).  Changes in the way data for race and Hispanic Origin status are now being coded in the 2020 ACS make the race data after 2020 inconsistent with the data before 2020.

Table 1 shows the distribution of young children (ages 0 to 4) by race and Hispanic Origin status in the 2019 and 2021 ACS.  Some groups show enormous (unrealistic) change.  For example, the  ACS data show the Hispanic White Alone population ages 0 to 4 decreased by 72 percent between 2019 and 2021.  At the same time, some groups showed enormous increases between 2019 and 2021.  The Non-Hispanic Some Other Race group more than doubled—up 102 percent.  The Hispanic Two Major Races group increased by a whopping 411 percent.

These kinds of huge changes over a two-year time period are highly improbably and these data are unlikely to reflect real demographic changes.  The big changes shown in Table 1 are more likely due to changes in how race and ethnicity are measured by the Census Bureau and reported by respondents.  However, since the 2021 race and ethnicity labels are the same as those in 2019, it would be easy for someone to think the data are comparable.

According to the Census Bureau (2021c), “The improvements made to the 2020 ACS race question design, processing, and coding are similar to changes made in the 2020 Census, which are presented in the blog entitled, ‘Improvements to the 2020 Census Race and Hispanic Origin Question Designs, Data Processing, and Coding Procedures’.

Beginning in 2020, the Census Bureau implemented changes to the race question based on extensive research and outreach over the past decade. For more information, please see 2010 Census Race and Hispanic Origin Alternative Questionnaire Experiment and the 2015 National Content Test.

Six key changes in the method used to categorize people by race and ethnicity in the 2020 Census (compared to the 2010 Census) are explained by the Census Bureau (2021a. no page number):

      1. In response to community feedback over the past decade, we added dedicated write-in response areas and examples for the “White” and the “Black or African Am.” racial categories.
      2. We provided six example groups for each of the “White,” “Black or African American,” and “American Indian or Alaska Native” racial categories. These examples represent the largest population groups within each of the geographically diverse population groups of each race category, as defined by the 1997 OMB standards.
      3. Based on successful previous testing, the term “Negro” was removed from the 2020 Census by updating the category “Black, African Am., or Negro” to “Black or African Am.” on paper questionnaires and “Black or African American” on electronic instruments.
      4. We reordered detailed Asian and Native Hawaiian or Other Pacific Islander checkboxes by population size.
      5. We changed the checkbox category “Guamanian or Chamorro” to “Chamorro” based on research and positive stakeholder feedback.

We updated the write-in instructions for the “Some Other Race” category to better solicit detailed reporting. The 2010 Census form included the instruction to “Print race,” but we updated the 2020 Census instruction to read “Print race or origin” to correspond with the overall question instruction to “Mark one or more boxes AND print origins

One of the biggest changes is how write-in responses are managed.  Before 2020 only the first 30 characters of write-in responses were analyzed. In 2020, the Census Bureau started analyzing the first 200 characters of the write-in response.   This led to a lot more people being recognized as having more than one racethus the big increase in people with two or more races or three or more races.

More information about this issue is provided by the U.S. Census Bureau in the documents listed in Box 1. These resources can help the reader find key material about race/ethnicity data in the 2020 Census as well as the ACS which will increase understanding of the complexities of race/ethnicity data today.

The U.S. Census Bureau (2021c) cautions users about comparing post-2020 ACS and pre-2020 ACS data this way.

“Differences between 2021 ACS and 2019 ACS may be the result of demographic changes, and/or differences in question wording (the ACS question on race was revised in 2020 to make it consistent with the 2020 Census race question), processing, coding updates, or methodological differences in the population estimates used as controls.  For more information, see the ACS Race User Note: Improvements to the Race Question.”

As far as I can tell, there is no crosswalk file that would allow one to gain a better understanding of how much of the change between 2019 and 2021 is due to changes in methodology compared to real demographic changes.

The changing methodology for categorizing people by race and ethnicity also raises questions about comparability of characteristics over time.  For example, is the poverty rate for Non-Hispanic White Alone young children in 2019 comparable to the one in 2021?  Table 1 shows this population decreased by 72 percent between 2019 and 2021.   It is difficult to believe that a change in population of the magnitude seen in Table 1 did not introduce some changes related to poverty status.

On one hand, the new way of measuring race and Hispanic Origin probably provides a more accurate picture of how people see themselves in our society (U.S. Census Bureau 2021b).  The new approach to measuring race and Hispanic Origin, undoubtedly reveals a level of complexity that the old method did not reveal.  On the other hand, the changes make it difficult to reliably assess recent changes in race and Hispanic Origin status using the ACS.

The key point I want to make here is that one should not compare the racial/ethnic distribution of children from post-2020 ACS with the distribution prior to 2020.  The changes are likely to reflect new methodology more than real demographic changes in the race and ethnicity of young children.

References

Bass. F. (2022). “Racial Ethnic Changes According to the Newly Released 2021 ACS 1-Year Estimates,” https://www.socialexplorer.com/blog/post/racial-ethnic-changes-according-to-the-newly-released-2021-acs-1-year-estimates-12939

Frey, W. (2015). Diversity Explosion: How New Racial Demographics Are Remaking America, Brookings Institution, Washington DC. Diversity Explosion: How New Racial Demographics are Remaking America: Frey, William H.: 9780815726494: Amazon.com: Books

Gutierrez, F. (2022). ”Use of ACS in Child Advocacy,”  presentation at the 2022 Applied Demography Conference

O’Hare, W. P. (2022). Use of the American Community Survey Data by State Child Advocacy Organizations – Count All Kids, Posted on Count All Kids website https://countallkids.org/resources/use-of-the-american-community-survey-data-by-state-child-advocacy-organizations/

Prewitt, K. (2013). What Is Your Race? The Census and Our Flawed Efforts to Classify Americans, Princeton University Press,

U.S. Census Bureau (2021a). “Improvements to the 2020 Census Race and Hispanic Origin Question Designs, Data Processing, and Coding Procedures,” Rachel Marks, Merary Rios-Vargas. U.S. Census Bureau, August 3, https://www.census.gov/newsroom/blogs/random-samplings/2021/08/improvements-to-2020-census-race-hispanic-origin-question-designs.html#:~:text=Based%20on%20further%20research%2C%20testing,recognize%20longer%20write%2Din%20responses.

U.S. Census Bureau (2021b). ”2020 Census Illuminates Racial and Ethnic Composition of the Country.”  Nicholas Jones, Rachel Marks, Roberto Ramirez, Merary Rios-Vargas,  August 12, https://www.census.gov/library/stories/2021/08/improved-race-ethnicity-measures-reveal-united-states-population-much-more-multiracial.html

U.S. Census Burearu (2021c). “ Improvements to the Race Question,”  U.S. Census Bureau, November  https://www.census.gov/programs-surveys/acs/technical-documentation/user-notes/2021-03.html