(January 2007) The U.S. Census Bureau plans to eliminate the "foster child" relationship category on its questionnaires for the 2010 Census and the American Community Survey. They will be counted with other children, but foster children's characteristics as a group will not be available.

Census Bureau officials had set a January deadline to approve the content of questions on the 2010 Census short form that goes to every U.S. household. Decisions about the 2010 Census form also affect the American Community Survey (ACS). The census survey is intended to replace the census long form, so the two data sets must be comparable.

Bureau officials said they had to eliminate one of the 15 relationship-to-householder categories because they had adopted a new questionnaire design in hopes of improving response rates and accuracy.1 In the census data collection process, the form designates one person in each household, usually the one who fills out the questionnaire, as the householder. Everyone else must be listed in a category that describes his or her relationship to the householder, such as spouse or boarder. "Foster child," which had been included with roomers and boarders in the 1990 Census, became its own category in the 2000 Census.

The elimination of the foster-child category was among a number of proposals for census-form revisions on the agenda in November at the Bureau's Joint Advisory Committee meeting. The Census Bureau is required to give Congress a list of topics it plans to ask about by April 1, which gives legislators a chance to voice their opinions.

Census officials said they decided to drop the foster child category because it had the fewest responses of any relationship option in the 2000 Census, and because census numbers do not match figures provided by state governments to the U.S. Department of Health and Human Services. Census Bureau figures count about 50 percent to 60 percent of the number of foster children included in the Adoption and Foster Care Analysis and Reporting System (AFCARS).

The 2000 Census 100 percent file shows 334,974 children in the foster child category,2 compared with 544,000 in AFCARS in September 2000.3 The 2005 ACS shows 307,000 children in foster care compared with 513,000 in the 2005 AFCARS report.

The figures provided by states include about 24,000 foster children aged 18 to 20, and census numbers only include foster children under age 18, but that explains only a small part of the difference between the two data sets. In addition, about 100,000 foster children in 2005 lived in group homes and institutions, which the Census Bureau classifies as group quarters.4 The Census Bureau does not identify foster children as such if they live in group quarters and not in households.

The rest of the gap between the two sets of data, experts believe, comes about because census counts miss and misclassify many foster children. Householders may not list foster children on their questionnaires: They often are in foster homes for only a short time, and the census respondent may not feel the foster child is part of the household. Or householders may list foster children in a different category. Because many are in "kinship care" with grandparents, aunts and uncles, or other family members, foster children may be listed as relatives on the census form. In 2005, according to state records, about a quarter of children in foster care (roughly 125,000 children) were in kinship care.5

Census Bureau surveys and the decennial headcount seem to capture a consistent proportion of children in foster care each year. But whether those numbers are representative of the foster child population is less clear. Analysis shows that data captured by the Census Bureau adequately reflects some characteristics of children in foster care, but not others.

In terms of age and gender distribution, 2005 American Community Survey estimates look very similar to state administrative data from AFCARS, especially considering the sampling errors in the census survey that are associated with these small categories. The proportions of foster children aged 15 through 18 in the ACS are slightly lower than the proportions in AFCARS, probably because the census survey does not account for the many teens in foster care who live in group homes.

There is a bigger gap between the two data sources for some racial and ethnic groups. About a third of the foster children counted by states are non-Hispanic black, compared with about a quarter of those in the 2005 American Community Survey. The opposite is true for Hispanics: State figures indicate that less than a fifth of foster children are Hispanic, but the Census Bureau reported that nearly one in four is Hispanic. Some of the discrepancy could be due to the fact that questions about race and Hispanic origin are not worded the same from state to state, in contrast to the uniform wording of census questions.

As the table (PDF: 34KB) shows, in general, the distribution of foster children by state in the American Community Survey looks similar to that in the AFCARS database, despite differing estimates of population size. For 42 states and the District of Columbia, the difference in proportions was less than one percentage point. The eight states with larger differences are California, Indiana, Texas, Florida, Pennsylvania, Illinois, Georgia, and Virginia. Of course, the ACS estimates have large sampling errors, and the difference between the two datasets in part reflects the small proportion of foster children in many states.

Foster children are a population of special concern because they often have been abused or neglected, then placed under government care. Many fall behind in school, and often have a host of other socio-emotional problems. State records on children in foster care focus heavily on their experience in the child welfare system. They have little information on the characteristics that census numbers are supposed to shed light upon, such as foster children's socio-demographic status or living arrangements.

In addition, access to the AFCARS data is not now readily available, aside from a series of key tables that are available on the federal Health and Human Services Department website, because it requires sophisticated programming capability. If the data on foster children are eliminated from the Census and the ACS as the Census Bureau plans, it suggests that we need to do more to make the data from other sources such as AFCARS more accessible.

William P. O'Hare directed the KIDS COUNT program at The Annie E. Casey Foundation from 1993 to 2006.


  1. U.S. Census Bureau, "Content Determination for the 2010 Decennial Census Program," accessed online at www.census.gov, on Jan. 10, 2007.
  2. U.S. Census Bureau, 2000 Census, SF 1, Table PCT15, "Nonrelatives by Household Types," accessed at www.census.gov, on Jan. 10, 2007.
  3. U.S. Department of Health and Human Services, "The AFCARS Report," accessed online at www.hhs.gov, on Jan. 10, 2007.
  4. U.S. Department of Health and Human Services, "The AFCARS Report."
  5. U.S. Department of Health and Human Services, "The AFCARS Report."