ATUS childcare for own children not living in the household

Hi all,

I have a couple of questions about ATUS measures for childcare for own children who do not live in the household. The questions relate to two observations: 1) The number of people who report having own children who do not live in the household (NONHHCHILD) is very low (~1%), and 2) The number of time episodes of primary and secondary childcare with own non-household children is also very low (<1% individuals report them).

What are the criteria used to define when a child is “own” vs “not-own” and when a child is “household” vs “not-household”? The statistic that only 1% of respondents have own children under 18 outside of the household does not seem credible, given separation and divorce rates. How are children who live part time with one parent classified? Do you have any insight as to why the prevalence in NONHHCHILD is so low?

I read in the documentation that respondents who report own children living outside of the household, the children are added to the roster so that they can be used in “who” records. If they were added as “household children” that could create errors (and the low prevalence of childcare for own non-household kids), but presumably they are added as “own non-household children under 18,” correct? Also, even if there was an error in classifying children as household or non-household in “who” records, this would not explain the low prevalence of just having own children under 18 outside of the household (NONHHCHILD).

Thank you!


Activities can be done with household members and people who are not part of the household. For example, RELATEW codes include spouse, parent, own household child, own non-household child under 18, friends, and coworkers. You are correct that own non-household children under 18 are given line numbers in the household and added to the household roster (see LINENOW values for activities with RELATEW=300 for evidence of this practice in the data). These individuals are identified in RELATE code 040. All households in which the time diary respondent reported any activities with a non-household child (i.e., there are any activities with RELATEW code 300) include a non-household child on the household roster.

Information on the presence of own non-household children is collected directly in the ATUS; the variable NONHHCHILD is not constructed by IPUMS. From the ATUS questionnaire, I can see that the definition of a non-household child is simply “a respondent’s child under the age of 18 who does not live with the respondent.”

The presence of own household children is determined based on the household roster. The household roster is initially constructed during the CPS interview, but is reviewed and confirmed, and edited if needed, at the time of the ATUS. You can see more detail on this on pages 8-9 of the questionnaire linked above.

In terms of defining whether a child is one’s own or not, I do not see any distinction made between biological, step, and adoptive children in the questionnaire. I believe it is safe to assume that relationships are self-reported and open to individual respondents’ interpretation.

To look into your concern about the low share of ATUS respondents with an own child under 18 outside of their household, I looked at the prevalence of own non-household children by marital status and age. You referenced high divorce and separation rates, and I see that about 11 percent of people in the 2003-2021 ATUS are divorced or separated. Among divorced individuals, 3.25 percent have a non-household child, and among separated individuals, 4.91 percent have a non-household child. These percentages are higher among younger individuals, who mechanically—based on their own age—have younger children on average. For example, among those under age 50, 6.88 percent of separated individuals and 6.87 percent of divorced individuals have a non-household child.

Thank you, Isabel! This is very helpful. The prevalence still seems low to me, but I will keep investigating. I will ask BLS if they have any other information about criteria to differentiate household vs non-household children or about the NONHHCHILD variable. Thank you!