I am wondering why there is not an age filter for NCHILD. For ACS PUMS the variable is only reported for those 15 and older.
I assume the question you are referring to in the ACS is the one that asks females age 15-50 whether they have given birth to any children in the last 12 months. This question is captured in the IPUMS USA variable FERTYR.
The IPUMS USA variable NCHILD is one of several IPUMS-constructed family interrelationship variables, created using algorithms that assign individuals to their most likely parent(s) and/or spouse within the household. After linking each household member to their probable parent(s) within the household (if any), the variable NCHILD simply sums up the number of children linked to each parent. You can read more about how we use information including age, sex, marital status, and relationship to household head to create NCHILD and the other family interrelationship variables in this paper by Gorsuch and Williams or on this page on the IPUMS USA site: New Family Interrelationship Pointer Variables in IPUMS USA.
The universe of NCHILD is all persons. The descriptions of MOMRULE and POPRULE list the different rules that are used to create parent-child links, which do include age restrictions in many cases:
We first prioritize links within a household based how clear the relationship is between the two people being linked (using RELATE). The links in the first priority level are direct links, so there are no age restrictions placed on these links. For links in the second priority level, the difference in the age of the “child” half of the link the “parent” half of the link must be 15 to 44 for mothers and 15 to 60 for fathers. For links in the third through fifth level, the same age difference restrictions apply.
Thank you Isabel for your thorough reply. It was actually PAOC from PUMS I was talking about. I think this is an equivalent variable in that it is women with children, so I would think one would want to filter for 15 and older.
No worries about this, I don’t think there are any persons under 15 that are reported having a child using NCHILD. For me, I would want to keep them out of the universe when computing percentages.
Thanks again.
That is a source variable, which has not been harmonized by IPUMS, but is included in the original data from the Census Bureau. The *PAOC source variables (e.g., US2023A_PAOC, available in the 2023 ACS sample) report the presence and age of own children in the household. The universe of these variables seems to be females age 16 and older. IPUMS does not use this particular variable to construct NCHILD.
The universe of NCHILD is all persons. Our rules for linking parents and children do not result in the linking of children to any parent under age 15, due to the age restrictions imposed on some of our links, and due to the age restrictions imposed on who in the household can be the householder.