Too Many Parents of Householder (identified via RELATE)

Hi all - I’m using the ACS 2018 5-year sample to parse out some different living arrangements for children, and have come up against a question I’m not sure how to answer.

Using the detailed relationship to householder measure (related), I calculated the number of people in the household who are identified as the householder’s parents (related==501; by year and serial). I expected this new household-level measure to have values of 0, 1, or 2. However, for a very small number of cases (~200), there are 3 or 4 parents of the householder present. Using momloc, momloc2, poploc, and poploc2, I can see that no more than two “parents” per household are actually linked to the householder (sometimes fewer). So who might these third and fourth parents be? Demographics make them look feasibly parental, but if they were step-parents, they would have been identified through the mom/pop pointer variables. If they were parents-in-law mistakenly coded on relate, they would have momloc/poploc values pointing to the householder’s spouse, where present. Any other thoughts on a relationship I might be overlooking?


RELATE reports information provided by the respondent about people residing in the household. In contrast, MOMLOC(2) and POPLOC(2) are family interrelationship variables created by IPUMS to identify the probably parent(s) of individuals; we do not link more than two parents to any individual. In the situation you describe, our algorithm selects among candidate parents to choose the most likely parent(s). The second mother’s location (MOMLOC2) and second father’s location (POPLOC2) are for identifying same-sex parents, not for persons who have more than two (candidate) parents.

Parents-in-law or step-parents are both good ideas that would not be captured by IPUMS family interrelationship variables. Regarding parents-in-law, IPUMS USA does not assign a second set of “parents” who look like in-laws as the spouse’s parents via MOMLOC/POPLOC. Step-parents are a bit more complicated. A RELATE code of step-child can be used to link a coresident child to the household and their spouse (who is presumably the child’s biological parent), but this level of clarity around step-parents in the IPUMS USA RELATE codes in recent years is unique. IPUMS family interrelationship variables search for a logical parent based on a number of criteria (RELATE value, age difference marital status, household composition) and will assign that person’s spouse as the other parent. For the case of step-parents you describe, the parent who looks most like the householder’s mother and her spouse will be assigned (though this person may actually be a step-parent to the householder). Here is a quick summary of the family interrelationship variables in IPUMS USA.