Non-sensible family relationships

I have restricted the ACS 2019 5-year sample to ages 15-19 and am attempting to identify households where the respondent is a biological child, and where possible the biological mother’s age at the respondent’s birth.

Some oddities make me question whether I am doing this correctly:
acs[RELATED == 101 & MOMLOC == 2], there are hundreds of respondents who are the household head (RELATED == 101), yet the mother is reported the household head’s spouse (MOMLOC == 2).

acs[RELATED == 301 & MOMLOC == 1], there are hundreds of respondents whose mother’s would have given birth to them above the age of 50 or below the age of 14 (MOM_AGE - AGE).

Are these expected patterns?


f RELATED=101 and MOMLOC=2, that means the individual’s mother has PERNUM=2. PERNUM=2 is the spouse if the householder is married, but if they are not married then the individual with PERNUM=2 can have any value of RELATED (except 101).

For cases where the mother is out of the expected range, in most cases these are the ages reported by the respondent. You can tell whether the mother’s age was allocated using the variable QAGE_MOM (or QAGE on the mother’s record). If you look only at the allocated (QAGE_MOM=4), you will see no cases above 50, and very few below 12, which aligns closely to the information described in the editing procedure for AGE. As for the few cases where the age difference is less than 12, there is this note on the editing procedure page:

The following situations will cause AGE or RELATE to be replaced with a missing value and then re-allocated: A child is not at least 12 years younger than the reference person or spouse and MARST is “Never married” - AGE will be re-allocated

I suspect that in these cases, the reference person’s spouse is an appropriate age to be the child’s parent, which means the ages will not be re-allocated.

Thank you. My understanding was that RELATED = 101 means they are the household head, and MOMLOC = 2 implies the mother is married to the household head. Otherwise could you clarify what a respondent with RELATED = 101 and MOMLOC = 2 means for the respondant?

Effectively I am trying to get mother figure’s age, but using MOM_AGE get many implausible values. It sounds like however using that in combination with QAGE_MOM should solve the issue.

MOMLOC = 2 does not imply the mother is married to the household head. It just means the individual’s mother is the second person listed on the ACS questionnaire. While this is often the householder’s spouse, when the householder is unmarried the second person listed can have any relationship to the householder.

I don’t think QAGE_MOM will solve your issue. You don’t want to only include those with QAGE_MOM = 4, because those are the cases with imputed age values for the mothers. I only highlighted this case, because these are the only cases that whose ages are forced to have a difference between 12 and 50 between the mother and the child. The non-imputed cases come directly from the questionnaire form. Implausible age values will often come from the respondents not fully understanding the form, or errors in recording their information. Unfortunately we don’t have more information on why the age difference is implausible in any given case. You’ll need to decide how to treat these cases in your analysis.