FTOTVAL for 1-person families in CPS

I’ve been working on an analysis of family income in the CPS and have been using FTOTVAL as my main variable of interest. I have tried to be careful about the difference between the definitions of family from IPUMS and the Census, and my results are roughly the same if I use FTOTVAL or sum INCTOT within FAMUNITs.

However, it was recently pointed out to me that the Census (e.g. https://www.census.gov/programs-surve…) defines a “family” as 2 or more related individuals living together. Yet FTOTVAL is populated for “families” with just one person in them. This makes sense to me as a matter of analytics, but it’s leaving me unsure of what to say when asked how I define “family income”.

Is there a place I can find the exact definition of family that is used to construct FTOTVAL? Is it just the income for families as defined by the Census but with individual income used for 1-person families? Or does the CPS just ask for family income even if people live by themselves and report what they say? Should I switch to my calcualted family income just to be clear about everything?

My question appears to be similar to that asked about the ACS here Does FTOTINC (family income variable) include results for households with a single person for the IPUMS-ACS data?, but my understanding is that FTOTINC in the ACS is created by IPUMS using FAMUNIT while FTOTVAL in the CPS comes straight from the Census.

FTOTVAL reports total income for the respondent’s family as they were reported to the interviewer; it is taken directly from the family record of the original CPS public use microdata samples.

FTYPE reports the Census family type of each person in a household. A one-person household is considered a “nonfamily householder”. In the case of a nonfamily household, FTOTVAL will be equal to INCTOT.

Whether you decide to use FTOTVAL or INCTOT with FAMUNIT to determine household income is up to you, but the results should be similar.