Attributing individuals to Census vs IPUMS families in the CPS

Is there a variable in IPUMS-CPS analogous to FAMUNIT, but using the Census family concepts? I don’t see any way to extract such a categorization from FTYPE, FAMKIND, and FAMREL.

Using the Census definitions, is every household member necessarily a member of some family as they are under the IPUMS definitions, or are single individuals (or families not related to the householder) not treated as families? Looking at the historical tables for family income by quintile, I am pretty sure that in the early years there are fewer people in families than households, though I do not know if this is still true.

I’m trying to do income accounting to check whether the individual incomes add up to the income totals and subtotals for families and households (and whether the individual income types add up to INCTOT). For that I need a family classification system and a set of subtotals that use the same rules. I know I could construct IPUMS family income totals myself, but then I don’t have much of a cross-check.

Am I correct in believing that IPUMS-CPS does not contain a variable analogous to FTOTVAL but based on FAMUNIT? I see that there is such a total in IPUMS-USA.

Variables in IPUMS CPS that describe families using Census family units are those which you have already identified: FTYPE, FAMKIND, and FAMREL.

The Census Bureau defines “primary families” as groups of persons related to the head of household, and “primary individuals” as household heads/householders residing without kin. So, they are not necessarily a member of some family under these definitions. For example, under the definitions of FTYPE, there is a code for a nonfamily householder, which includes a householder living alone (a one-person household).

FTOTVAL in IPUMS CPS is taken directly from the family record of the original CPS data and relies on the definition of family unit used by the Census Bureau. You are correct in that IPUMS CPS, unlike IPUMS USA, does not have an analogous variable to FTOTVAL based off of IPUMS derived family unit variables.

Again, as you noted, if you are interested in a particular family unit variable, whether it’s a Census defined variable or an IPUMS defined variable, you could then use the income variables of interest to you to construct a total family income variable. However, I will suggest the idea of offering an analogous family income variable based on FAMUNIT to our IPUMS CPS Team as it sounds like it would be valuable to researchers.

Hi Michelle!
It is not that hard to create from what is already there, but it might be a good idea for y’all to create it anyway, to highlight the (true) fact that your definition makes sense and the Census definition doesn’t. I cannot imagine any policy question where it would be desirable to exclude e.g., a group consisting of a married couple with children, all unrelated to the householder, from the category of “families.”