How is FAMSIZE calculated when unrelated individuals in household?


I would like to create a variable similar to POVERTY in the ACS IPUMS for the ASEC files. It seems like it should be easy using the OFFTOTVAL and FAMSIZE variables in the ASEC files. But it seems like there are cases of unrelated individuals in the same Household that are summed to determine family size. The attached is an excel file with an example of one serial number from the 2013 ASEC. It seems like there should be 3 families here. Two single individuals and one 2-person family. The poverty cutoff variables are consistent with that interpretation. So my question is why are the first 3 grouped into one family for the purposes of determining family size? Or am I mis-interpreting? I have have interpreted these data correctly, will I need to calculate my own family size variable to make my POVERTY variable like the variable in the ACS?


sample.xlsx (8.5 KB)

The conflicting information you are seeing is actually coming from two different ways families are identified in the CPS. The first set of family identification variables: FTYPE, FAMREL, and FAMKIND are all generated by the Census Bureau using their family definitions. These variables should match up with the official poverty measure variables as you noticed. FAMSIZE is created by IPUMS-CPS and corresponds to FAMUNIT.

I hope this helps.