Replicating family poverty estimates


I am trying to replicate the number of families in poverty from CPS 2018 published in Table 4 here: I am not able to replicate the number of families (with our without children) in the universe, however (83,508,000). I am using the ASEC family weights (asecfwt) and have limited my sample to householders of households that contain a family member related by birth, marriage, or adoption. My number underestimates the number of families in the universe, however. Here is my Stata code.

recode relate (201 301 501 701 901 1001 = 1) (101 1114 1115 1241 1242 1260 = 0), gen(fammember)
bysort serial: egen family2 = max(fammember)
gen familyhhh2 = .
replace familyhhh2 = 1 if relate==101 & family2==1
svyset [pweight=asecfwt]
recode offpov (99 = .), gen(offpov2)
svy, subpop(familyhhh2): tab offpov2 //n=83,103,090

Results are identical if I do:

gen familyhh33 = .
replace familyhhh3 = 1 if relate==101 & ftype==1
svy, subpop(familyhhh3): tab offpov2 //n=83,103,090

Any ideas what families I am excluding?


I think the issue is with your sample. The 2018 published numbers come from the 2019 ASEC. This is because most ASEC questions ask about the previous calendar year. Re-run the numbers with 2019 ASEC and see if it replicates the Census publication.