What is the best way to analyze household level data in the CPS ASEC?

Hi,

I’m trying to analyze household level data in the CPS ASEC. I need to estimate HH tenure and # of own children. I’m trying to benchmark it to Census’s “Families and Living Arrangement” tables but am way off base.

I would like to create 1 observation for each HH, which variable should I use to do that? Serial? HRHHID? Relate = 101? They all produce different estimates.

Related to this, why are the #s of unique serial #s lower than the # of HH level records in the Samples page of IPUMS website?

Thanks!

The PERNUM variable counts the number of individuals within each household. So, if you keep only observations with PERNUM==1, you will be left with one observation per household.

When you download IPUMS CPS data in the (default) rectangular format, all non-response households will be dropped because they have no person level data available. This is why the number is lower than that listed on the ASEC sample sizes page. If you download IPUMS CPS data in hierarchical format, then the number of households will match the number listed on the ASEC sample sizes page.

…That’s an embarrassingly simple solution, I should have seen that for myself!

I thought it had to do with non-response HH but just needed to make sure.

Thanks so much!