Issues with inconsistent WKSWORK2 and WKSUNEM2 universes

I’m trying to estimate people who were in the labor force the previous year using the ASEC sample. I’m having the issue that there are a number of people who, for example, are coded as “2” for WKSWORK2 (worked 14-26 weeks last year), but are also coded as “9” in WKSUNEM2 (not in universe). However, the universe for WKSUNEM2 is listed as “Persons aged 15+ who worked between 1 and 51 weeks last year”. Thus, according to the value for WKSWORK2, these people should be in the universe for WKSUNEM2.

Is this just a record error, or I am doing something wrong? Interestingly, this problem only appears in ASEC surveys from 2014-2019 (this issue doesn’t arise in 1962-2013 as far as I’ve found).

Any help would be appreciated. Thank you!

Our IPUMS CPS team is looking into this; in the meantime, I would suggest using/recoding the single week versions of the variables (WKSUNEM1 and WKSWORK1).

Thank you! I’m using the intervalled estimates because I’m trying to compare workers from 1962-2019, and the single week versions are only available for later samples.

I have received confirmation from our IPUMS CPS team that this is an error; these cases should be in universe and have a value of 0. We are working on implementing a fix for this and will deploy it with the next IPUMS CPS release. In the meantime, you can correct the issue in 2014-2019 by recoding WKSUNEM2 to 0 if WKSUNEM1 is equal to 0 and WORKLY is equal to 2. As a thank you for helping us improve IPUMS CPS data, we would love to send you an IPUMS mug. Please email with a mailing address where we can ship this token of our appreciation.

1 Like