Generalizability of CPS Cases with County and Metro Identifiers


I am linking CPS 2020 cases aged 18+ to county and metro area data from HUD. I am only able to include those CPS cases with either County FIPS or Metro FIPS identified.
Given the CPS sampling and weighting procedures, is it fair to generalize estimates from this subsample to “all US adults living within metropolitan areas”?

Because not all metro areas are identified in the CPS, I would caution against your proposed generalization. Also, it sounds like you are including identified counties as well as identified metro areas in your subsample. I would direct you to this IPUMS working paper to learn more about some of the quirks of CPS geography: Generalizability of CPS Cases with County and Metro Identifiers.

Thank you Kari,

Can you share the link to that working paper? The hyperlink in your response is to this post’s page.

Sorry about that! I am pasting the correct link here and updating the previous link as well so future readers of this post don’t run into the same issue: