To compute unemployment rates from Ipums-I, should I use perwt as pweights or fweights? Results differ.

I’m computing country-level unemployment rates using EMPSTAT from IPUMS-I. Should I use the PERWT weight as pweights of fweights in Stata? (The variable description sounds as if fweights would be appropriate, but a lot of online sources point to pweights.)

One might think that it doesn’t matter, but for some countries, it does.

Thanks for your time!

Based on what it sounds like you are trying to do, I think you should use the pweight option. In general fweights are used when the “weight” indicates the number of duplicated or collapsed observations. That being said, sample weighting methodology is still a debated area of empirical work. This resource from the Carolina Population Center is a helpful guide and provides general rules of thumb.