I am using a panel dataset of the CPS October supplement from 2003 - 2013. I want to be sure to get unbiased and efficient standard errors, but I cannot figure out the best way to account for the sampling design of the CPS in Stata. I am very familiar with svyset in Stata, but I can’t find a strata variable in the CPS.
(1) Are there replicate weights for the October supplement? Or, (2) do a primary sampling unit and strata variable exist that can be used to svyset the data?
Unfortunately the CPS does not provide primary sampling unit information and IPUMS has never released non-ASEC replicate weights before. We have documentation on IPUMS provided replicate weights for March ASEC samples, but it seems that you’ll need different replicate weights for a different month’s supplement. An alternative may be to use the replicate weight files provided by NBER. They have dedicated files for October supplement replicate weights, but only for years 2010-2015. They also have documentation and instructions for using non-March ASEC household and person replicate weights.