I noticed that starting in 1994, there are respondents who are coded as ELIGORG == 1 but are coded as N.I.U. for EARNWEEK2 (9999.99). I was wondering if this was an error, as the one found for those coded N.I.U. for EARNWEEK (999.99) here in the March 2025 revision:https://cps.ipums.org/cps-action/revisions
Thanks for your inquiry. I had a look at this today but haven’t found the answer yet. I will continue my investigation at the beginning of next week.
I confirmed that if the respondent has a value of $0 for EARNWEEK2 in the source data, they are being coded as NIU. I was able to verify this using the unharmonized version of EARNWEEK2 variable (UH_ERNWK_B5). It appears that we are not consistent in our handling of $0 values between EARNWEEK and EARNWEEK2, and based on my analysis, the $0 in the BMS are valid and should be handled as $0, not NIU. I have made a note with the IPUMS CPS team to investigate this further and to ensure our data and documentation are consistent across the earnings variables in the Outgoing Rotation Group Earner Study.
In the meantime, you may wish to use the unharmonized version of this variable ([UH_ERNWK_B5) for your analysis.