I noticed that in the 1880 100% sample, it appears that many people coded in the occ1950 scheme as 290 (Managers, Officials, and proprietors (n.e.c) are actually listed in occ 302 as “Keeping House.” This leads to the strange result that 20% of women in 1880 are listed as Managers and Proprietors, instead of “keeping house,” which is obviously wrong. Can this be fixed? I thought I used the 1880 100% sample last year, and it did not have this problem. Thanks!
The IPUMS USA team is currently looking into this. I’ll post here again once we have an answer to your question.
Our historical team assigns the values for OCC and OCC1950 in the 1880 full-count data file using slightly different inputs, which are based on interpretations of occupation strings. This is likely the source of the discrepancy you are seeing. The IPUMS USA team is aware of the discrepancy and will explore if the string assignments for these cases seem appropriate in light of this. As this is likely an error and comparing string assignments in full-count data files is not a quick process, you might choose to reassign the cases with an OCC value of 302 to a different category of OCC1950 (e.g., 980 or 981) if this seems appropriate for your analysis.