I understand that I should not expect IPUMS USA online query results to match exactly what I would find in similar Census ACS data tables, but I am finding discrepancies far outside the margins of error and I’m trying to understand this. I am starting with Table B05006, Place of Birth for Foreign-Born, and limiting to Rhode Island (statefip=44, and using perwt). While the totals do not seem wildly far apart (162,413 +/-4,078) in B05006 vs 173,412 in IPUMS), though latter indeed outside MOE, the distribution by continent/region does seem wildly different. So, for Africa, the B05006 count in 19,814 +/-1,952, while the IPUMS count is 12,474 or 12,566. In both cases I am using ACS 5-year estimates for 2019-2023. Any ideas on how to explain this sort of difference?
For anyone interested, I think I solved the problem. Although B05006 include Capo Verde, IPUMS USA classifies it under Atlantic Islands rather than as part of Africa, so we have added 16020 in and now the total is probably within MOE range.
I’m glad to hear that you were able to figure out the source of this discrepancy! I’ve shared a note with my colleagues to look into moving the BPL code for Cape Verde to (West) Africa for better consistency.