For MET2013, how does IPUMS determine the omission and commission errors as population percentages rather than as area percentages? Is this process replicable using publicly available data?
All 2013 metropolitan areas correspond to sets of counties.
All 2010 PUMAs correspond to sets of counties and/or census tracts, and all census tracts nest within counties.
Determining tract-to-county relationships is simple because complete tract identifiers include codes for the state and county where the tract lies.
To produce a population crosswalk between 2013 MSAs and 2010 PUMAs, we first join the 2013 MSA delineation file to a 2010 tract population file (from IPUMS NHGIS)–linking by county codes–and then join to the 2010 tract-to-PUMA relationship file.
We can then sum the tract populations for MSAs and for PUMAs to get numerators and denominators for the omission and commission errors.
(We use the same process to determine which PUMAs to associate with which MSAs, associating a PUMA with an MSA if a majority of the PUMA’s population is in the MSA.)
Creating a crosswalk between 2000 PUMAs and 2013 MSAs is somewhat more complicated. Let me know if you’d like to know how we do that, too.