Does anyone know if there is a geography correlation file from PUMAs to Metropolitan Statistical Area ?
You can use Geocorr to generate correspondence files for pretty much any pair of census geographies from the same time period, including PUMAs to metro areas (i.e. core-based statistical areas, which include both metropolitan and micropolitan areas).But in this case, I'd specifically recommend using the linking files available through IPUMS USA's MET2013 variable description here, as long as you're OK using 2013 metro area definitions. That page includes both crosswalks and "PUMA match summaries by MSA," which summarize how well each metro area matches up with its corresponding PUMAs. (We also plan to add a MET2023 variable to IPUMS USA in the next year or two, along with new correspondence files between new and old PUMA definitions.)
Likewise, if you obtain microdata via IPUMS USA, you needn't use a correspondence file at all. You could just include the MET2013 variable in your data request and have metro areas identified for you.
With geocorr the only correspondence that I can find is PUMA2012 to Metro/non metro (2 geographies). I can't seem to make a correspondence file taking PUMA ID to MSA id.
Are you using CBSA as your target geography?
Right, as I mentioned, metro areas are "core-based statistical areas" (i.e., CBSAs, as Glenn says). That's one of the options in Geocorr.
One more question. I need have a single "Non-CBSA" area that corresponds to geographies (census blocks) that fall outside any CBSA. I need the "allocation factor" and "second allocation factor" Can you think of a simple way to get this ? I guess that the geocorr "rural" part of the urban/rural designation might work
Typically in Geocorr, a code of "99999" means "Not in any ____"
In this example, 31.5% of the population of Missouri PUMA 00300 is not located in any CBSA.
Thanks Glenn !