Which is the lowest level of geography in the ACS PUMS 5 year samples: PUMA or county (when available)?

I see that in the ACS PUMS 5-year data, the countyfips variable is available for some of the counties in some of the states. Are these, when weighted, applicable to the county? In previous questions, the answer appears to be yes:

However, on the Census and IPUMS websites, it still says that PUMAs are the lowest level of geography available:

“The smallest identifiable geographic unit is the PUMA, containing at least 100,000 persons. PUMAs do not cross state boundaries.” https://usa.ipums.org/usa/sampdesc.shtml#us2014c

So which one is correct? Is PUMA or county the lowest level of geography that can be tabulated using the PUMS data? And for what years/samples does this apply?

If it supposed to be applicable to the county level, I believe I’ve found an error when using the perwt variable for Canyon County, Idaho (county fips 027) in the 2014 5 year PUMS for Idaho (state fips 16). I’m finding that I get the correct estimate for national-level and state-level for Idaho, but not for Canyon County. The population estimate I get for Canyon County is 79,070 but it should be closer to 195,353 (based on ACS Table DP05 for 2010-14).

Thank you in advance for your help.

The PUMA variable is the lowest level of geography within the IPUMS data. Regarding the discrepancy you’ve uncovered between the ACS table and the IPUMS data, it seems there is an error in the data. Basically, many of the multi-year samples include two sets of PUMA boundary definitions. This is because the US Census Bureau redraws PUMA boundaries every 10 years based on population information gathered from the most recent decennial census. So, what show up in the IPUMS data at the moment are observations associated with only one of these PUMA boundary definitions. You can check if any given county is identified for the full 5-year sample, by making sure there are observations for each MULTYEAR represented in the sample. We are looking into the resolving this error soon.

We like to reward our users for uncovering errors with a IPUMS mug. If you could please email ipums@umn.edu with your mailing address we will send you a mug.