PUMAs not listed in crosswalk


I was having an issue where I was getting the wrong estimates for my geography (Philadelphia) based on what is on the Census Bureau’s published tables. I was able to determine that the issue was based on how I was filtering the data.

I was originally filtering based on the PUMAs listed for Philadelphia in the crosswalk table (PA, 3201 - 3211). This was giving me low estimates. But when I instead filtered based on state and county (42 and 0101, respectively), I get the right estimates. However, filtering this second way pulls in PUMAs 4101-4111. I can’t find those PUMAs on the crosswalk table.

This fix works but ulimately I’m trying to do some mapping with this data and want to ensure that I’m getting the right estimates. Any thoughts on where these additional records are coming from would be appreciated. Thanks.


It sounds like this is a side-effect of using the ACS 5-year files. Basically, because the multi-year files are combining several of the single year files and because PUMA boundaries change over time, the resulting combination in your data set can be quite confusing. Since you are seeking estimates for Philadelphia, I’d encourage you to look into the CITY variable. Philadelphia is uniquely identifiable in most years and at the bottom of the Comparability Tab there are many crosswalks available between CITY and PUMA.

I hope this helps.