Unable to reproduce basic population count in PUMS

I am using the 2017 American Community Survey 1-Year Estimates. I am unable to reproduce the total population shown in American FactFinder for a given area using the PUMS.

For example, I am interested in the total population of Erie County, Pennsylvania. Using the PUMS data and PUMA geographic identifier, I get a total population of 261,758. But when I search for the population using FactFinder (same survey and year) and select the corresponding PUMA, I get a total of 274,541.

Should I be able to reproduce this value using PUMS, and if so, what might be the cause of the discrepancy?

(The PUMAs for Erie County are 101 and 102 according to the MABLE geographic database.)

Hi Michelle,

In general, population totals from PUMS will not exactly match the totals from census summary tables because they have a different sample basis. (The annual ACS PUMS samples are each 1% of the U.S. population, which is less than the complete ACS sample used for summary tables.)

That said, when I derive a population total from IPUMS USA for Erie County, Pennsylvania, I get 275,475, which is much closer to the FactFinder number than what you produced.

I get the same population total whether I use the PUMA variable, or more simply, the COUNTYFIP variable (with COUNTYFIP = 49 for Erie County and STATEFIP = 42 for Pennsylvania).

Are you sure you’re using the correct state code to identify Pennsylvania? I’m not sure what else you might be doing differently.

Ok. Thank you for your help. I am using the correct state code; I’ll review my code again and see if I overlooked an error there.