Hi. I downloaded the 2022 and 2017 five-year ACS data to look at occupational codes by PUMA. However, the 2017 includes 15.7 million records but the 2022 only 5.57 million records. Is there a reason for this?The 2022 five-ear ACS includes 2018-2022 (about a million records for each year).
Am I misunderstanding something? Thanks.
There are 15,721,123 person records in the 2022 5-year ACS. You can check this quickly by tabulating any variable in the online data analysis system for the 2022 5-year ACS, making sure to select “none” under weight options to view the unweighted number of person records in the sample. For example, I tabulated the variable SEX for this sample without weights and see the total number of persons is roughly 15.7 million:
Below is the same table for the 2017 5-year ACS. You can see there are 15,758,937 person records in this sample.
If you have restricted your analytical sample in some way or perhaps accidentally dropped a number of records during data cleaning, that could explain why you see a smaller number of person records than are present in the sample. I do not see any issue with your most recent IPUMS USA extract, which is a rectangular, person-level extract that includes both the 2017 5-year ACS and 2022 5-year ACS.
