I’m looking to merge the ACS SPM yearly research files to the 1-year ACS data accessed through IPUMS USA. According to the ACS SPM documentation, it instructs users to merge them to the ACS public use files using the SERIALNO and SPORDER variables. In the ACS IPUMS data, the variables I was able to find were SERIAL (the equivalent to SERALNO, as it uniquely identifies households) and PERSONNO (the equivalent to SPORDER, I believe, as it should uniquely identify the person in each household).
After doing a 1:1 merge on these variables for one wave (year 2022), only half of the total observations were matched. Of those that were matched, it appears the demographic indicators included in both datasets, including age and race, vary between the ACS main file and ACS SPM indicators.
Has anyone else has used the ACS SPM files? Any insight and suggestions would be much appreciated. Thanks in advance!
The variable SERIALNO is a household identifier present in the ACS PUMS data from the Census Bureau, and the ACS SPM research files. The variable is formatted differently in these two data sources from 2016-forward. In the ACS PUMS data from the Census Bureau for 2016-forward, the variable starts with a numeric prefix for the year, such as 2022. It then includes letters indicating whether the record is from a housing unit or group quarters unit (either GQ or HU). The last digits of the serial number are a numeric identifier. In the ACS SPM research files, the SERIALNO variable consists of just the final numeric identifier, and does not include the year prefix or GQ/HU label.
The variable CBSERIAL in IPUMS USA data corresponds to SERIALNO in the ACS PUMS data, with the one modification that it does not include the GQ or HU label. It does include the year prefix.
The IPUMS USA variable CBPERNUM is the same as the original Census Bureau variable SPORDER. We do not have any variable called PERSONNO in IPUMS USA.
You can merge the ACS PUMS from the Census Bureau with the ACS SPM research files by modifying SERIALNO so they are formatted identically in both data files. You can also merge the ACS PUMS and/or the ACS SPM research files onto ACS data from IPUMS USA, again making sure you modify SERIALNO or CBSERIAL appropriately so their formats match. The merge must use both the household serial number (SERIALNO in the PUMS and SPM data and CBSERIAL in the IPUMS data) and the person number (SPORDER in the PUMS and SPM data and CBPERNUM in the IPUMS data). Of course, you’ll need to rename the variables so they have the same name in the files being merged.