I am using CPS data from 2004 to 2006, and I would like to generate an individual identifier. I tried to group over (hhrid hhird2 pulineno) but this does not work. Some suggestions?
Thank you for your question. Which variable(s) you should use to uniquely identify records depends on the samples you are using and your purpose for using a unique identifier. Within a single sample (e.g. the December 2021 basic monthly data, or the 2022 ASEC), the combination of SERIAL and PERNUM uniquely identifies person-records. If you are pooling multiple samples (e.g. all BMS data for 2021 or multiple years of ASEC data), you will also need to include sample information such as year and month.
If, instead, your goal is to link the individuals over time, you can use the variable CPSIDP. CPSIDP is an IPUMS-created variable that uniquely identifies individuals across CPS samples. More information about linking households and individuals across CPS samples is available in this working paper.