Clarity of SERIAL variable documentation

Reading through the DESCRIPTION for SERIAL, it notes:

SERIAL is an identifying number unique to each household in a given survey month and year. All person records are assigned the same serial number as the household record they follow. A combination of YEAR, MONTH, and SERIAL provides a within-sample unique identifier for every household in IPUMS-CPS; YEAR, SERIAL, and PERNUM uniquely identify every person in the database within sample.

I think this would perhaps benefit from making clear what is meant by “within sample.”

Thanks for the suggestion. The IPUMS CPS team is now reviewing this universe statement for clarity.

To answer your question, each month of the CPS survey is considered a separate sample in IPUMS. Also Basic Monthly surveys are separate samples from ASEC surveys (even though they overlap in March). The samples are all listed on the Sample Selection page. “Within sample” in this context means that these variables uniquely identify a household within each sample. However, since a given household can appear in multiple samples (they can be interviewed up to eight times), these variables do not uniquely identify households across samples. Including YEAR and MONTH automatically makes this combination different for a given household in each of the samples that they appear in. Cross-sample unique identifiers are available in the variables CPSID (household-level) and CPSIDP (person-level). More information on the longitudinal aspect of CPS can be found here.