Does anybody know how to construct a family file with ACS data: HH variables and person data of all members of the HH in one record?
The best way to achieve this data structure is to create a rectangular extract in the extract system and then reshape it wide. Here is documentation on reshaping data in Stata and reshaping data in R.
A rectangular extract includes only person records, but you can add household variables to these extracts, and they will be included on all the person records in each household. Households are uniquely identified within samples using the SERIAL variable. Persons are uniquely identified within samples using a combination of SERIAL and PERNUM.
Since the household variables have the same values for every household member, these variables do not need to be indexed. Only the person variables (e.g, age, sex, employment status) need to be indexed for each person. You can use PERNUM to index these variables within each household, identified by SERIAL.
Your data file would go from having a structure like this, where each person has their own record:
To having a structure like this, where each household has its own record, with person variables included, indexed by PERNUM:
Please note that the concept of a family is distinct from the concept of a household. The guidance I provided will allow you to create a household-level extract with person-level information included. For households that include more than one family, this process would look slightly different if you wanted to create a family-level extract rather than a household-level extract. Please review our overview of subfamilies and family interrelationships, and feel free to reply here in the forum if you have questions on applying this basic idea of reshaping to households with more than one family.
Thank you very much Isabel. Will let you know how it works.
Oleh
In a message dated 6/11/2026 12:24:41 PM Eastern Daylight Time, notifications@ipums.discoursemail.com writes: