It sounds like you are using the FOODSTAMP variable from IPUMS ATUS, which reports if anyone in the household received SNAP benefits (food stamps) in the last 30 days; please correct me if I am wrong. I will provide information on the out of universe cases for the IPUMS ATUS variable FOODSTAMP and share some information about questions about SNAP benefits in CPS for reference.
As noted on the description tab for FOODSTAMP, this variable is collected as part of the Eating and Health module of the ATUS. Respondents that are not in households selected to complete the Eating and Health module will be out of universe for this variable. You can learn more about the Eating and Health module on the IPUMS ATUS sample documentation page or on the Bureau of Labor Statistics’s page about the Eating and Health module data. The Eating and Health module data dictionary notes that persons who did not complete this module are those who ended the ATUS interview before reaching this section or who ended the interview during this section (see page 4).
The IPUMS ATUS variable differs from the information about food stamps collected from the CPS in a number of ways. First, the IPUMS CPS variable FOODSTAMP reports the family market value of food stamps and is collected as part of the Annual Social and Economic Supplement fielded in March. A similarly named IPUMS CPS variable, FOODSTMP reports if one or more members of the household received SNAP benefits (food stamps) in the prior year (as opposed to the last 30 days reference period used in IPUMS ATUS). The Food Security Supplement of the CPS, fielded in December, also asks about SNAP receipt in the past year (FSFDSTMP) as well as each month of the year (FSSTMPJAN, FSSTMPFEB, etc.). I would not expect these CPS values to be available as part of ATUS unless a user were to link ATUS respondents to the CPS (I am linking to more information on the panel aspect of CPS data, including linking them to ATUS data).