Hello awesome Staff!
I was wondering about why there are NIU observations for the FOODSTMP variable which has a universe of “All interviewed households and group quarters.” I see in the description that “Households that were not questioned were coded as ‘no’ for . . . FOODSTMP.” Should I assume that IPUMS enters households not actually asked the question into NIU and it’s up to the data user to recode themselves? Is this also the case for the other “several variables relating to the receipt of means-tested benefits”?
Thank you so much!
Households who are NIU for FOODSTMP are sampled households that were not interviewed for the CPS. Household type (interviewed versus non-interviewed) is identified in HHINTYPE. Non-interviewed household records are only included in hierarchical data extracts, which include both household records and person records. In person-level extracts, the default extract type for IPUMS CPS, only interviewed household members are included. Households that were interviewed in the CPS but whose income likely precluded them from being eligible to receive food stamps were not asked questions about food stamps. These respondents have FOODSTMP=1 (no). They are not coded as NIU. This coding decision was made by the Census Bureau and is present in the original data; this is not an edit made by IPUMS.
The description of FOODSTMP states:
“For some variables relating to means-tested government assistance, including FOODSTMP, information was collected only from households whose estimated income fell below a given threshold. During the first month that a household entered the survey, and one year later, the CPS interviewer asked the respondent to estimate the family’s total income in the past twelve months, by choosing one of fourteen broad categories. Only households in which the household members’ combined income fell below a given level (e.g., under 30,000 dollars in the 1980s) were questioned about means-tested program benefits. Households with estimated incomes above the threshold were presumed to neither qualify for nor receive such benefits, and were not asked these questions, to limit the length of interviews. Households that were not questioned were coded as “no” for several variables relating to the receipt of means-tested benefits, including FOODSTMP.”