Not in Universe in EDUC


I have a question about “Not in universe (NIU)” in the EDUC variable in ATUS. I understand that the universe for EDUC is age 15+ so anyone younger than 15 would not have values in EDUC. But I noticed that some observations age 15+ also are NIU (obs = 12,903). This is quite large for missing values. Am I misunderstanding something here?

Thank you!
Hyun Ju

1 Like

The universe tab for EDUC notes that data is collected only for CPS household members age 15+ (learn more about the ATUS sample being drawn from households existing the Current Population Survey). The description tab for EDUC offers more detail, specifically that the variable “was drawn from the final CPS interview, conducted 2-5 months before the ATUS interview” and “not updated at the time of the ATUS interview.” EDUC will therefore only be available for respondents who have data from their household’s final CPS interview. If a household did not respond to their final CPS interview or if the respondent moved into the household between the time of the CPS and ATUS interviews, they will not have any data for EDUC.

I looked at 2021 ATUS data to confirm this universe statement. I used the variable AGE_CPS8 to get the person’s age at the time of the CPS interview (when they were asked about their educational attainment) and PRESENCE to determine if the person was present in the household for the CPS, ATUS, or both. Using AGE_CPS8 instead of AGE (measured at the time of ATUS), I can see that all 661 records over the age of 14 who are NIU for EDUC in the 2021 ATUS are only present in the household at the time of the ATUS interview. Accordingly, these individuals will not have an EDUC value as they were not in the household when the data was collected at the time of the CPS. Please follow up if you are encountering cases where the combination of AGE_CPS8 and PRESENCE do not explain the NIU cases for EDUC.