I am wondering if there is a reason why the variables CNCERV and CNBRES have such a high number of NIUs for the 2015 sample.
Each has over 100,000 cases that are not in the universe.
The universe tabs for CNCERV and CNBRES note that these questions are only asked of sample adults who reported being told that they had any type of cancer (CANCEREV = 2). In the 2015 NHIS, this included 3,289 sample adults. Respondents outside of the universe criteria are coded as NIU. These NIU respondents consist of 70,117 individuals who were not sample adults (including both children and non-sample adults) and 30,350 sample adults who had never been told that they had cancer. The 2019 NHIS redesign removed non-sample adults and children from the survey, which significantly decreased the number of NIU cases for these variables.