Wondering about dementia and Alzheimer’s related questions, and if these interviews are done via proxy? In addition, what is the question asked to collect responses to SMOKESTATUS variables?
There are a number of technical variables that can be used to identify who responded to survey questions in the IPUMS Health Surveys datasets. I am not aware of any dementia/Alzheimer’s related questions available through IPUMS MEPS, but am including information for both MEPS and NHIS.
In IPUMS MEPS, the variable RESP indicates if the person is the respondent who answered the survey questions on behalf of the reporting unit; the variable PROXY notes if the person responding is not a member of the responding unit.
For IPUMS NHIS, the appropriate variable used to identify proxy respondents depends on the universe of the question. For questions asked of sample adults, see the variables PROXYSA (sample adult needs a proxy), SAPROXYAVAIL (knowledgeable proxy available for sample adult), and SAPROXYREL (relationship of proxy to sample adult). The sample adult section of the questionnaire allows for a proxy response if the selected adult has a physical or mental condition that prevents them from responding, but does not include additional restrictions on who answers questions about senility, dementia, or Alzheimer’s disease. For questions asked of all persons in the household (or not restricted based on sample adult/sample child status), the variable FAMRESPFLAG indicates if the person is the family respondent, or the main person who answers questions about the family’s health.
I appreciate your post about proxy responses in the NHIS. Using the variable FAMRESPFLAG, a large share (43%) of my sample are not primary respondents which suggests that NHIS relies on proxy reports for their health statuses. Given that all adult members of the household 17 years of age and over who are at home at the time of the interview responded for themselves, I wonder if some respondents who were not family respondents actually reported their health statuses. Can you please comment on the accuracy of FAMRESPFLAG in identifying adult respondents (not sample adults) who were not at home at the time of the interview? Was FAMRESPFLAG derived from the original variable FINTRO2 in the NHIS? I suspect that the variable is based on KNOW2 (Who in the family would you say knows about the health of all the family members?), not FINTRO2 (Are (READ NAMES) at
home now?). Thank you so much for your help.
The interview procedure for the NHIS is to select one adult to be the primary respondent for the family and provide information for all children and adult family members. As you note, adult members of the family who are home at the time of the interview may respond for themselves, but this doesn’t necessarily mean that they do. I do not know of a variable in the public use data that indicates if the individual responded directly to family questionnaire items, or if all responses are from the family respondent. FAMRESPFLAG only indicates if the person was the official family respondent; it is comparable to the original NCHS variable FMRPFLG. It cannot be used to identify if the person was at home at the time of the interview. At least in 2018, the questionnaire items KNOW2 and FINTRO2 are not included in the public use data files. You can look at the NHIS Sources tab of any IPUMS NHIS variable to see the original NCHS variable(s) associated with an IPUMS variable.
Thank you so much.