Reports of smoking in a national survey: data from screening and detailed interviews, and from self- and interviewer-administered questions.

PURPOSE: This study compares responses to questions about smoking in a brief screening interview with those from a subsequent, more detailed interview; it also compares responses to self-administered questions and questions administered by interviewers. The data are from the 1994 National Household Survey on Drug Abuse (NHSDA).
METHODS: About 22000 respondents completed the main questionnaire of the 1994 NHSDA. Earlier, a member of each sample household had been asked to provide screening information, including smoking status, for each person in the household. Then, one or more persons in the household were interviewed about their own smoking and drug use; for some respondents, the questions about smoking were self-administered and for others they were administered by an interviewer. We examined discrepancies between reports about smoking from the screening data and main interview data; we also compared the results across the two versions of the main interview smoking questions (self and interviewer-administered).
RESULTS: The screening data produced lower estimated rates of smoking than did the main interview data, particularly when proxies provided the screening data. In the main interviews, self-administered questions produced higher estimates of the prevalence of smoking than interviewer-administered questions, but only for adolescents.
CONCLUSIONS: Proxies can provide some information about smoking, although the data are likely to be biased for younger age groups and for infrequent smokers. For adolescents, self-administration appears to elicit more candid reports about smoking than interviewer administration. In addition, multiple items may help to capture smoking reports by persons who are reluctant to admit they have smoked recently or whose status as smokers is unclear.