Effects of interview mode on bias in survey measurements of drug use: do respondent characteristics make a difference?

Three recent empirical studies have provided strong evidence that self-administered questionnaires (SAQs), compared with interviewer questioning, substantially improve the reporting of drug use in population surveys. Specifically, SAQs appear to diminish underreporting bias. Two of these studies previously reported that this effect of interview mode varied significantly across gender, race/ethnicity, and age. Data from a randomized experiment embedded in the 1990 National Household Survey of Drug Abuse (NHSDA) field test were reanalyzed to test for those interaction effects. To better replicate prior studies, the NHSDA field test sample was restricted to people ages 18 to 45 (N = 1,877). The results of our statistical analyses generally replicated the finding of a main effect of SAQs on the reporting of drug use. However, only weak evidence was found to support the hypothesis that the advantage of SAQs varies substantially by the gender, race/ethnicity, or age of the respondent.