Self-reported drug use data: what do they reveal?

The purpose of this study is to examine self-reported marijuana and cocaine use responses from two nationally representative surveys. We compared prevalence rates across birth cohorts for multiple years of the National Household Survey of Drug Abuse (NHSDA) and also analyzed longitudinal inconsistencies in self-reported drug use between two waves of the National Longitudinal Survey Youth Cohort (NLS-Y). We found the percentages of respondents admitting use within the past month, year, and lifetime were comparable to other findings and were consistent with the declining trend in drug use in the late 1980s. A comparison of lifetime prevalence rates revealed seemingly inconsistent reports between 1985 respondents and their birth cohorts in 1990. Using the longitudinal NLS-Y data, we found that roughly one-fifth of the people who had admitted using marijuana or cocaine in their lifetime on the 1984 survey subsequently denied ever having used in 1988. The majority of these cases were people who reported having used infrequently. The subsample of women had similar patterns. In addition, we discovered that women who had been pregnant between the two surveys were more likely to inconsistently deny having ever used, while those who were currently pregnant responded more honestly about their past use. Overall, we found that although most people are willing to provide accurate accounts of their use, the researcher should be aware that under-reporting or complete denial does occur. Most importantly, external factors appear to contribute to the rate of inaccurate reporting.