Comparing Prevalences of Use Across NHSDA Years

In the 1994 NHSDA, new instrumentation for measuring drug use was introduced, and that instrumentation has been in use since 1994. These changes were implemented to improve the validity of the drug prevalence estimates, but they also precluded being able to make direct comparisons with estimates published in 1993 and earlier years. Therefore, a subsample of respondents was administered the old instrumentation during the 1994 NHSDA to provide a means for assessing the magnitude of the difference in measures of drug use produced by the two different instruments. Then measures of drug use for 1993 and earlier were adjusted, using these two sets of measures available in the 1994 NHSDA, so that all measures of drug use from the NHSDAs since 1979 would be directly comparable to those generated in 1994 and later with the new instrumentation. As a result, all 1979-1993 estimates shown in this chapter are different from NHSDA estimates for those years published in 1993 or earlier.

A second important consideration to keep in mind while assessing the trend data presented in this chapter is that the NHSDA is based on a different sample of respondents each year. The survey is not designed to track changes in drug use behavior of individuals over time, but rather to measure the prevalence of drug use in a target population at a specific point in time. Estimates from different years can be compared to assess changes in the prevalence of drug use in the household population of the United States. Because a new sample is drawn each year, small fluctuations in the prevalence estimates across years are expected due to random sampling error. The tables presented in this chapter indicate when the differences in the estimates between 1997 and previous years are greater than would be expected as a result of sample differences.

Not only is a new sample drawn each year, but it is also important to recognize that the actual membership of the target population changes from each year to the next. For example, the household population aged 12 or older in 1997 includes 12 year olds who were not in the 1996 target population. Although the changes from year to year in the total household population membership are relatively small, the changes in membership for specific age groups are more pronounced. Indeed, the target population aged 12 to 17 in 1997 included none of the youths who were in that age group back in 1991. Thus, it is important to bear in mind that noteworthy differences in the prevalence of drug use across years may be due to the fact that membership of the target populations has changed, rather than (or in addition to) changes in the behaviors of individuals within a target population. This is especially true for the population subgroups defined by age.