Because of the change in methodology in 1994, many of the estimates from the 1993 and earlier NHSDAs are not comparable to estimates from the 1994 and 1995 NHSDAs. Since it is important to describe long-term trends in drug use accurately, an adjustment procedure was developed and applied to the pre-1994 estimates. This adjustment uses the 1994 split sample design to estimate the magnitude of the impact of the new methodology for each drug category. To improve the precision of the adjustment used for low-prevalence drugs, data from the large 1993 sample were combined with the smaller 1994-A sample (both used the "old" questionnaire). Essentially, if the new methodology is estimated to produce a 20% higher prevalence rate for the use of a particular drug by a demographic group, then estimates based on the old methodology (for that drug) are adjusted upward by 20%. A more complete description of the adjustment method is given in Appendix 2.
Readers need to be aware that all 1979-93 data shown in this report are different from previously published NHSDA estimates for 1979-93. Because the adjustments were developed from sample survey data, they are subject to sampling error and, therefore, may in some cases introduce additional variation into trends. This is particularly true for estimates of rare behaviors and for small subgroups.
This page was last updated on June 16, 2008.