Survey Methodology

The sample for the 1997 NHSDA was designed so that study results could be used to make inferences about the U.S. civilian, noninstitutionalized population aged 12 or older. This 1997 surveyed population was identical to the surveyed population of the 1991, 1992, 1993, 1994, 1995, and 1996 NHSDAs, but 1997's differed somewhat from pre-1991 NHSDA populations, which were restricted to the household population in the 48 contiguous States. Alaska and Hawaii were included in the sample for the first time in 1991, as were civilians living on military bases and persons living in noninstitutional group quarters, such as college dormitories, rooming houses, and shelters. Although this change introduced some minor inconsistencies between the samples of the 1991-1996 NHSDAs and pre-1991 NHSDA samples, the impact on trends in drug use estimates is generally inconsequential. Specific details of the sample design and weighting procedures are given in Appendix D.

The sample design was a multistage area probability sample. A national probability sample of dwelling units was selected in 1997 from the same 115 primary sampling units (PSUs) used in 1995 and 1996, plus a total of 18 supplemental PSUs from Arizona and California. PSUs are defined as counties (administrative subdivisions of States) or groups of counties, such as metropolitan areas. As did the 1991-1996 NHSDAs, the 1997 sample design used a composite size measure methodology and a specially designed within-dwelling unit selection procedure to meet specified precision constraints for subgroups defined by age and minority group membership. The sample design oversampled blacks, Hispanics, and young people to improve the accuracy of estimates for those populations. In addition, beginning in quarter 2 of the 1997 NHSDA, residents of Arizona and California were oversampled to provide direct survey estimates for these State populations. The basic plan involved several selection stages: the selection of primary areas (e.g., counties), subareas (blocks or block groups) within these primary areas, sample dwelling units within subareas, age domains within sampled dwelling units, and eligible residents (if any) within the sampled age domains.

To reduce the number of required screenings, two selections per household were allowed. Approximately 101,000 listings were screened to identify sufficient dwelling units to yield the Hispanic and black age-domain samples. In each selected dwelling unit, a roster recording the age, race/ethnicity, and gender of all household members aged 12 or older was completed. Two, one, or no household members were selected to be interviewed using a random sampling procedure, with selection probabilities based on the race/ethnicity of the head of household and the ages of household members. The procedure was designed to ensure adequate sample sizes for the four age groups (12 to 17, 18 to 25, 26 to 34, and 35 or older) and the three racial/ethnic groups (Hispanic, black, and white). In 1993, 1994, and 1995, persons aged 18 to 34 identified as current cigarette smokers by the household screening respondents also were oversampled, although this practice was abandoned in 1996.

Trained interviewers interviewed respondents in person in their homes. At the conclusion of data collection for the study, sample weights were constructed that reflect the various stages of sampling described above. The sample weights then were adjusted to account for sample persons who could not be found at home or who refused to participate. In pre-1991 surveys, the weights were adjusted for nonresponse with weighting-class adjustments. In the 1991 through 1997 surveys, a generalized raking procedure based on logistic regression was used. The logistic adjustment is an extension of the weighting-class procedure that achieves more nonresponse bias reduction through the use of additional predictors of response. Finally, these weights were poststratified to U.S. Bureau of the Census projections of the number of persons in the U.S. civilian, noninstitutionalized population. The 1993 through 1997 estimates were poststratified using population projections based on the 1990 Census. The final adjusted weight was used in deriving estimates.

SAMHSA, an agency in the Department of Health and Human Services, is the Federal Government's lead agency for improving the quality and availability of substance abuse prevention, addiction treatment, and mental health services in the United States.    Site Map | Contact Us | Accessibility | Privacy Policy | Freedom of Information Act  Disclaimer | Department of Health and Human Services | SAMHSA | White House | USA.gov * PDF formatted files require that Adobe Acrobat Reader® program is installed on your computer. Click here to download this FREE software now from Adobe. What's New Highlights Topics Data Drugs Pubs Short Reports Treatment Help Mail OAS