unique to drug use surveys, do hinder inferences on the levels and trends in the proportions of people who are consuming illegal drugs in the United States. These problems, however, do not imply that the data are uninformative or that the surveys should be discontinued. Rather, researchers using these data must either tolerate a certain degree of ambiguity or must be willing to impose strong assumptions. The problem, of course, is that ambiguous findings may lead to indeterminate conclusions, whereas strong assumptions may be inaccurate and yield flawed conclusions (Manski, 1995; Manski et al., 2000).
There are practical solutions to this quandary. If stronger assumptions are not imposed, then the way to resolve an indeterminate finding is to collect richer data. Data on the nature of the nonresponse problem (e.g., the prevalence rate of nonrespondents) and on the nature and extent of invalid response in the national surveys might be used to both supplement the existing data and to impose more credible assumptions. A sure remedy is to increase the frequencies of correct responses in surveys.
Nonresponse. Nonresponse is an endemic problem in survey sampling. Approximately 15 percent of the students surveyed by MTF fail to respond to the questionnaire and approximately 25 percent fail to respond to the NHSDA.12 These nonresponse rates are similar to those achieved by the National Survey of Family Growth, which also asks for sensitive information, and not much higher than those of the Current Population Survey, which is used to measure the unemployment rate. The committee recommends a systematic and rigorous research program (1) to understand and monitor nonresponse and (2) to develop methods to reduce nonresponse to the extent possible. The inferential problems that may arise in the absence of a better understanding of nonresponse are elaborated below.
The MTF and NHSDA data are uninformative about the behavior of nonrespondents. Thus, these data do not identify prevalence unless one makes untestable assumptions about the responses that nonrespondents would have given if they had responded. A simple example illustrates the problem. Suppose that 100 individuals are asked whether they used illegal drugs during the past year. Suppose that 25 do not respond, so the nonresponse rate is 25 percent. Suppose that 19 of the 75 respondents used illegal drugs during the past year and that the others did not. Then the reported prevalence of illegal drug use is 19/75=25.3 percent. How-
The 25 percent nonresponse rate for the NHSDA includes both unit (household) and element (person) nonresponse. The 15 percent nonresponse rate cited for MTF includes student nonresponse only. Schools that refuse to participate in the MTF survey are replaced by similar schools. School nonresponse and replacement are addressed later in the chapter.