Dec 31, 1989
This Note evaluates the impact of baseline nonresponse in Project ALERT (Adolescent Experiences in Resistance Training), a longitudinal experiment that tests the effectiveness of a school-based drug prevention program for seventh- and eighth-grade students in 20 treatment and 10 control schools in California and Oregon. Of the intended sample, 16 percent failed to participate in some or all of the baseline data collection. The analysis addresses four issues: (1) the number of students lost because of baseline nonresponse; (2) the ways each group of nonrespondents — parent refusals, student refusals, and absentees — differed from respondents; (3) the ways nonresponse changed the sample; and (4) whether nonresponse introduced differences among the treatment groups. The authors find that the steps taken to reduce baseline nonresponse in Project ALERT — passive informed consent, careful explanation of data privacy, and makeup sessions for absentees — proved successful. Besides holding baseline nonresponse to 16 percent, these steps sharply limited differences between the baseline population and baseline sample and avoided substantial differences among the treatment groups.