U.S. flag

An official website of the United States government, Department of Justice.

NCJRS Virtual Library

The Virtual Library houses over 235,000 criminal justice resources, including all known OJP works.
Click here to search the NCJRS Virtual Library

Subject Attrition in Prevention Research (From Drug Abuse Prevention Intervention Research: Methodological Issues, P 213-234, 1991, Carl G. Leukefeld and William J. Bukoski, eds. - see NCJ-140135)

NCJ Number
140147
Author(s)
A Biglan; D Hood; P Brozovsky; L Ochs; D Ary; C Black
Date Published
1991
Length
22 pages
Annotation
This discussion of subject attrition in prevention research focuses primarily on school-based smoking prevention and drug abuse prevention studies.
Abstract
The available evidence suggests a significant rate of attrition in both types of studies. Attrition in school-based studies is generally due to student absenteeism, student transfers, drop-outs between initial interview and postintervention assessment, failure by the student to come to the assessment, decision by the student or parent to discontinue participation, or failure to complete the questionnaire properly. Subject attrition can threaten the internal validity of a study that depends on the researchers' confidence that any differences between experimental conditions are due to the experimental variable that was being studied rather than to extraexperimental factors. The authors recommend randomizing the units of study in order to maximize the equivalence of attrition across groups. Attrition also threatens a study's external validity to the degree that subjects who are missing from postintervention assessments are systematically different from the remaining subjects. However, the authors note that using schools, rather than individual students, as the unit of analysis can obviate the effects of attrition. Attrition rates can be reduced if researchers try to attain the largest possible proportion of students, return to assess absentees, and track missing students. 2 tables, 20 references, and 1 appendix