EBP Exchange - UMB School of Social Work

This very interesting article by James Derzon of Battelle uses meta-analysis to identify risk factors associated with substance abuse. Then, the occurrence of substance abuse among youth who have these risk factors is assessed. Most non users of alcohol and marijuana can be identified from having 5 or fewer risk factors. But, more interesting is the finding that it is not until 14 risk factors that the number of users of these substances becomes larger than the nonusers. That is, there are many youth with very high numbers of risk factors who are nonusers. Another way to look at this is that if you, then, targetted the intervention on children with 14 or more risk factors, you would exclude 71% of the users–greatly reducing your prevention effects. The paper goes on to predict true and false positive rates and selection accuracy using Monitoring the Future data and conclude that even though a less targetted program will result in smaller effect sizes for the intervention, the benefits of a universal program will be greater because there are so youth who could benefit from an effective intervention but would have been screened out. The methodology of this analysis could be instructive to the design of other prevention programs.  Deron, J. H. (2007).  Using correlational evidence to select youth for prevention programming. Journal of Primary Prevention, 28, 421-447.