This meta-analysis of the literature, involving experimental research studies (n=35) using control/comparison groups, focuses on the impact of involving families in the treatment of young offenders. In general, family intervention treatment significantly reduces the recidivism of young offenders compared to traditional non-familial responses to youth crime. Methodology, however, is an important determinant of recidivism in that less rigorous experimental designs tend to produce significantly lower rates of recidivism compared to more rigorous designs. Studies that used the most rigorous methods displayed a zero mean effect from treatment. The paper suggests the possibility of a relationship between "how we evaluate" and "what works" in correctional treatment research, and recommends further research examining the relationship using a larger sample of research studies. As policymakers, program managers, and researchers work to develop and implement the most effective correctional treatment practices, evaluation design may need to be more carefully constructed and monitored. Figures, tables, note, references, appendix