The major literature reviews of juvenile delinquency treatment research have generally come to discouraging conclusions with regard to the efficacy of treatment for preventing or remediating delinquent behavior. Those literature reviews and their conclusions, however, may not fully and accurately reflect the information available in existing studies. Many new studies have been reported since those reviews were completed, the reviewers have not systematically analyzed study details for factors associated with treatment outcome, and little account has been taken of the low statistical power of many studies or the role of sampling error in the distribution of study outcomes. Application of the relatively new statistical technique of meta-analysis is proposed as a way to better diagnose the nature of the apparent failure of delinquency treatment and identify the most promising directions for further study. Meta-analysis is based on a detailed quantitative coding of the characteristics of each study and the treatment effect size that results on each dependent variable of interest. Statistical models developed for such data permit identification of homogeneous distributions of effect size, estimation of population effect size values, construction of confidence limits around the population estimates, and analysis of the study characteristics that contribute to the effect size variance. The goals for the proposed delinquency treatment meta-analysis are to: (1) Quantitatively estimate the mean overall treatment effect size, and its statistical confidence limits, for the population of trustworthy studies and thus bring some precision and impartiality to the controversy over the effectiveness of delinquency treatment. (2) Provide a detailed analysis of the relationships between the effects of delinquency treatment and the circumstances of treatment in order to identify any factors associated with outcome. (3) Analyze the relationship between the methodology used in studies of delinquency treatment and the resulting effect sizes in order to identify and assess those methodological practices that introduce bias or excessive error into the results.
