Applying an empirically derived effect size distribution to benchmark the practical magnitude of interventions to reduce recidivism in the USA

• Published in: Journal of experimental criminology 2024-09, Vol.20 (3), p.817-841
• Main Authors: Gies, Stephen V., Nichols, Lindsey M., Mojekwu, Frank, Guerette, Rob T., Tanner-Smith, Emily E.
• Format: Article
• Language: English
• Subjects: Crime prevention; Criminology and Criminal Justice; Intervention; Law and Criminolgy; Political Science; Recidivism; Social Sciences
• ISSN: 1573-3750; 1572-8315
• DOI: 10.1007/s11292-023-09561-1

Objectives The purpose of this study is to apply an empirically derived effect size distribution to benchmark the practical magnitude of interventions aimed at reducing recidivism at the individual level. Methods We conducted a systematic review and meta-analysis of crime intervention evaluations. To establish a framework for benchmarking the magnitude of these interventions, we generated means, medians, tertiles, and interquartile ranges from these analyses. Results The results of the overall meta-analytic models revealed that crime intervention programs were associated with statistically significant reductions in recidivism regardless of outcome type ( k = 74, n = 293, OR [odds ratio] = 1.42, SE = 0.05, p = 0.0001, 95% CI [1.30, 1.57]). Conclusions Overall, the results from the current study have several important implications for the crime prevention field. Most importantly, the study provided evidence that the tradition of using generalized guidelines for interpreting effect sizes as small/medium/large should be avoided given that they are devoid of context and ignore important variations in effects across interventions and outcomes. Moreover, this study provided an alternative framework to benchmark the practical magnitude of crime intervention programs aimed at reducing recidivism at the individual level.