A meta-analysis update on the effectiveness of early self-control improvement programs to improve self-control and reduce delinquency

41 eligible studies, all randomised controlled trials, children aged 10 and under

Mostly United States; children under 10.

An RCT-only meta-analysis, which is a real strength, undermined by what those RCTs measure. The self-control instruments are researcher-designed rater scales such as the Kansas Reflection-Impulsivity Scale and the Kendall and Wilcox Self-Control Rating Scale, and the problem-behaviour instruments are the Child Behavior Checklist and Eyberg Child Behavior Inventory - parent and teacher report throughout, at post-test, with no blinding. NO objective long-run outcome - achievement, earnings, convictions - appears in this meta-analysis or its 2010 predecessor. Numbers here were obtained via the National Institute of Justice CrimeSolutions profile rather than from the paper.

Self-control - 53 effect sizes from 36 randomised trials, overall mean effect size 0.32. Delinquency and problem behaviour - 36 effect sizes from 22 randomised trials, overall mean effect size 0.27. Both are post-test effects on rater-report instruments. Confidence intervals were not obtained.

Low for the randomised contrasts. The interpretive risk is not heredity but measurement - non-blind raters scoring instruments that the programme trained them to think about.

An update of the same team's 2010 Campbell review with seven additional studies, and consistent with it. What has never been replicated - or attempted - is the demonstration that these rating changes correspond to any objective outcome.