Evaluation of a Whole-School Change Intervention: Findings from a Two-Year Cluster-Randomized Trial of the Restorative Practices Intervention

13 middle schools (7 intervention, 6 control), 2,834 students, two years

Maine, US; middle schools.

Two-year cluster-randomised trial of the IIRP restorative-practices whole-school model, evaluated by RAND. Primary outcomes are student-reported school climate and related self-report scales — the weak measurement class this archive discounts — with the honest consequence that a null here is a null on the programme's own preferred yardstick.

No statistically significant differences between intervention and control schools on student-perceived school climate. The positive results in the paper are correlational within-arm associations - students who reported personally experiencing restorative practices reported better school climate and connectedness, better peer attachment and social skills, and less cyberbullying victimisation. That is a self-selected exposure measure and cannot carry a causal claim.

Low for the randomised contrast; the within-arm exposure analysis is fully confounded.

A second independent null for whole-school restorative practices, alongside Huang et al. 2023. Together with the Pittsburgh trial's negative achievement estimates, the randomised record for restorative practices is one partial positive on suspensions and two nulls.