The Impact of Restorative Practices on the Use of Out-of-School Suspensions: Results from a Cluster Randomized Controlled Trial

18 schools randomised (9 intervention, 9 control); 5,878 students in the analytic sample

US K-12 (six elementary, six middle, four high, two combined schools); one year of outcome data, 2018-19.

Cluster-randomised trial with school-level assignment and administrative suspension records as the outcome. Two limitations the authors state matter for interpretation - four of the nine control schools were already using some restorative approach, and data collection was truncated by the March 2020 school closures.

No significant main effect on the likelihood of suspension (B = -0.001, p = 0.95), and no variation by school level, gender, race or disability status (chi-square(5) = 8.4, p = 0.14). One interaction survived: students with a PRIOR suspension were much less likely to be suspended in intervention schools — 11% predicted probability versus 33% in control schools (d = 0.77). A companion analysis of the same sample found students in intervention schools less likely to receive a discipline incident record (11.1% vs 18.2%).

Low. School-level randomisation.

A direct test of the Pittsburgh result and it does not reproduce the main effect. The surviving subgroup finding — a large effect concentrated on previously suspended students — is a different claim from the universal one restorative practices are sold on, and it is a single subgroup in a single trial.