State-Level Restorative Justice Policy and Students' Sense of School Safety: A Difference-In-Differences Analysis

Background: Restorative justice (RJ) practices have emerged as a promising approach to improve school climate and reduce exclusionary discipline practices that have disproportionately harmed Black and Brown youth. Since 2009, over 25 states have adopted school-based RJ legislation that varies widely in strength and scope, but commonly positions RJ as a pathway to reducing exclusionary discipline and improving school safety. Existing studies have largely focused on the district and school levels, leaving limited macro-level evidence on whether state RJ policies translate into broad changes in students’ school experiences. This study begins to fill this gap by examining whether state-level RJ policy adoption is associated with students’ sense of safety at school, a key component of school climate, and whether this association varies by race/ethnicity.

Methods: We linked the Youth Risk Behavior Surveillance Survey (2007-2023) with a state-level RJ education policy database across 39 states (analytic sample: 1,028,442 high school students). Sense of safety was measured as a binary indicator of missing school due to feeling unsafe. RJ policies were grouped into three functional categories: alternative-to-discipline practices (e.g., restorative circles, peer mediation), staff training, and administrative practices (e.g., strategic plans, funding), and coded based on statutory strength: permissive, encouraged, or mandatory. Using a difference-in-differences design with survey-weighted probit regressions, we estimated associations between each policy category and sense of safety, adjusting for demographic covariates and state and year fixed effects. Average marginal effects (AMEs) were calculated, and interaction terms tested for moderation by race/ethnicity.

Results: Between 2007 and 2023, the proportion of students reporting feeling unsafe doubled from 6.04% to 12.87% (p < .001), with persistent disparities by sex and race/ethnicity. Statutory strength was the primary indicator of school safety, with mandatory policies showing the strongest associations across all three policy types (alternative-to-discipline: AME = 1.1 percentage points [pp], p = 0.011; staff training: AME = 2.6 pp, p < .001; administrative: AME = 1.4 pp, p = 0.009). Mandatory policies were most consistently associated with increased sense of safety among Black, Hispanic, and Native Hawaiian/Pacific Islander students (AME range: 2.0 to 10.9 pp, p < .05), while significant associations for White students emerged only under mandatory staff training, and no significant associations emerged for American Indian/Alaska Native students.

Conclusion & Implications: Findings suggest that statutory strength may be an important factor in RJ policy design, with mandatory policies showing the strongest associations with sense of safety. Associations varied across racial and ethnic groups, suggesting that policy mandates alone may not be sufficient to produce equitable outcomes, and implementation warrants further study. Safety is foundational to equitable learning environments, and situating school-based RJ as an educational equity intervention has implications for practitioners seeking to reduce racial disparities in school safety. School social workers are positioned to advocate for stronger RJ legislation and to work alongside students and schools to support implementation and equitable safety outcomes. Future research should continue to explore how RJ policies influence broader school climate outcomes and for whom.