Restorative Practices in Schools: Tiers, Scripts and Evidence

By the Transforming Education editorial team. Reviewed for accuracy against the sources listed below.

Restorative practices ask a different question after something goes wrong. Instead of "what rule was broken and what is the punishment?", they ask "who was harmed, what do they need, and whose job is it to make things right?" The approach also works before anything goes wrong, through daily language and routines that build the relationships a repair process depends on.

This guide covers the tiers of practice, circle scripts and restorative questions you can use this week, what the research honestly shows, and the implementation mistakes that explain most of the disappointing results.

The continuum of practice

The International Institute for Restorative Practices (IIRP) describes a continuum that runs from informal to formal: affective statements, affective questions, small impromptu conversations, circles and formal conferences. The informal end is where most of the work happens. IIRP's handbook argues that when a teacher sets a more positive tone day to day, the serious problems that need formal responses diminish in intensity and frequency.

Affective statements tell a student how their behavior affected you. IIRP's examples contrast "Stop teasing Sandy" with "I'm uncomfortable when I hear you tease Sandy," and "Nice job on your project" with "I am so impressed with the results of your final project." Specific and emotive beats general and procedural, in both directions.

Affective questions draw out what the student is thinking and feeling, and who else was affected. Small impromptu conversations use those questions with two or three students on the spot. Circles build community proactively and respond to conflict. Formal conferences bring together the person who caused harm, those harmed and supporters, with a trained facilitator, for serious incidents.

Tiers of practice

Schools usually organize the continuum into three tiers that map onto a multi-tiered system of supports.

Tier 1 is not optional. Responsive circles and conferences depend on students having practiced speaking and listening in a circle when the stakes were low, and on adults having a relationship to draw on. Schools that start at Tier 3 because discipline data is urgent usually find the process hollow.

Circle scripts

Community-building circle (10-15 minutes)

Arrange chairs in a circle with no desks. Use a talking piece; only the person holding it speaks. Passing is allowed.

1. Opening (1 minute). "Welcome to our circle. One voice at a time, speak to the circle, listen to understand. You can pass."
2. Check-in round (5-8 minutes). One question, one sentence each. Rotate between light and reflective prompts: "One word for how you're arriving today." "Something you're looking forward to this week." "A time someone helped you recently." Our SEL check-in questions include dozens more.
3. Optional second round (3-5 minutes). Connect to classwork: "One thing that was hard in yesterday's lesson and what you tried."
4. Closing (1 minute). "Thanks for being here. One word to close."

Responsive circle after conflict (20-40 minutes)

Use only after the group has practiced community circles. Prepare participants individually first.

1. Opening and norms. Restate the norms and the purpose: "Something happened that affected this group. We're here to understand it and decide how to make things right. We're not here to punish."
2. What happened. Round one, each person who was present answers "What happened from your point of view?"
3. Impact. Round two: "Who has been affected and how?" Include people who were not directly involved but noticed.
4. Needs. Round three: "What do you need to feel okay about this?"
5. Agreement. Round four: "What are you willing to do?" Write the agreements down with names and dates.
6. Closing. One word, and a scheduled check-in.

Try this: Run a community circle twice a week for a month before you attempt a responsive circle. Students need to trust the format before they will trust it with something that hurt.

The restorative questions

IIRP's restorative questions come in two sets. Print them on a card and keep it in your pocket.

When things go wrong (for the person who caused harm):

• What happened?
• What were you thinking of at the time?
• What have you thought about since?
• Who has been affected by what you have done? In what way have they been affected?
• What do you think you need to do to make things right?

When people are affected (for those harmed):

• What did you think when you realized what had happened?
• What impact has this incident had on you and others?
• What has been the hardest thing for you?
• What do you think needs to happen to make things right?

The questions are deliberately neutral. "What were you thinking of at the time?" assumes thinking happened and invites the student to explain it; "What were you thinking?!" shouted across a room assumes it did not.

What the evidence shows

The research on restorative practices is more mixed than advocates or critics usually admit. Four rigorous studies illustrate the range.

Pittsburgh, 2015-2017. RAND's randomized evaluation of the Pittsburgh Public Schools' restorative practices initiative, known as PERC, in 44 schools found that days lost to suspension fell 36% in schools that implemented restorative practices compared with 18% in the comparison schools over the same period, and teachers in implementing schools rated overall school climate as improved. The research brief also reports what did not change: academic outcomes did not improve, and suspensions did not fall for middle school students or for violent offenses. RAND called the approach promising, particularly for elementary schools.

Maine, two-year trial. A cluster-randomized trial by Acosta and colleagues, published in 2019 in the Journal of Youth and Adolescence, enrolled 14 Maine middle schools (13 analyzed, with 2,771 students in the analytic sample) and tested IIRP's Restorative Practices Intervention over two years. Intent-to-treat results were not significant for any of the 11 outcomes. The authors note the intervention delivered only a modest amount of restorative practice, not much different from control schools. Students who personally reported more restorative experiences did report better climate, connectedness and peer attachment and less physical and cyberbullying victimization, but those are correlations, not causal effects.

Northeastern urban district, 2018-2019. A cluster-randomized trial by Huang, Gregory and Ward-Seidel, published in Prevention Science in 2023, randomized 18 K-12 schools (5,878 students) to a Whole School Restorative Practices Project combining 15 hours of staff training, coaching, school-based restorative coordinators and weekly community-building circles. After one year there was no statistically significant overall difference in the likelihood of out-of-school suspension. The program did reduce suspension for students who had been suspended the prior year: in control schools 33% of those students were suspended again, versus 11% in intervention schools. The planned second year was cut short by the pandemic.

Chicago Public Schools. An NBER working paper by Adukia, Feigenberg and Momeni, later published in the American Economic Review, studied the district's adoption of restorative practices and found that suspensions decreased, with effects driven by Black students, an estimated 15% decrease in out-of-school arrests, and null effects on test-score value added, which rules out meaningful average declines in achievement.

Research: Read together, the studies suggest restorative practices can reduce exclusionary discipline and improve adult-rated climate when implemented with enough intensity, especially in elementary schools and for students with prior suspensions. They have not been shown to raise test scores, and thin implementation produces thin results.

Implementation pitfalls

Announcing policy before building skill. Telling staff that suspensions will drop before anyone has practiced an affective statement produces resentment and quiet non-compliance. Sequence training, coaching and circles first; change discipline codes second.

Underdosing. The Maine trial's own authors attribute the null result partly to how little restorative practice students actually experienced. One training day is not an implementation.

Skipping the adults. Staff who have never sat in a circle cannot run one with conviction. Hold staff meetings in circles, use affective statements in feedback and let teachers experience repair with each other.

Confusing a conversation with a consequence. A restorative conference is demanding for the student who caused harm; it is not "getting off." Make the agreement concrete, written and followed up.

Ignoring middle school dynamics. Two of the four studies found weaker or no effects in middle grades, where peer status is everything. Keep repair private when possible, prepare participants individually and avoid formats that expose a young adolescent in front of classmates.

Dropping the data. Track suspensions, referrals, circle frequency and conference agreements by grade and subgroup, and look at them with staff each quarter. Restorative practices are one strand of a wider approach; they work best alongside the preventive strategies in our classroom management guide, the predictability and regulation routines of trauma-informed teaching, and explicit teaching of perspective-taking from the social awareness toolkit.