Promoting Social and Emotional Competencies in Italian Schools: A Two-Stage Randomized Evaluation of RULER and an Embedded Digital Decision-Making Activity

RULER is a school-based social and emotional learning approach centered on emotional literacy. Teachers assigned to RULER receive three training modules totaling approximately 21 hours, continued implementation support during the school year, and access to structured classroom materials and tools. The approach includes the Charter, Mood Meter, Meta-Moment, Blueprint, and activities focused on recognizing, understanding, labeling, expressing, and regulating emotions.
Teachers in the grade assigned to RULER are expected to integrate these tools and practices into ordinary classroom routines during the 2026/27 school year. The comparison grade is not assigned to receive the first-year RULER intervention and continues its usual educational activities.
At the second stage, RULER classes meeting prespecified technological and operational eligibility criteria will be assigned either to continue with RULER only or to receive a separate classroom-based digital decision-making activity delivered in a single school session lasting no more than two hours. The activity is designed by the research team and informed by experimental and behavioral economics. Students will make individual decisions in structured digital tasks that may include interactions with classmates.
The activity is an additional experimental component and is not part of RULER, its training program, or its standard classroom tools. Its final tasks, target social and emotional competence, incentive structure, within-activity behavioral measures, and scoring rules will be preregistered in a versioned addendum before the second-stage randomization.

2026-08-31

2027-06-30

1. Stage-1 primary endpoint: Post-test SELIS Self-Awareness domain score.
2. Stage-2 primary endpoint: One of the five predefined SELIS domain scores, selected ex ante to match the focal competence targeted by the digital activity. The focal competence and corresponding SELIS domain will be specified and frozen in a versioned addendum before the Stage-2 randomization.

Self-Awareness was selected as the primary outcome for the Stage-1 RULER comparison because it is the student-level SELIS domain most closely aligned with RULER’s first-year emphasis on emotional awareness and emotional literacy. It captures students’ perceived ability to understand their own emotions, thoughts, and values and how they influence behavior across contexts.
Grade 4 students will complete the SELIS form for grades 3–5, and grade 7 students will complete the form for grades 6–8. Responses will be coded from 0 to 3, with higher values indicating greater perceived ease in performing the described social and emotional skill.
The primary domain score will be the mean of the age-appropriate Self-Awareness items when at least 80 percent of the domain items have valid responses. Scores will be standardized separately by grade using the pooled initial-measurement mean and standard deviation among respondents in the primary 12-institution randomized cohort, across Stage-1 assignment conditions. The same fixed grade-specific parameters will be applied to post-test and follow-up scores.
The primary Self-Awareness measure will use all Self-Awareness items included in the relevant age-appropriate form. As a sensitivity analysis, the score will be reconstructed using only the Self-Awareness items included in the 41-item common core shared by the two grade-span forms. These procedures constitute the project scoring of the Italian SELIS adaptation and do not reproduce the official SELIS 1–999 Rasch scale.
The primary endpoint is the student post-test planned for late May 2027. The initial student measurement is scheduled after the beginning of teacher training but before the planned introduction of RULER tools and activities in the classroom.
Stage 2 will have a distinct, estimand-specific primary endpoint for the randomized incremental contrast. It will be one of the five predefined SELIS domain scores, selected ex ante to match the focal competence targeted by the digital activity. This domain is part of the common SELIS questionnaire administered to all study students and will be constructed using the same age-appropriate form, response coding, 80-percent item-completion rule, and fixed grade-specific standardization across study conditions. The focal competence and corresponding SELIS domain will be frozen in a versioned addendum before the Stage-2 randomization.

1. Post-test SELIS Relationship Skills domain score (key secondary).
2. Post-test SELIS Self-Management domain score (key secondary).
3. Post-test SELIS Social Awareness domain score.
4. Post-test SELIS Responsible Decision-Making domain score.
5. Post-test project overall SELIS index.
6. Follow-up SELIS Self-Awareness, Relationship Skills, Self-Management, Social Awareness, and Responsible Decision-Making domain scores, and project overall SELIS index.
7. Harmonized school administrative indicators of attendance.
8. Harmonized school administrative indicators of assessments and school achievement.
9. Harmonized school administrative indicators of disciplinary or conduct outcomes.
10. Delayed INVALSI outcomes in Italian, mathematics, and English at the next scheduled national assessment.

Relationship Skills and Self-Management are designated as key secondary outcomes. Relationship Skills is prioritized because first-year RULER practice is expected to support students’ perceived ability to communicate, cooperate, seek help, and navigate interactions with teachers and peers. Self-Management is expected to be somewhat more distal, as changes in emotional regulation and stress management may require sustained practice and transfer across contexts. Social Awareness and Responsible Decision-Making are broader secondary outcomes that are theoretically related to RULER but are expected to change less proximally during the first year. Follow-up SELIS outcomes will assess the persistence and possible broadening of effects. Each secondary SELIS domain score will use the same response coding, 80-percent item-completion rule, and grade-specific initial-measurement standardization specified for the primary outcome. Follow-up scores will use the same construction and fixed standardization parameters.
The project overall SELIS index will be the equally weighted mean of the five grade-standardized domain scores when all five domain scores are available. This composite will then be rescaled separately by grade using its pooled initial-measurement mean and standard deviation in the primary 12-institution randomized cohort; the same fixed parameters will be applied at post-test and follow-up.
School administrative and INVALSI outcomes will be treated as distal exploratory outcome families. Their exact construction, aggregation level, harmonization, and multiplicity rules will be frozen before the research team accesses the corresponding post-treatment outcome data.

The study uses a two-stage clustered randomized design.
At Stage 1, each of the 12 probabilistically assigned institutions contributes a grade 4 cohort and a grade 7 cohort. These institutions were assigned to one of two configurations: RULER in grade 4 and comparison in grade 7, or comparison in grade 4 and RULER in grade 7. Two additional institutions had their assignments fixed ex ante because of documented feasibility constraints. These two institutions participate in data collection but are excluded from the primary Stage-1 causal analysis.
Stage-1 randomization used a prespecified constrained assignment procedure. The institution and class rosters, assignment protocol, randomization code, and SHA-256 hashes were frozen and deposited with the co-investigators on July 5, 2026, before the official draw. Of 162 structurally admissible institutional allocations, 41 satisfied the prespecified balance criterion, and one allocation was selected uniformly from this acceptance set on July 8, 2026, using a reproducible seed derived mechanically from the frozen roster.
At Stage 2, RULER classes meeting prespecified technological and operational eligibility criteria will be randomly assigned at approximately a 1:1 ratio to RULER only or RULER plus the digital decision-making activity. Eligibility will be defined using criteria that do not depend on study outcomes and will be assessed and frozen for both grade cohorts before the Stage-2 randomization. A versioned addendum filed before that randomization will report the eligible roster, blocking rules, exact conditional assignment probabilities, and the procedures required to combine the two assignment stages. The two institutions fixed at Stage 1 may contribute to the randomized Stage-2 comparison if their RULER classes meet the eligibility criteria.
The primary Stage-1 estimand is the intention-to-treat effect of the embedded RULER-only strategy relative to the comparison condition in the frozen cohort of 12 randomized institutions. RULER-only observations arise both from RULER-assigned classes that are not eligible for the additional activity and from eligible RULER-assigned classes randomized not to receive it. This contrast will therefore be estimated using the exact joint assignment probabilities implied by the 41-allocation Stage-1 acceptance set and the Stage-2 blocked randomization; it will not be estimated as an unweighted comparison of the observed groups.
The pooled Stage-1 estimand will assign equal weight to the student-average grade-specific effects for grades 4 and 7. Grade-specific effects will also be reported.
The primary Stage-2 contrast will compare assignment to RULER plus the digital decision-making activity with assignment to RULER only among eligible RULER classes. Its primary outcome will be the SELIS domain corresponding to the activity’s prespecified focal competence, measured at the common student post-test in both Stage-2 arms. The focal competence, matched SELIS domain, and Stage-2 outcome-analysis procedures will be frozen in the versioned addendum before the Stage-2 randomization.
Within-activity choices, response times, and operational logs will be observed only for students who participate in the activity within classes assigned to it. They will support prespecified within-arm analyses of their correspondence with the matched SELIS self-report measure and will not be treated as between-arm causal outcomes.
A key policy-relevant secondary estimand will compare assignment to the overall RULER-centered pathway, comprising RULER only or RULER plus the activity as generated by the Stage-2 design, with the comparison condition. A secondary total-package comparison of RULER plus the activity with the comparison condition will be defined in the Stage-2 addendum.
Analyses including all 14 institutions for the Stage-1 contrast will be reported as supportive and will not form part of the primary randomized causal analysis.
The frozen July 2026 class roster defines the confirmatory Stage-1 cohort. Any subsequently activated sections invited to participate in measurement will constitute an extended measurement stratum and will be included only in secondary or sensitivity analyses.
The initial student measurement is scheduled after initial teacher training but before the planned student-facing implementation of RULER. The post-test is planned for late May 2027, after implementation of the additional digital activity. The follow-up is planned during the 2027/28 school year.
The comparison grade does not receive direct RULER implementation during the study year. Because RULER and comparison grades belong to the same institution, indirect between-grade exposure will be documented and considered when interpreting the estimates.
All main estimates will follow assignment, irrespective of actual implementation or participation. Post-assignment measures of implementation intensity will not be used to redefine the primary treatment groups. Detailed estimation, inference, missing-data, multiplicity, and covariate-adjustment procedures will be specified in a versioned analysis plan before the research team accesses the corresponding post-treatment outcome data. Estimation and inference for each randomized contrast will respect its corresponding assignment mechanism, including the Stage-1 constrained randomization and, where relevant, the Stage-2 blocked randomization.
Implementation resources will be measured prospectively to support a separately prespecified cost-effectiveness analysis linked to the trial’s effectiveness estimates.
Within classes assigned to receive RULER, teacher-completed ratings of students’ observable social-emotional competence are planned as part of implementation monitoring and descriptive multi-method triangulation. These ratings are not included in the prespecified confirmatory impact-outcome family. The instrument and edition, administration schedule, rater assignment and continuity rules, and scoring procedures will be frozen and time-stamped before the first rating wave. The initial rating wave is expected in late October or early November 2026, after teachers have had sufficient opportunity to observe students in regular classroom settings.

Not available

Stage 1 used computerized constrained randomization. The institution and class rosters, assignment protocol, randomization code, and SHA-256 hashes were frozen and deposited with the co-investigators on July 5, 2026. The procedure generated 162 structurally admissible allocations, retained the 41-allocation acceptance set satisfying the prespecified balance criterion, and selected one allocation uniformly from that set on July 8, 2026, using a reproducible seed derived mechanically from the frozen roster.
Stage 2 will use computerized class-level random assignment among eligible RULER classes, with an approximately 1:1 allocation within prespecified school or school-site blocks. Eligibility criteria, blocks, any allocation constraints, and exact conditional assignment probabilities will be frozen before the draw.

Stage 1: school institution.
Stage 2: class among RULER classes meeting prespecified technological and operational eligibility criteria.

Yes

Stage 1 includes 12 randomized school institutions. Two additional institutions had their assignments fixed ex ante and are excluded from the primary Stage-1 causal analyses. The full roster contains 127 classes.
Stage 2 is expected to include approximately 54–60 eligible RULER classes; the final number will be reported before the Stage-2 randomization.

The frozen roster includes 2,389 planned students across 127 classes. The primary Stage-1 causal population consists of 2,070 planned students across 110 classes in the 12 probabilistically assigned institutions. Actual participation and subsequent sample flow will be reported separately.

The observed Stage-1 assignment produced:
- Configuration A, RULER in grade 4 and comparison in grade 7: 7 institutions, all probabilistically assigned.
- Configuration B, RULER in grade 7 and comparison in grade 4: 7 institutions, of which 5 were probabilistically assigned and 2 were fixed ex ante.
Across all 14 institutions:
- RULER-centered pathway assignment: 62 classes and 1,182 planned students.
- Comparison: 65 classes and 1,207 planned students.
Within the 12 probabilistically assigned institutions used for the primary Stage-1 causal analysis:
- RULER-centered pathway assignment: 53 classes and 1,007 planned students.
- Comparison: 57 classes and 1,063 planned students.
Among the approximately 54–60 eligible RULER classes, about half will be assigned to RULER plus the activity and half to RULER only. RULER-assigned classes that are not eligible for the activity will also remain in the RULER-only condition but will not belong to the Stage-2 randomized contrast.