Are School- and Home-Based Social-Emotional Learning Interventions Complements or Substitutes? Experimental Evidence from Rural China

T1: The school-based program consists of weekly SEL classes delivered by trained teachers.
T2: The home-based intervention provides structured, low-cost parent–child activities delivered via an app.
T3: The combined intervention integrates both T1 and T2 (from a cross-cutting design).

2026-04-27

2028-12-31

social emotional skills, mental health, student's academic performance

Academic performance is measured using administrative test records.
Social-emotional skills and mental health are measured using standard instruments (CES-D, WCSD). complements with reports on relationship etc.

Take-up among teachers and parents; student behavioral responses; beliefs of students and parents; self-reported parent–child relationships; and other non-cognitive skills.

We implement a cross-randomization design in which school-based and home-based interventions are assigned independently, with the school as the unit of randomization. We stratify schools into three groups based on school size. Within each stratum, we independently randomize the 149 schools along two dimensions. First, schools are assigned to receive the school-based intervention (74 schools) or not (75 schools). Second, they are independently assigned to receive the home-based intervention (74 schools) or not (75 schools).

Because the two randomizations are conducted independently, this procedure yields four groups: (i) school-only treatment, (ii) home-only treatment, (iii) combined school and home treatment, and (iv) pure control, each comprising approximately 37–38 schools. Within each school, all Grade 3 classes are included in the study sample.

Schools assigned to the school-based intervention receive a structural SEL curriculum delivered by local teachers selected by each school. At the start of each semester, the research team conducts a brief SEL training for participating teachers and provides accompanying materials, including teacher guidelines, student textbooks, instruction slides, and teaching props. Teachers deliver one SEL class per week and log completion through a mobile application after each session. In addition, SEL teachers are randomly assigned to discussion teams, which are led by a stellar teacher selected from within the team, who is incentivized to arrange monthly discussions.

Schools assigned to home-based intervention receive a program of parent-child activities designed by the research team, covering three core domains: self-awareness, self-management, and interpersonal skills. To ensure accessibility for diverse family contexts. The activities are low-cost and easily implementable, requiring no specialized equipment and drawing on materials and scenarios from daily family life. All activity materials (e.g., instructions and content) are delivered through a mobile application. Parents and children are instructed to complete one activity per week, with each session lasting approximately 30 minutes.

The control group receives no intervention from the research team during the study period.

Not available

Randomization done in office by a computer.

School level

Yes

149 schools

approximately 4,950 primary students

In the first randomization, 74 schools are assigned to the school-based SEL intervention and 75 to control. In the second randomization, 74 schools are assigned to the home-based SEL intervention and 75 to control.
Hopefully, each group contains approximately 37 schools: T1 (school-based SEL only),
T2 (home-based SEL only), T3 (combined school- and home-based SEL), and the control group (neither intervention).

Schoo-/home-based SEL intervention vs control, 74 schools per arm, an average of 35 students per school, an ICC of 0.05, 80% power, a 10% two-sided test, for test score, and baseline covariates explaining 50% of the variation (R2=0.5), MDE=0.084 SD; for SEL outcomes, R2=0.2, MDE=0.106 SD. With 37 schools per group, the MDE is 0.12 SD for test score, and 0.15 SD for SEL