Compassion Training in Adolescents: A Cluster-Randomized Trial with 12-Month Follow-Up (Sun, 2026)
Adolescence is a time when the ability to manage emotions and the sense of right and wrong are both developing quickly, and schools are increasingly asked to support both. Compassion-based programs are one candidate, yet few trials have followed students for long after the lessons end. A cluster-randomized trial in Frontiers in Psychology, led by Yutong Sun of Nanjing Normal University, tracked students for a full year.
The study enrolled 876 students aged 12 to 18 from 34 classes in four secondary schools in China. Whole classes, not individual students, were randomly assigned: 17 classes (447 students) received compassion training, and 17 classes (429 students) received relaxation training as an active comparison. Both arms had eight weekly 45-minute sessions plus 10 to 15 minutes of daily home practice. Students completed questionnaires before the program, right afterward, and at three and twelve months. The measures covered emotion regulation, using the Emotion Regulation Questionnaire, and prosocial moral development, a study-specific measure that combined prosocial behavior, moral reasoning and moral identity.
Compassion training produced larger gains than relaxation training on both outcomes at every follow-up point, and the differences were statistically significant each time. Immediately after the program, the advantage was about half a standard deviation for emotion regulation and slightly less for prosocial moral development. At twelve months it had narrowed to roughly a third and a quarter of a standard deviation, which amounted to about two to two and a half points on each scale.
The pattern over time was one of a steep rise during the program followed by partial fading. Both groups drifted down afterward, and the compassion group dropped more, but its scores stayed above where students started and above the relaxation group at twelve months.
Interestingly, the study also tested whether the two outcomes strengthened each other. Better emotion regulation predicted later prosocial moral development at every interval, while the reverse held only in the first two. Since the compassion and relaxation groups did not differ significantly in these links, the idea that compassion training specifically tightens the connection was not supported.
The findings come with clear caveats. About 40% of students had dropped out by twelve months, and those who stayed differed from those who left in age, sex and socioeconomic status. Only four schools in one country took part, the prosocial moral development measure had not been validated before, and program delivery was checked in only about a quarter of sessions. The author calls the results tentative.
Even so, a lasting advantage over an active comparison is a useful signal for school-based programs. Future studies will need larger and more varied samples, validated outcome measures and lower attrition. Readers interested in how compassion toward oneself may relate to attitudes toward others can also see our recent essay on self-compassion and social attitudes.
Reference and link: https://doi.org/10.3389/fpsyg.2026.1969277
