Social-emotional learning, or SEL as it is often abbreviated, remains one of those topics where discussions veer off in all directions remarkably quickly. For some, it is almost the solution to everything that goes wrong in education: from bullying to learning deficits, and from well-being to citizenship. For others, it has become nothing more than a catch-all term that wastes too much time at the expense of “real knowledge”.

A new meta-analysis in Review of Educational Research attempts to make at least one part of that discussion more concrete: do SEL programs effectively lead students to exhibit more prosocial behaviour? So, specifically: more helping, collaborating, supporting, sharing, or considering others?

The researchers, led by Yu Hung, reviewed 66 studies involving more than 52,000 primary and secondary school students. Their general conclusion is quite clear: on average, SEL programs have a positive effect on prosocial behaviour. The calculated average effect size was approximately g = .24.

Now, 0.24 might not sound impressive if you just look at the number. But in real educational research, in real schools, with complex human interactions, this is certainly not insignificant. The authors themselves refer to researchers who state that effects around 0.25 can already be quite meaningful for school-wide interventions. Especially when it concerns behaviour that repeats daily in social contexts.

At the same time, this is yet another study where the nuance is particularly interesting. The spread between the studies was large. Some programs had little effect, while others had a fairly strong effect. The estimated true effects lie between approximately -.06 and +.53. This, therefore, also means that SEL does not work automatically. Some approaches seem to work much better than others.

One of the most striking findings of the meta-analysis is that “more” did not necessarily mean better. Surprisingly, programs with fewer than nine hours of explicit SEL instruction had greater effects on average than programs with more hours. Also, programs running longer than one school year performed less well on average than programs of approximately half a school year to one school year.

That feels somewhat counterintuitive. Much educational logic implicitly assumes that more time should automatically yield greater effect. But perhaps something else is at play here. Implementation fatigue, for example. Or the fact that long-term programs are harder to sustain consistently and with high quality. Or simply that teachers in overcrowded curricula experience limits on how many separate SEL lessons remain feasible. Of course, that does not mean that short and superficial is always better. But it does suggest that quality, focus, and integration are likely more important than simply adding extra hours.

Also striking was what else made no major difference. Effects did not differ significantly between primary and secondary education. Urban versus rural contexts also made little difference. Just like who delivered the program—teachers, researchers, or other supervisors—yielded comparable results on average. This also deviates from other educational themes.

But even there, we must remain cautious. Some categories simply contain a few studies. For example, there were remarkably few studies in secondary education and hardly any that used purely interactional approaches, focusing primarily on teachers’ daily interactions rather than on separate lessons.

Many SEL programs remain strongly curriculum-oriented: separate lessons, separate modules, separate exercises. While at the same time one can argue that prosocial behaviour originates primarily in daily culture, routines, and relationships, in how teachers respond to conflicts, how class discussions unfold, how collaboration is organised, and how adults themselves deal with respect, empathy, and boundaries. Perhaps that is also why some relatively limited interventions still show effects: because they succeed in changing something in daily interactions rather than merely adding extra lesson content.

Methodologically, this meta-analysis actually looks quite solid. The researchers used robust statistical techniques, investigated publication bias, and were remarkably cautious in their interpretations. That alone makes it a breath of fresh air in a domain where very grand claims sometimes circulate.

My impression after reading? This study certainly supports the idea that schools can influence prosocial behaviour. But at the same time, it implicitly undermines the simplistic idea that SEL is just an extra box that you have to teach “more” to get better people.