A flexible, integrated, and scalable approach to early social-emotional development: Experimental evidence from Brazil

We propose to conduct a randomized evaluation of a set of ready-made games and activities (“SEL Kernels”) that teachers can implement at various points during the day (e.g., before/after recess) to practice cognitive, emotional, and interpersonal skills, as well as character and mindset.

The intervention includes: two days of training (one at the start and a “refresher” later), an online course that teachers can complete at their own pace (which earns them a certification from the Ministry of Education), 30+ cards with the games/activities, support via WhatsApp, and online videos modeling implementation. Teachers will be expected to use the activities at least three times per week and twice per day.

The SEL Kernels materials were developed by Prof. Stephanie Jones. They include quick, fun, teacher-led exercises that can be easily integrated into daily routines and transitions. They are designed to develop children’s cognitive, emotional, and interpersonal skills, as well as character and mindset. After each game, teachers and students are expected to engage in discussions to build children’s metacognition, promote the use of a shared vocabulary around the skills, and identify opportunities to apply these skills.

We have chosen this intervention for three main reasons: it does not require curriculum reforms (which take a long time and are often resisted by stakeholders), it is easy to implement (it is distilled to a set of instructions in “flash cards”), and it does not take up a large share of lesson time (and thus do not need to compete with other scheduled activities). Further, while it has yielded promising results, and it has already been implemented in Sao Paulo and other parts of Brazil, it has not yet been evaluated by itself at scale in a low- or middle-income country (see section on potential policy impact). Finally, the materials and online teacher training course are already available in Portuguese, which will enable us to move quickly with the evaluation once it is approved.

We hypothesize that the intervention will improve the frequency and type of student-teacher interactions; in turn improving the five skill domains directly targeted by the intervention (cognitive, emotional, and interpersonal skills, as well as character and mindset); reducing the incidence of students’ behavioral problems; and potentially raise students’ numeracy and literacy skills (see, e.g., Barnes, Bailey, and Jones 2020, Jacob and Parkinson 2015, Jones, Bailey, and Jacob 2014).

2026-05-01

2026-09-30

Executive function (inhibitory control, short-term memory, and cognitive flexibility), emotion recognition, risk preferences, cooperation, trust, and grit

Numeracy and literacy skills

We plan to include 200 public pre-schools in the two municipalities that we have pre-selected in Sao Paulo (Bauru and Guarulhos). In pre-school centers with more than one classroom, we will randomly select only one classroom. In classrooms with more than 15 students, we will randomly select 15 students (see the section of the application on statistical power for the explanation for this number). Thus, the evaluation will ultimately include 200*1*15=3000 public pre-school students.

We will randomly assign 200 pre-school centers to receive the intervention during or after the study, which will allow us to estimate its causal effect by comparing the control and treatment groups at endline (end of the year), adjusting for measurements at baseline (start of the year). We will stratify our randomization by municipality to ensure all municipalities can receive the intervention in some schools and to improve the comparability of control and treatment centers at baseline.

We plan to undertake two main rounds of data collection: a pre-intervention baseline, which will include students’ demographics and social-emotional skills; and a post-intervention endline, to include student-teacher interactions and students’ behavioral problems, social-emotional skills, and numeracy and literacy skills.

Not available

Randomization done in office by a computer

Pre-school center

Yes

200 pre-school centers

3000 public pre-school students

100 control and 100 treatment pre-school centers