It Takes a Class: Peer Learning and Children Social Networks

We evaluate the effects of the adoption of a peer learning platform by teachers. This online platform helps students learn the bases of computer sciences, using two key pillars to foster interactions: (i) in-person peer help---every time students need help, they ask their classmates rather than their teacher, and (ii) in-person peer evaluations---every time students finish a task, they ask their classmates to evaluate them rather than their teacher.

The ministry recommends that teachers use the platform with their students 45 minutes per week.
The assignment of helpers and graders aims at enhancing the diversity of interactions within each classroom. When a student requests help or an evaluation, all other students see a pop-up image indicating that someone is seeking assistance or an assessment. However, they do not see who made the request. In addition, to ensure that students who rarely interact with others get an opportunity to do so via the peer learning platform, those who have frequently helped or graded their peers will experience a slight delay before seeing the pop-up. The first student to respond to a request for help or evaluation will be assigned to be the helper or grader.

2023-11-14

2024-07-06

This project investigates how the adoption of a peer learning platform affects four main outcomes:
(i) the number and diversity of work-related interactions between students of a class;
(ii) the density and diversity of students’ friendship networks;
(iii) students’ cooperation and students’ prosociality;
(iv) students’ well-being measured by their level of self-esteem, school satisfaction, frequency of positive and negative emotions, and prevalence of bullying.

See the pre analysis plan for the secondary outcomes.

To study the effects of peer learning on student social networks and their non-cognitive skills and well-being, we partnered with the French Ministry of Education to evaluate the effect of the adoption of a peer learning platform by teachers of primary school students (4th and 5th grades). The peer learning platform was designed by the “Ecole 42”, a private, non-profit and tuition-free IT school in France.

We plan to evaluate the effects of this platform adoption using a Randomized Controlled Trial (RCT) during the academic years 2023/2024. With the help of the French Ministry of Education, we will recruit 300 volunteer primary school teachers in Autumn 2023. 200 teachers will be randomly assigned to the treatment group---these teachers will be able to use the new peer learning platform from January 2024 onward---while 100 teachers will be assigned to the control group---they will only have access to the new peer learning platform in September 2024. Among treated schools, we will implement two treatment pathways, one in which teachers only use the peer learning platform for 45 minutes per week (100 classes), and another one in which teachers are encouraged to use peer learning outside of the platform and during their daily lessons (100 classes). Teachers will receive information on how to do so.

We will randomize the control status and the two treatment pathways at the school level: all volunteer teachers from the same school will be assigned the same status.

We will measure the causal effects of peer learning by comparing the evolution, over one academic year, of student outcomes in the treated and control groups. We will also explore whether the two treatment pathways have differential effects.

The experiment will continue during the academic years 2022/2023 with a larger sample of 300 schools with at least one class per school (at least 6,375 students). We will randomly assign schools to the treatment and control to avoid spillover effects between treated and control classes of a given school.

During the two years of the experiment, classes or schools will be assigned to the treatment or control condition using a stratified random lottery with strata defined on the basis of school location and school status.

Not available

Computerized randomization using Stata.

Schools

Yes

We aim at recruiting 250 schools with an average of 1.2 volunteer teachers per school.

If we succeed in recruiting 250 schools with an average of 1.2 volunteer teachers per school and an average of 20 participating students per class, our expected sample size will be of about 6,000 students. Scale-up phase (2022/2023): With 25 students per class, an expected consent rate of 85%, 300 schools and at least one class per school, our final sample will contain at least 6,375 students.

We will assign one third of volunteer schools to the control group, one third to the first treatment pathway, and one third to the second treatment pathway. If we succeed in recruiting 250 schools, we will therefore assign about 83 of them to each status.

We computed the minimum detectable effects on the main outcomes for which we have data from the pilot phase of the experiment: -Number of classmates with whom students have worked during the previous week: +0.34. -Probability that students have worked with a classmate of the opposite gender: +0.05. -Probability that students have worked with a classmate of the opposite social background: +0.06. -Number of close friends: +0.62. -Probability that students are isolated: +0.02. -Probability that students have a close friend of the opposite gender: +0.05. -Probability that students have a close friend of the opposite social background: +0.03. -Number of coins that students contribute to the common moneybox when playing with random students from different schools: 0.17. -Probability that students exhibit a high level of cooperation when playing with random students from different schools: 0.06. The minimum detectable effects are computed with a sample size of 250 schools with 24 students per school, and using information from the pilot experiment on the mean, standard deviation and intra-class correlations of the above-mentioned outcomes.