Within-School Randomization Spillover
RCTClinical Trial
Individual-level randomization of students into treatment and control groups within shared educational facilities creates severe peer spillover effects and social contamination.
Picture this
Assigning students within the same school to separate treatment and control groups is like giving medicine to half the passengers on a bus while leaving the rest unmedicated in full view of each other. Daily interaction between passengers leads to shared information, resentment, and skewed outcomes across both groups.
What the evidence says
Individual student randomization within schools amplified social stigmatization and generated unavoidable treatment spillovers across treatment and control arms, contributing to zero-session attendance rates of 70.0% in CBT and 97.0% in math tutoring.
- Who was studied
- N = 5,274 students across 20 public federal upper secondary schools in Mexico City.
- How
- Two-step randomized controlled trial (school-level selection followed by student-level randomization within schools) paired with a qualitative process evaluation.
What to do
Utilize cluster randomization at the school or classroom level rather than individual student randomization when evaluating behavioral interventions in public school settings.
From the source
"All these complex implementation challenges were compounded by the student-level randomization creating treatment and control groups within each school... Given this, as well as the fact that teachers who knew the students were directly involved, some level of spillover across treatment groups was probably unavoidable."
Addressing High School Dropouts with a Scalable Intervention: The Case of PODER