ERCT: A Simple Standard for Making Sense of Educational RCTs
Twelve questions for reading educational randomised controlled trials more critically.
ERCT: Educational randomised controlled trials
ERCT gives us a consistent way to read educational randomised controlled trials.
A trial can report an impressive effect and still leave important questions unanswered. How long did it run? What did the comparison group receive? Was the outcome an independent exam or a test designed around the intervention?
ERCT is a practical checklist of twelve criteria, organised into three levels. It helps make those questions visible. It is not a universal certification of a study, and a checklist cannot replace reading the evidence.
Twelve questions, three levels
- Class randomisation
- Randomise at class level to address classroom spillovers.
- Exam assessment
- Use meaningful exam-based outcomes.
- Term duration
- Study the intervention over at least a term.
- Documented control
- Describe the comparison condition.
- School randomisation
- Consider school-level allocation.
- All exams
- Look for effects beyond the target subject.
- Year-long duration
- Observe a full year of learning.
- Balanced control
- Consider whether the control is a fair comparison.
- Graduation tracking
- Follow longer-term educational outcomes.
- Reproduction
- Look for reproduced results.
- Independent conduct
- Distinguish independent evaluations.
- Preregistered protocol
- Specify the study plan before seeing results.
Articles about educational evidence
Education as Medicine
An argument for testing educational practice with the same seriousness we bring to medical evidence.
The 2 Sigma Problem and the AI Tutor Dream
What the two-sigma tutoring claim tells us—and what the original evidence leaves unresolved.