A Methodological Review for the Data Must Speak Positive Deviance Research

Insights from positive deviance, behavioural sciences, implementation research and scaling science

A boy at the billboard in a school in Niger.


The pandemic has aggravated a learning crisis and put global goals in jeopardy. And yet, even in the most challenging educational contexts, some schools outperform others located in similar contexts and with an equivalent level of resources. Why do these exceptional schools, known as ‘positive deviant’ schools, achieve improved outcomes in learning, retention, equity and gender equality?

Data Must Speak (DMS) – a global initiative implemented since 2014 – aims to address the evidence gaps to mitigate the learning crisis using existing data. DMS’s research component is co-created with ministries of education. It relies on mixed methods to generate knowledge, alongside practical lessons about ‘what works’, ‘why’ and ‘how to’ scale grassroots solutions for national policymakers and the broader international community of education stakeholders.

The research utilizes innovative and complementary approaches of positive deviance, behavioural sciences, implementation research and scaling science to identify and scale up behaviours and practices of ‘positive deviant’ schools. This methodological review presents key definitions, concepts and methodologies of those approaches to guide and inform the development and implementation of the DMS research at country level. By drawing on existing examples from research on education and other fields, this review also offers best practices and lessons learned from those approaches that can be used as a common reference and standard language for future application.

The DMS research is currently active in 14 countries: Brazil, Burkina Faso, Chad, Cote d'Ivoire, Ethiopia, Ghana, Lao PDR, Madagascar, Mali, Nepal, Niger, Tanzania, Togo, and Zambia. It is co-financed by the Jacobs Foundation, Hewlett Foundation, KIX (IDRC/GPE), NORAD, Schools2030, and internal UNICEF resources.

Document cover
Lorena Levano Gavidia; Cirenia Chavez; Alvaro Fortin; Luca Maria Pesando; Renaud Comba
Publication date
English, French