UNICEF - Learning Together

Learning Together: Reflections from co-creating education research across 15 countries DATA MUST SPEAKUNICEF, the United Nations agency for children, works to protect the rights of every child, everywhere, especially the most disadvantaged children and in the toughest places to reach. Across more than 190 countries and territories, we do whatever it takes to help children survive, thrive, and fulfil their potential.

UNICEF Office of Strategy and Evidence – Innocenti accelerates progress for children by working to ensure that policies and programming are informed by high-quality evidence. As the global custodian of child-related official statistics, it works closely with governments and partners to strengthen national statistical systems. Through world-leading data, research and foresight, it underpins UNICEF’s global leadership on children’s rights and serves as the organization’s hub for setting strategy and monitoring programmes. With the active engagement of young people and other partners, it supports advocacy and dialogue aimed at improving the lives of children everywhere. Published by UNICEF Office of Strategy and Evidence – Innocenti Via degli Alfani, 58 50121, Florence, Italy Web: unicef.org/innocenti Email: innocenti@unicef.org Social media: @UNICEFInnocenti on BlueSky , Instagram, LinkedIn, and YouTube Suggested citation Herman, Ryan S., and Jessica L. Bergmann, Learning Together: Reflections from co-creating education research across 15 countries, UNICEF Innocenti, Florence, 2026.

© United Nations Children’s Fund (UNICEF), 2026

Cover photo: © UNICEF/Lao People’s Democratic Republic Except for the UNICEF logo, photos and where another © appears, this work is licensed under CC BY 4.0. The UNICEF logo and enclosed photos may only be used in exact copies of the publication.

Where another © appears, you must ask the listed copyright holder to use the item outside exact copies of the publication.© UNICEF/UNI623856/DejonghAcknowledgements The Data Must Speak (DMS) positive deviance research is a collaborative effort made possible by

Where another © appears, you must ask the listed copyright holder to use the item outside exact copies of the publication.© UNICEF/UNI623856/DejonghAcknowledgements The Data Must Speak (DMS) positive deviance research is a collaborative effort made possible by ministries of education, UNICEF Country and Regional Offices, UNICEF Innocenti – Global Office of Research and Foresight, and key partners from education sectors across the 15 countries where the research has been implemented. The development of this report was led by Ryan Shawn Herman and Jessica Bergmann. The authors would like to thank all team members who engaged in the critical reflections and discussions to arrive at this report, especially Erica Aiazzi, Michèle Boujikian, Luisa Cardoso, Renaud Comba, Ana Luiza Fonseca Minardi and Alexis Le Nestour. Special thanks also go to Jessica Bergmann and Renaud Comba, who oversaw the implementation of the research; to the additional members of the DMS research team at UNICEF Innocenti for their contributions; to the administrative and communications colleagues who made the production of this report possible; and to Linda Jones for her overall guidance of the DMS research and the education unit at UNICEF Innocenti. The authors would also like to thank Matt Brossard for his conception, vision and creativity, as well as for his general support for Data Must Speak’s work. Finally, sincere thanks go to a coalition of donors who have supported the DMS positive deviance research and our learning journey from the outset: Global Partnership for Education Knowledge and Innovation Exchange, a joint endeavour with the International Development Research Centre (Canada), Hewlett Foundation, Jacobs Foundation, Norad, Schools2030 initiative (Aga Khan Foundation) and UNICEF internal resources. The views expressed herein do not necessarily represent those of the donors or their respective Boards. Additionally, the findings, interpretations and conclusions expressed in this paper are those of the authors and do not necessarily reflect the views of UNICEF. Data Must Speak research coalition of donors:© UNICEF/UN0438268/RamasomananaAcknowledgements Why this report?

findings, interpretations and conclusions expressed in this paper are those of the authors and do not necessarily reflect the views of UNICEF. Data Must Speak research coalition of donors:© UNICEF/UN0438268/RamasomananaAcknowledgements Why this report? The Data Must Speak positive deviance research Reflections from our research journey Reflection 1: Consistently include school actors as change agents and users of evidence Reflection 2: Intentionally adapt co-creation to partners’ needs, capacities and contexts Reflection 3: Dedicate ongoing technical assistance to sustainably improve education data systems Reflection 4: Integrate and value mixed-methods approaches for deeper insights Reflection 5: Strengthen country ownership for lasting scaling and policy impact Reflection 6: Plan for adapting education research for use in conflict settings Conclusion 4 7 16 19 23 26 29 32 14–15 36–38 9–13© UNICEF/UNI448445/Laxmi Prasad Ngakhusi

1. Why this r eport?Learning Together: Reflections from co-creating education research across 15 countries

8 UNICEF Innocenti – Global Office of Research and Foresight aims to generate quality evidence that can improve decision-making, with the hope that better decisions can result in real impact on children’s learning and well-being. Within the education unit at UNICEF Innocenti, we believe that co-creation with partners throughout the research life cycle is key to ensuring that evidence remains demand-driven, aligned with local needs and, ultimately, used. Even the most carefully designed research plans can encounter setbacks, despite thorough risk analyses and established mitigation measures. Key personnel may transition to new work, data may become lost or corrupted, funding can run short, and changes in country leadership can shift priorities and focus. Sometimes, the reflection questions asked at the end of a project cycle are the ones that could have guided better planning, strengthened risk anticipation and mitigation, and achieved closer collaboration between all actors from the start.

and changes in country leadership can shift priorities and focus. Sometimes, the reflection questions asked at the end of a project cycle are the ones that could have guided better planning, strengthened risk anticipation and mitigation, and achieved closer collaboration between all actors from the start. Throughout the four years of the Data Must Speak (DMS) positive deviance research , our team has come together on many occasions and asked the following questions: ■ What have we learned from our experience as a diverse team of researchers, working across various countries to co-create the DMS positive deviance research? ■ How can we do better now, and what could we do better the next time? Why this report? While these conversations are not always easy, they are invariably productive – and critical for creating a culture of learning and that encourages the kinds of risk-taking, creativity and continuous adaptation that are needed to overcome the current challenges facing education systems globally. These conversations have driven the development of this report, which shares six of the DMS positive deviance research team’s reflections from co-creating and implementing the research across 15 countries. For each reflection, the report presents an overview of what the DMS research team did and where we could have done things differently, as well as insights for ourselves and others on what considerations could be made. We took inspiration from organizations such as Engineers without Borders Canada , BRAC and One Acre Fund , all of which, in different ways, have generated and shared reflections that can guide others in their own work. We hope this report contributes to building a culture of knowledge-sharing, inviting those working in education research – and in other sectors, too – to reflect on and share their lessons learned so that, together, we can continue to grow in developing more thoughtful, inclusive and effective research practices.2. The Data Must Speak positive de viance

research

those working in education research – and in other sectors, too – to reflect on and share their lessons learned so that, together, we can continue to grow in developing more thoughtful, inclusive and effective research practices.2. The Data Must Speak positive de viance

research © UNICEF/UNI408813/Érico HillerLearning Together: Reflections from co-creating education research across 15 countries 10 Launched in 2014, the DMS initiative was designed to help countries unlock the power of existing data to expand access to education and improve learning for all. Although education systems around the world invest in collecting large amounts of data from their schools, these data could be better leveraged to support policymaking. DMS strives to ensure that data and evidence drive policy decisions, and that government officials at all levels are empowered to analyse and use data to manage education systems efficiently. In 2020, the DMS initiative expanded to include the positive deviance research component, which aims to identify and amplify effective practices and behaviours that contribute to improved educational outcomes. The DMS positive deviance research leverages existing data to identify positive deviant schools – schools that outperform their peers despite operating in similar contexts and with similar resources – and the practices and behaviours that contribute to their success. The DMS positive deviance research was implemented in 15 partner countries in collaboration with ministries of education (MoEs), development partners, local academics, civil society organizations (CSOs), and school actors including head teachers, teachers and decentralized administration officers (see Figure 1). The Data Must Speak positive deviance research

Source: A uthors

Note: T his map does not reflect a position by UNICEF on the legal status of any country or territory or the delimitation of any frontiers.

Figure 1: DMS positive deviance research partner countries Brazil (State of Maranhão)

Source: A uthors

Note: T his map does not reflect a position by UNICEF on the legal status of any country or territory or the delimitation of any frontiers.

Figure 1: DMS positive deviance research partner countries Brazil (State of Maranhão) ChadCôte d’Ivoire Ethiopia Ghana Burkina Faso Nepal Mali Niger Togo United Republic of Tanzania (Mainland and Zanzibar) Zambia Lao People’s Democratic Republic Viet Nam11

Learning Together: Reflections from co-creating education research across 15 countries

Figure 2: Stages of the DMS positive deviance research The DMS research includes multiple stages, all co-created and co-implemented with in-country partners (see Figure 2 ): ■ Stage 1 involves a quantitative analysis of education management information system (EMIS) data and, where available, learningoutcome data. This analysis is used to identify contextual factors and school resources that are correlated with better school performance, such as head teacher and teacher characteristics (gender, qualifications, etc.), as well as school characteristics such as infrastructure and class sizes. At this stage, only a small part of school performance can be explained by the available administrative data. ■ In Stage 2, a sample of positive deviant schools – those that are outperforming their peers despite operating in similar contexts and with similar resources – and comparison schools is identified. ■ In Stage 3, data are collected from positive deviant and comparison schools to identify the practices and behaviours that may contribute to differences in student outcomes. ■ In Stage 4, implementation research, scaling science and behavioural sciences are leveraged to support MoEs in identifying opportunities to scale these practices and behaviours to more schools in the country. ■ Stage 5 focuses on dissemination at the country level to support uptake and

differences in student outcomes. ■ In Stage 4, implementation research, scaling science and behavioural sciences are leveraged to support MoEs in identifying opportunities to scale these practices and behaviours to more schools in the country. ■ Stage 5 focuses on dissemination at the country level to support uptake and use for policymaking, while also using the evidence to inform global dialogue and advocacy on improving learning. For more information about our research approaches, please see our series of methodological briefs . Stage 3 Understanding school-level positive deviant behaviours and practices Stage 2 Identification of positive deviant schools and school typology Stage 1 Analysis of resources and context associated with school performance Stage 4 Investigating levers for optimum scale Stage 5 Country-level knowledge use and global mobilization (Behavioural sciences)(Quantitative research) (Positive deviance) (Participatory implementation research & scaling science) Source: A uthors12 Learning Together: Reflections from co-creating education research across 15 countries The DMS positive deviance research shows the power of existing country data sets to improve education systems. The research uses these existing data to identify the local education solutions that are already proven to be cost-effective and readily implementable, and that may be more likely to sustainably scale since they are already rooted in local realities. It has generated valuable knowledge by identifying the inputs, practices and behaviours that are associated with positive school performance in specific contexts. Created by I Putu Dicky Adi Pranatha from Noun Project Created by Romaldon from the Noun Project Created by Larea from Noun Project As a research team, we are proud of what we have accomplished collectively with our country partners over the past four years: Supporting research co-creation and capacity development for MoEs and other education partners through over 50 technical sessions and capacity-building workshops Collecting data-rich insights from more

As a research team, we are proud of what we have accomplished collectively with our country partners over the past four years: Supporting research co-creation and capacity development for MoEs and other education partners through over 50 technical sessions and capacity-building workshops Collecting data-rich insights from more than 1,100 schools and 11,000 school-level and decentralized education actors across 12 education systems through primary data-collection activities to identify positive deviant practices and behaviours that improve education outcomes for children Producing more than 75 country-level research reports and global public goods, which have been used to inform education sector plans, education sector analyses, Global Partnership for Education Partnership Compacts, and other funding proposals and/ or partnership agreements for education sector strengthening and programmes13Learning Together: Reflections from co-creating education research across 15 countries Created by Muhammad Shabraiz from Noun Project Created by miftakhudin from Noun Project Created by Andika Cahya Fitriani from the Noun Project The research has uncovered a range of effective practices and behaviours used by positive deviant schools to improve education outcomes, including: Improving teachers’ skills by encouraging school-based collaboration, mentoring and training, providing targeted feedback and coaching, and leveraging local oversight and shared accountability for better teaching practices Adopting student-centred pedagogies , including through active learning strategies, cooperative learning and peer-support networks, and differentiated instructional practices Preserving available learning time by monitoring and improving teacher attendance, while also reducing student absenteeism and recovering lost learning Leveraging community support , with communities contributing financially to school resources and functioning, as well as providing operational support for the school Strengthening collaboration through effective communication , including by promoting internal communication at the school, and engaging parents and caregivers in school life and student learning Cultivating a safe, supportive school environment by ensuring children’s safety, amplifying their voices and agency, and

Strengthening collaboration through effective communication , including by promoting internal communication at the school, and engaging parents and caregivers in school life and student learning Cultivating a safe, supportive school environment by ensuring children’s safety, amplifying their voices and agency, and ensuring a supportive environment for all school actors

T he full report presenting the synthesis of results spanning DMS partner countries is available online. 1 In total, the DMS positive deviance research was implemented in 15 countries across three continents. However, three countries (Burkina Faso, Niger and Viet Nam) did not advance to the primary data-collection stage to identify positive practices. Reflections from the research include experiences drawn from all 15 countries.Learning Together: Reflections from co-creating education research across 15 countries 14

3. Reflections from our r esearch journey

© UNICEF/UNI413207/Tesfaye15

Learning Together: Reflections from co-creating education research across 15 countries As we look back on the implementation of the DMS positive deviance research, we share six key reflections from our work with fellow researchers, policymakers and governments, implementing agencies, development partners, and donors.

These reflections, and this report more broadly, were developed through a co-creation process with the DMS research team. In the first stage, reflections were synthesized across the various sensemaking meetings, team retreats and weekly team meetings that took place throughout the implementation of the research between 2021 and 2024. Initial reflections were discussed over three different working sessions with the DMS team in 2023–2024, providing an opportunity to prioritize, revise and refine the six reflections. Individual consultations with team members were conducted to solicit further information and reflections across the various countries, and multiple reviews were carried out, and meetings with team members held, to ensure the information presented accurately reflected country-level nuances and experiences.

and refine the six reflections. Individual consultations with team members were conducted to solicit further information and reflections across the various countries, and multiple reviews were carried out, and meetings with team members held, to ensure the information presented accurately reflected country-level nuances and experiences. Reflections from our research journey Each of the six reflections presented in this report follows the same structure: What we did: ■ The concrete steps and activities the DMS research team carried out in practice throughout the planning, implementation and/or dissemination of the research What we could have done differently: ■ The steps that we did not always take and what, with the benefit of hindsight, we believe could have been done differently Insights for ourselves and others: ■ What considerations, based on our experiences, could be made earlier in the research planning phase, or adaptively throughout the implementation of the research, for future improvement ■ Together, these reflections not only informed iterative adaptations as the research was scaled to new countries, but also provided critical takeaways that will continue to inform the DMS team’s research work moving forward.Learning Together: Reflections from co-creating education research across 15 countries 16 © UNICEF/UNI490638/Studio 19 Reflection 1 Consistently include school actors as

change agents and users of evidence Created by elastic1studio from the Noun Project17

Learning Together: Reflections from co-creating education research across 15 countries Reflection 1: Consistently include school actors as change agents and users of e vidences At the heart of the DMS positive deviance research lies the identification of the practices and behaviours – adopted by head teachers, teachers, parents and community members – that contribute to positive school performance. Throughout our research, we integrated and amplified the voices of school actors in various ways.

But believe we could have engaged schoollevel actors even more meaningfully.

What we did: School actors, including head teachers,

head teachers, teachers, parents and community members – that contribute to positive school performance. Throughout our research, we integrated and amplified the voices of school actors in various ways. But believe we could have engaged schoollevel actors even more meaningfully.

What we did: School actors, including head teachers, teachers, decentralized administration officers and other in-country stakeholders, were actively involved in the co-creation of the research, including in the following phases: ■ Research kick-off: Head teachers, teachers, decentralized administration officers, MoE officials and other in-country partners participated in co-creation workshops to help align our research questions, design a workplan, and create data-collection instruments that reflected school-level realities, interests and needs. ■ Data collection: School actors were engaged through primary data collection, with the goal of gathering their expertise and understanding their perceptions, mindsets and beliefs around what contributed to their school’s performance. In the Lao People’s Democratic Republic, pre-service teachers were trained as enumerators, fostering even stronger teacher engagement and ownership. ■ Research validation: After data had been collected and analysed, head teachers, teachers and decentralized administration officers participated in the final validation workshop and reflected on how they could take the results forward in their schools.

What we could have done differently: One important step missing from our research process was to consult with school actors, as users of evidence, and to ask how the research results could be translated into formats most useful to them. During co-creation workshops, we could have asked school actors – including head teachers, teachers and decentralized administration officers – where they got their information and evidence from, and how we could adapt some of the DMS research outputs to their preferred channels or formats. By translating the DMS evidence specifically for these actors, we could have been more successful in bridging the evidence gap that tends to exist between researchers,

information and evidence from, and how we could adapt some of the DMS research outputs to their preferred channels or formats. By translating the DMS evidence specifically for these actors, we could have been more successful in bridging the evidence gap that tends to exist between researchers, policymakers and practitioners. Including, in research proposals, activities to monitor dissemination efforts and the extent to which they have supported uptake by these users is also something we would consider in the future. School leader Decentralised administration officerLearning Together: Reflections from co-creating education research across 15 countries 18 Insights for ourselves and others: Exploring more effective channels for communication with school actors could have supported increased and sustained engagement. We learned that WhatsApp/ Telegram groups were commonly used to facilitate within-school, school-to-school and/or school-to-home communication. Collaboration mechanisms, including through school clusters or support from decentralized administration officers, also support school actors in exchanging best practices and solutions for existing challenges. In addition to engaging school actors in our co-creation workshops, we could have more intentionally thought through ways to tap into these broader networks to gather insights and disseminate research results. ■ Plan for the continuous and meaningful engagement of school actors from the outset, ensuring consideration is given to both time and financial resources to sustain engagement. Undertake a mapping of various networks, platforms and tools that could be leveraged, and work with in-country partners to determine how best and most meaningfully to engage school actors without creating additional burden. ■ Engage with school actors reflecting diverse contexts and realities. Owing to various constraints, it is often easier to engage with school actors from the capital city than from elsewhere in the country. While their perspectives are valuable, they do not represent the diverse contexts, needs and circumstances of all schools. At the planning stage, it is therefore important to clearly define which school actors to engage with and to ensure they are included. This includes budgeting adequate

city than from elsewhere in the country. While their perspectives are valuable, they do not represent the diverse contexts, needs and circumstances of all schools. At the planning stage, it is therefore important to clearly define which school actors to engage with and to ensure they are included. This includes budgeting adequate time and resources for their participation, and, where necessary, meeting stakeholders where they are through localized consultations. ■ Be intentional and ask school actors how they access, use and engage with evidence and data before developing a research dissemination plan. Evidence can only be effectively translated to the school level if this is done hand-in-hand with end users. The DMS positive deviance research is exploring various evidence translation approaches in Ghana, involving school actors throughout this new research to understand what works in strengthening evidence use in schools.19

Learning Together: Reflections from co-creating education research across 15 countries

© UNICEF/UN0275125/Quarmyne Reflection 2 Intentionally adapt co-cr eation to partners’ needs, capacities and contexts Created by symbolic from Noun ProjectLearning Together: Reflections from co-creating education research across 15 countries 20 The co-creation approach adopted across our 15 partner countries helped to ensure that the research remained contextually relevant and useful for stakeholders. Our approach to co-creation evolved over time and was adapted across countries as we learned where we could more meaningfully engage our partners, who play a central role in addressing education challenges within their own contexts.

What we did: We collaborated closely with various stakeholders, both within and outside the central MoE, throughout our research life cycle. We did this by creating two groups of

country-level stakeholders and partners: ■ The Consultative Working Group comprises 20–25 members, including decision-makers from the MoE, as well as school actors, development partners and CSOs, and local academics. This group provides highlevel input into the research design, methodologies, timelines and uptake strategies, and ensures the research reflects the various interests and

decision-makers from the MoE, as well as school actors, development partners and CSOs, and local academics. This group provides highlevel input into the research design, methodologies, timelines and uptake strategies, and ensures the research reflects the various interests and needs at the country level. ■ The Technical Working Group is a smaller, more specialized group composed of monitoring and evaluation experts, analysts and statisticians from the MoE. This technical team is involved in the day-to-day implementation of the research, including data-merging and analysis, sampling and data collection, and report-writing. In four participating countries, we also had a dedicated researcher, based in-country, to support research co-creation and implementation.What we could have done differently: Discussing expectations for co-creation with our partners from the outset could have helped us better adapt our co-creation approach for each country. Our experiences taught us that there was no one-size-fits-all approach to co-creation. We experimented to find the proper balance for building ownership and capacity with partners tailored to their specific needs, capacities and interests. For example, for DMS primary data-collection activities, some MoEs hired and deployed their own enumerator teams, building skills for future, potentially ministryled research exercises. In other countries, MoE colleagues reviewed technical proposals when contracting data-collection firms and oversaw field activities, as they were more likely to contract these services again in the future. We could have better understood our partners’ expectations, and thus ensured better adaptation to each country’s needs, by asking what time they had available, how it could best be used, and whether the intent was to support capacity-building or sustainability. Reflection 2: Intentionally adapt co-creation to partners’ needs, capacities and conte xts21

Learning Together: Reflections from co-creating education research across 15 countries Rather than relying on intensive workshops

best be used, and whether the intent was to support capacity-building or sustainability. Reflection 2: Intentionally adapt co-creation to partners’ needs, capacities and conte xts21

Learning Together: Reflections from co-creating education research across 15 countries Rather than relying on intensive workshops as the primary modality for co-creation, we could have