How some villages in Northern Ghana keep their forest while everyone else loses theirs
We are genuinely honoured to have been part of this project from beginning to end. As GIZ Data Lab, we have accompanied the Data Powered Positive Deviance (DPPD) method over the years as a whole and cornerstone of our work, we designed and facilitated the DPPD Masterclass out of which the team around Anthony and Joy eventually formed which carried out a promising use case study in Northern Ghana. We have supported the team continuously ever since, both technically and financially, and now the study has been published in the Environmental Management journal (https://link.springer.com/article/10.1007/s00267-026-02566-2). This fills us with gratitude, because it shows how a training idea grew into a genuine, peer-reviewed piece of research with real impact on the ground.
Fittingly, independent of this specific case, we also once took a step back and looked at what seven years of DPPD work at the GIZ Data Lab add up to as a whole. Our colleagues Andi and Basma captured this structural review of DPPD practices in two separate blog posts, which we link here:
- Blog post 1: https://dppd.medium.com/reviewing-the-data-powered-positive-deviance-method-what-we-found-part-1-ca1345efe2a8
- Blog post 2: https://dppd.medium.com/reviewing-the-data-powered-positive-deviance-method-what-we-suggest-part-2-a09027bc7f7d
The starting question
In Ghana’s Upper West Region, many villages’ livelihoods are closely tied to trees. Trees provide firewood, fodder, food, shade and income, and they protect soils against erosion, especially in a region with irregular rainfall and long dry seasons. And yet tree cover is disappearing in many places, driven by agriculture, firewood extraction, bushfires and overgrazing.
What is striking is that neighbouring villages with very similar starting conditions perform quite differently. Some retain their forest cover, others lose it steadily. This exact difference was the starting point for the study by Anthony Kwabena Sarfo, Michael Darko Ahwireng, Raphael Nawrotzki, Basma Hassan, Nelson Nyabanyi N-yanbini, Robin Nowok and Joy Heitlinger.
The methodological approach
The study uses Data-Powered Positive Deviance (DPPD). The core idea behind classic Positive Deviance is out there for some time and originally comes from health research. It assumes that within any group there are actors who, despite facing the same constraints as everyone else, achieve noticeably better outcomes than average, and that their behaviour can hold solutions that might work for others too. DPPD extends this approach by bringing in systematic data, in this case satellite data, to reliably identify these outliers in the first place.
Concretely, the study proceeded in three phases.
First, the team checked whether the problem was actually suited to this approach. Forest conservation is strongly behaviour-driven, people decide every day whether to leave a tree standing or cut it down, which is a good precondition for DPPD.
The second phase was about finding the outliers. The team analysed Sentinel-2 satellite imagery from 2018 to 2023 and derived the NDVI vegetation index from it, which allows conclusions about the density and health of plant cover. To filter out seasonal noise from cropland, the team deliberately used images from the dry season in December, when crops have already withered and the remaining green signal mostly comes from trees. Because ecological conditions vary from place to place, villages were first grouped into comparable ecological clusters through clustering, and only then were rates of change compared within these clusters. This produced a ranking of 200 villages, from which the 12 villages with the strongest forest gain and the 13 with the strongest loss were selected. Through propensity score matching, comparable pairs of positive and negative outliers were then formed, resulting in 10 so-called Positive Deviant villages and 10 Negative Deviant villages for the fieldwork.
In the third phase, the team went out into the field. In all 20 villages, focus groups were held with traditional authorities and with the wider village population, complemented by in-depth interviews with individuals actively engaged in forest conservation. In total, this amounted to 40 focus groups and 40 individual interviews, conducted in the local languages Dagare, Waali, Brifo and Sisali.
What positive deviance villages do differently
The analysis reveals a pattern. In villages with stable or growing tree cover, trees are seen as an economic asset. Cashew, mango, shea and moringa generate income, and are therefore planted and cared for rather than cut down. In the comparison villages, by contrast, trees are more often seen as a free resource for charcoal and timber. One quote from a focus group captures this well, people know that cutting trees isn’t good, but everyone does it anyway.
Cultural practices also play a role. In four of the ten positive deviance villages, actively maintained sacred groves and taboos still forbid the felling of certain trees. Where these beliefs have eroded, this protective mechanism disappears too.
A third pattern concerns land surveying. Because formal property boundaries are largely absent across much of Northern Ghana, some villages use trees to mark farmland. This reduces land disputes while simultaneously increasing tree cover, a double benefit from a single practice.
Fourth, functioning local institutions matter. Six of the ten successful villages have active forest committees that report violations to the chief and enforce sanctions. In the comparison villages, such committees often exist only on paper, or not at all.
And finally, knowledge sharing plays a role, whether through family tradition, exchange with other villages, or NGOs and government programs.
The Golden Mix
Out of these thematic areas, the team distilled three practices that, based on criteria such as transferability, cost, sustainability and local acceptance, are best suited for scaling to other villages. This combination is referred to in the study as the “Golden Mix”.
First, the economic use of tree products, for instance through processing shea or cashew. Second, the use of trees to mark property boundaries. Third, building forest committees for local self-governance.
What is notable about this Golden Mix is that it wasn’t invented from the outside, but derived from what already works in the successful villages. That is exactly the core idea of Positive Deviance, you don’t look for an ideal theory, you look for what is already demonstrably working on the ground.
What came out of it
The study doesn’t stop at analysis. The Golden Mix is now being implemented, with support from the GIZ Data Lab and the Sector Network Rural Development Africa, in the ten negative deviance villages, funded through the Climate and Nature Working Group.
For us at the Lab, this is exactly the moment we’ve been working toward as we keep developing the DPPD format over the years. A method that finds people who, under difficult conditions, have already found good solutions, and then uses that to let others benefit from them. That this has now turned into a scientific publication and a concrete implementation across ten villages in Northern Ghana is something we’re particularly happy about.
Acknowledgements
We sincerely want to thank the immensely motivated and well-structured team around Anthony, who brought clear, evidence-based knowledge into the shaping of this activity over the years. Without Anthony’s direction and oversight in steering the whole process from an initial idea to a scientific publication, this project wouldn’t have been the same. A huge thanks also goes to all advising and participating colleagues at GIZ Ghana and headquarters, who supported, co-funded, and carried this process for so long. We also want to thank GIZ itself for allowing the Lab to experiment, innovate, and support teams in their innovative activities. Without such structures, we believe we wouldn’t see this kind of interdisciplinary, international collaboration between teams from around the world.
If you want to learn more about DPPD and its activities, as well as its critical evaluation, feel free to browse our blog or the DPPD blog on Medium.
