Logo Tropentag

Tropentag, September 20 - 22, 2017 in Bonn

"Future Agriculture: Social-ecological transitions and bio-cultural shifts"

Learning from the Most Successful: Prioritizing Rural Development Interventions by “Positive Deviance” Analysis

Jonathan Steinke1,2, Jacob Van Etten2, Mark van Wijk3

1Humboldt-Universität zu Berlin, Germany
2Bioversity International, Information services and seed supplies, Costa Rica
3International Livestock Research Institute (ILRI), Livestock Systems and the Environment, Kenya


Rural households in developing countries pursue multiple livelihood goals simultaneously, such as food security, income, and a healthy nutrition. These goals are often dynamically interconnected, meaning that development interventions targeted at improving one livelihood domain may cause concurrent losses in another. Researchers and policy-makers have answered calls for holistic, integrated development by formulating new paradigms, such as 'sustainable intensification', yet without prescribing concrete strategies. Hence, which innovations and practices lead to the desired changes in specific context remains to be determined.

System modelling is a widespread strategy to integrate different livelihood domains, and make ex-ante predictions about the outcomes of specific technological or institutional interventions. But models are subject to uncertainty and incompleteness, and frequently suffer from narrowly defined system boundaries. These limitations make it difficult to derive concrete recommendations for rural development interventions.

To support the selection of intervention strategies for holistic development, we propose an alternative approach. The “positive deviance” concept rests on the observation that, in many rural communities, some households achieve higher livelihood success than others, although facing similar resources, challenges, and trade-offs. These positive deviant households (PDs) likely do things differently, be it farm management, resource allocation, or off-farm activities. We suggest that involving PDs in empirical, qualitative research may reveal successful and potentially uncommon behaviours embedded in local context. These may be crucial inputs to meaningful development interventions.

We designed a methodology for identifying PDs, using a lean data approach and based on household success in five key livelihood domains. We surveyed 521 rural households in Tanzania, identifying 54 PDs. Here, positive deviance chiefly stemmed from success in nutrition, income, and low environmental impact. Interventions in these domains are thus more promising than interventions for food security or gender equity. We also systematically selected 18 PDs for in-depth interviews and farm visits, revealing several on- and off-farm success strategies, such as investments into crop storage, and meticulous allocation of labour resources during periods of land preparation and sowing.

This pilot study demonstrates the potential of positive deviance analysis for rapid prioritisation of rural intervention options, grounded in the real-life experiences of the target population.

Keywords: Development interventions, lean data, positive deviance, research methods, sustainable intensification

Contact Address: Jonathan Steinke, Humboldt-Universität zu Berlin, Unter den Linden 6, 10999, Berlin, Germany, e-mail: j.steinke@cgiar.org

Valid HTML 3.2!