Rethinking ‘adoption’ in African agriculture

Published on 2 September 2016

Dominic Glover

Rural Futures Cluster Lead

James Sumberg

Emeritus Fellow

Is it meaningful and informative to say that a new technology has been ‘adopted’? What information is conveyed by such a statement, and how should we understand it? Does the concept of adoption help us to understand and evaluate technological change? These questions are central to efforts to measure the impact of agricultural research for development (AR4D) in African agriculture.

Every year, millions of dollars are invested in agricultural research. To justify this investment, agronomists and social scientists need to be able to show that their work produces tangible benefits for farmers and consumers. The conventional way to measure impact has been to ask, ‘has the technology been adopted?’ and ‘what have been the impacts of adoption?’

In a recent open access paper – written in collaboration with Jens Andersson of the International Maize and Wheat Improvement Centre (CIMMYT) — we argue that the adoption concept is badly flawed. Consequently it leads project evaluators to inaccurate and misleading conclusions about the return on investment achieved by AR4D. We argue that a new concept is needed, based on a better understanding of the process of technological change, but which will still provide a practical way to measure the impacts of investment in agricultural research.

The problems with ‘adoption’

The concept of ‘adoption’ reflects an idea of technology as something embodied in technical objects and artefacts such as machines, improved seeds or chemicals, which do not change as they travel from one place to another. From this perspective, farmers are portrayed as clients receiving these ‘technology packages’, which emerge fully formed from a long pipeline of scientific research. The farmers’ only decision is whether or not to accept the package and to ‘implement’ it on their farms.

The problem is that observational studies of technological change on the ground find very little that resembles this neat, transactional image of technology adoption. In reality, the practice of new methods, use of new tools and learning of new techniques is a multidimensional process that involves learning through observation, reflection, selection, experimentation and adaptation. It involves much more than simply unpacking a technology package and ‘plugging it in’ or ‘switching it on’.

Moreover, the transition from one set of farming practices to a new set, often requires negotiation among a wide network of actors including household members, labourers and neighbouring farmers, not to mention agricultural extensionists, input suppliers and consumers.

In this process of local adaptation and learning, the new technology may end up being used or practised – if at all – in ways that are quite different from what the designers envisaged. The change process may be incremental or iterative rather than clear and decisive. And the process is not linear, because old technologies may continue to be used alongside new ones, and the new technologies may be abandoned if they prove to be inappropriate, unproductive, too costly or too difficult.

When an evaluator goes into a community looking only for ‘adoption’, she is in danger of reporting both false negatives and false positives. Imagine a simple case of a new crop variety. After a significant investment has been made in plant breeding and extension, the evaluator finds little trace of the new variety in the field. Is this a case of the variety not being adopted, or is it just invisible to the evaluator because, for example, it fits an important niche for only one farmer in 20; it is known locally by one or more different names; it is being grown in mixtures with other varieties; or it has been crossed with existing varieties and in so doing enriched the local gene pool?

False positives are equally problematic. Imagine that 70 per cent of farmers in the target area have started growing the new variety. The investment looks like a resounding success, but what if, in the process, the genetic diversity available in the local seed system has been sharply reduced, increasing the community’s vulnerability to drought or disease? ‘Adoption’ overlooks this kind of effect and fails to report on it.

Moving on from ‘adoption’

We suggest that a new concept of technological change is needed, one that benefits from insights from the sociology and anthropology of science and technology. These perspectives confirm that technology is not something farmers simply receive or adopt, but something they do or make for themselves. To do so, they employ various resources including knowledge, energy, time, skill and bio-physical materials and processes, as well as networks of social and economic relationships.

The case of a new crop variety being planted by farmers for the first time comes closest to the conventional understanding of adoption, but even then, the embedding of a new seed in a wider farming system will entail user experimentation and adaptation.

None of this means that scientific research or technology development are irrelevant, but we should recognise that science is just one of the resources farmers and communities may use to improve their farming practices and outcomes.

We suggest that the new concept should be capable of appreciating change that is emergent, incremental and iterative, as well as changes that are partial and selective.

It should also be able to encompass technologies of different complexity and scale, ranging from a new variety of an established crop through to a more complex transition to a new cultivation system that involves interconnected changes in crop or land management methods – as we see with the example of conservation agriculture – or even a reorganization of community institutions, as observed with the system of rice intensification (SRI).

Developing and testing such a tool is a research agenda that we hope to pursue at IDS in collaboration with other partners in the near future. We feel that the new concept should be practical and useful, in other words it should provide a basis for robust estimates of the effects of investment in AR4D, while remaining easy to use and cost-effective.


About this opinion

Related content