Scientific research has historically been assessed by the level of citations a publication or researcher has – the more the better being the mantra. The reasoning being that the “credible” researcher (or significant work) would automatically lead to citations or popularity: the more “credible” leading to more citations. Problems with this model are quite obvious, as it leads to a “publish or perish” mentality and encourages “popular” or trendy research. In addition, frequently cited publications are, at times, cited for their controversial nature, and not necessarily for their significance or impact in terms of research. But what does that mean for agricultural research?
The final product of agricultural research should, at the end of the day, have a measurable positive impact on the lives of the poor. If that is taken as a given, then we must reconsider our current evaluation models for agricultural scientific research. Various other strategies have been considered to address some of the shortcomings of the above model. However, most of these strategies aim to include the end-users either in the developing of the project or in training at the tail end of the project.
Is this enough? Is there not a better way to measure impact? How can we better link outputs to results? What about accountability?
During a workshop held last year at the GFAR meetings in India, Frank Rijsberman, Sanjini De Silva, and I presented to the participants a model based on Outcome Contracting (you can access the article here). The basis of this model is accountability, both in terms of project design and funding. If the primary goal of our work and research is poverty reduction, should we not be held accountable for it? In the new model, researchers, along with the end users, partners etc, identify the impact pathway of any particular project, and decide up to which point the project can be held responsible. Accountability is established and funding, or partial funding, is awarded upon achieving the intended goal.
Can such an inclusive model work in our current environment? How would that affect our current approach to research? What would be the drawbacks for using such a model?