Assessing Innovations in International Research and Development Practice

Enhancing impacts of international development interventions has become a central issue of the twenty-first century. Conventional monitoring and evaluation (M&E) tools either focus on efficiency (output-to-input relationships) or strive to demonstrate a logical progression from specific actors and factors of an intervention to development impacts (inputs => activities => outputs => outcomes => impacts). However, in complex adaptive systems there is neither such a linear results chain nor can impacts be unambiguously attributed to an actor or a factor. Therefore, alternative ways of doing M&E focus on outcomes - the changes in behaviour and social relations - rather than on impacts, such as poverty reduction, environmental protection and social inclusion. Innovation systems thinking, particularly in renewable natural resource, agriculture and rural development, informs that the dominant paradigm of impact assessment should be complemented by social innovation assessment, providing research and development actors with critical learning lessons. This paper integrates two distant bodies of literature - the literature on impact assessment of research and development interventions, and the literature on social psychology of assessing learning and innovations. Based on case studies of a series of projects implemented in India and Nepal under DFID's 11-year Renewable Natural Resources Research Strategy (RNRRS) programme between 1995 and 2006, a social innovation assessment tool was developed and implemented. The tool includes questions about critical incidents and modes of stakeholder interactions to be ranked on a four-point scale depending on how often the statements apply to the respondents' work environments. The social innovation assessment provides critical learning lessons for social innovation generation and overall performance improvement in collaborative research and development interventions at the organisational, network and system levels.