International development cooperation in recent years has seen an increased interest in evidence and evidence-based policies and practices. The central idea of the evidence movement is for policies and practices to be based on the best available scientific research about what works, what does not, and the reasons why. However, the evidence notion is by no means unambiguous and what constitutes evidence is highly debatable. In this article, evaluation is defined as evidence-based provided it makes explicit use of a counterfactual. This can be done through experimental or quasi-experimental research designs.
Evidence-based evaluations have appeared later in international development cooperation than in areas such as medicine, social work, and education. The lack of impact evaluations has been increasingly recognised. Many researchers and evaluators, and indeed administrators, managers and politicians, call for and look for evidence of results of development spending. But a brief review of evaluation practice suggests that such evidence is not forthcoming, despite the significant amounts of money spent on evaluation. This article sets out to explore why, and discuss how to improve the situation. It has three interrelated purposes.
1 Descriptive: given a strict definition of what constitutes a sound and reliable design for evidence-based evaluation, how often do evaluations live up to such standards?
2 Explanatory: why do so few evaluations use experimental or quasi-experimental enquiry methods? We focus on how evaluation processes are determined and what role terms of reference (ToRs) play in deciding research design.
3 Assuming more evidence-based evaluations would be desirable, our third purpose is to produce hypotheses on how to generate more of these, to discuss what changes this necessitates among those who commission and use evaluations, and among those who conduct them and/or are subjected to them.
This article comes from the IDS Bulletin 39.1 (2008) Evidence‐based Evaluation of Development Cooperation: Possible? Feasible? Desirable?