Development support must be able to prove its effectiveness. Impact evaluation is a way to generate this information. The thesis is about the design of these impact evaluations and how research methods can be combined to obtain credible evidence on effectiveness. It contrasts two approaches to impact evaluation design, ‘randomistas’ (Does it work?) and ‘realistas’ (For whom does it work, and under what conditions?), and distils seven principles for research design that create synergy between these two approaches. The thesis covers various development interventions. The main research concerns a Bolivian grant fund that supports investments in processing by farmer groups. To assess the effectiveness of this fund it was necessary to develop and test a new tool to measure organisational strength of these groups, called Tension Containment Capacity and apply a new method of data analysis, Qualitative Comparative Analysis. Interestingly, grants to the older, larger and stronger organisations proved particularly unsuccessful.