in: Ole Winckler Andersen / Beate Bull / Megan Kennedy-Chouane (eds.), Evaluation methodologies for aid in conflict, London: Routledge, 154-174
Impact evaluations are supposed to provide reliable knowledge about how statebuilding support is best designed and delivered in order to be effective. Yet an analysis of major statebuilding evaluations raises concerns about their methodological rigour and the validity and generalisability of their findings. Few studies follow rigorous social science standards of inferring causality. Most fail to explicate a theory of change, to construct a credible counterfactual, and to make use of quantitative methods where possible. As a consequence, they have a “conservative” bias: instead of testing theories of change, they tend to reproduce the conventional wisdom underlying external interventions in the first place.