What Works, Why, and Where: Integrating Realist Evaluation, Behavioural Science, and Systems Thinking

Abstract

Evaluation commissioners need more than a verdict on whether a program worked. They need evidence to inform cost-effectiveness decisions, support scaling, and guide complementary program design. Realist evaluation provides the right causal structure to generate that evidence: in this context, the program activates this mechanism, producing this outcome. This article addresses three problems with realist evaluation: the misconceived debate between realist and counterfactual approaches, which are independent decisions not competing alternatives; poor specification of mechanisms and context; and the failure of evaluations in complex systems to capture how the context evolves over time and at scale. The solution integrates realist evaluation with behavioural science and systems thinking and allows for any data collection and analysis methods. Behavioural science operationalises mechanisms and context using a model of behaviour called “COM-B”, and explains why the same program produces different effects for different people through mindsets – making findings precise enough to inform cost-effectiveness decisions and complementary program design. Systems thinking maps feedback loops, leverage points, and emergent outcomes – identifying where to intervene for lasting or transformative change and what contextual conditions must be in place before scaling. The realist behavioural systems evaluation framework is illustrated through a case study and five-step practical guide.