Policy Evaluation and Adaptive Governance: Learning What Works, Where and Why

Abstract

Policy evaluation is central to accountable public management, but an average treatment effect is rarely sufficient for governing complex programmes. Decision makers need to know how an intervention produced outcomes, for whom, under which institutional conditions, at what cost, over what time period, and with what unintended consequences. This editorial argues for an evaluative model that combines causal inference with implementation analysis, theory, context, equity, cost, and organizational learning. Randomized and quasi-experimental designs remain powerful tools for estimating effects, while process evaluation, realist and theory-based approaches, qualitative inquiry, administrative-data analysis, and mixed methods can explain mechanisms and transferability. Key challenges include attribution, selection bias, implementation variation, weak outcome measures, short follow-up, missing data, publication incentives, and external validity. Priorities include linked administrative data, longer-term follow-up, replication across settings, mechanism analysis, distributional effects, cost-effectiveness, adaptive experimentation, mixed-method evaluation, and transparent reporting of null and adverse findings. Evaluation should support policy learning and adaptation rather than merely classify programmes as successful or unsuccessful.