Scientists Map Why Evidence Fails to Reach Government Policymakers

Governments around the world have spent the past two decades pouring money and political capital into evidence-based policymaking. Evaluation offices have been created, clearinghouses launched, and entire behavioural insights teams embedded inside ministries. Yet a stubborn and well-documented problem persists: rigorous research that could improve public programmes frequently never influences the decisions it was meant to inform. A new Perspective published in Nature Human Behaviour argues that the field has been diagnosing this failure incorrectly, and that fixing it requires treating the adoption of evidence not as a single act of persuasion but as a chain of distinct behavioural and organizational hurdles that research has largely failed to map.

The article, authored by a team spanning Harvard Kennedy School, New America, Minnesota Management and Budget and ESADE Business School, introduces a unified framework built on three sequential questions. Is the evidence useful, meaning does it speak to a problem policymakers actually face? Is it usable, meaning can decision-makers realistically interpret and act on it within their institutional constraints? And is it used, meaning does it ultimately change what governments do? The authors contend that a substantial production–adoption gap has opened because researchers and funders have concentrated overwhelmingly on the first stage, producing more and better studies, while paying far less systematic attention to the second and third stages where most evidence quietly dies.

The scale of the problem is striking when laid out in the literature the authors synthesize. Implementation science has estimated that it takes an average of seventeen years for research evidence to change clinical practice, and there is little reason to believe policy moves faster. Systematic reviews of barriers to evidence use among policymakers consistently identify the same culprits: poor timeliness, irrelevant framing, lack of trust, and absence of intermediaries who can translate findings into actionable recommendations. Meanwhile, surveys of civil servants in the United States and Australia reveal that many public employees have never encountered the evidence base relevant to their portfolios, and that those who have often lack the statistical training to interpret it.

One of the most provocative threads in the new framework concerns the psychology of the people on the receiving end. Policymakers, like everyone else, are boundedly rational and susceptible to motivated reasoning. Experimental work shows that politicians often update their beliefs selectively, discounting evidence that threatens their ideological commitments, and that they appear more resistant to debiasing interventions than the general public. Field experiments with local officials across multiple countries have tested whether presenting impact estimates changes behaviour, with mixed results. The authors argue that this growing body of behavioural evidence on evidence reception remains fragmented across political science, economics and psychology, and that integrating it is essential to understanding why even high-quality, well-communicated findings go unused.

The framework also draws attention to the supply side of the problem: the characteristics of the evidence itself. Credibility matters, a finding that stretches back decades in communication research, but so do practical attributes such as cost, feasibility and the political palatability of the intervention being tested. Conjoint experiments with US state government officials suggest that decision-makers weigh implementation difficulty and price alongside effect sizes, sometimes favouring a less effective but easier-to-run programme over a more effective but operationally demanding one. Studies of policy diffusion show that effectiveness information travels unevenly across jurisdictions, and that ideological alignment between the source of evidence and the adopting government can matter as much as the underlying results, a pattern documented in recent work on evidence-based policy adoption.

Intermediary organizations occupy a central place in the agenda. The United Kingdom’s What Works Centres, evidence clearinghouses in the United States, and a growing ecosystem of knowledge brokers all exist to bridge the production–adoption gap, yet the authors note that scholarship on these intermediaries is surprisingly thin. Are they neutral boundary organizations or politically savvy policy entrepreneurs? Which brokering strategies, skills and organizational forms actually increase uptake? Large-scale studies using tools such as Altmetric and Overton to trace the societal impact of policy research are beginning to provide answers, and the authors highlight emerging work on AI-augmented clearinghouses and large language models for evidence synthesis as potentially transformative, if unproven, technologies for making evidence cheaper to find and digest.

Scaling presents another underexplored bottleneck. A randomized evaluation conducted in one context may not transport to another, and formal work on the transportability of experimental results remains largely confined to technical journals. Evidence from two prominent government nudge units suggests that effect sizes observed in academic settings often shrink when interventions are deployed at administrative scale, and megastudies involving hundreds of thousands of participants have revealed substantial heterogeneity in what works. The authors argue that the field lacks a systematic understanding of when and why findings generalize, and that policymakers’ justified scepticism about external validity is rarely treated as a researchable question in its own right rather than an obstacle to be overcome with better messaging.

Organizational learning offers the final pillar of the framework. Governments are not individual decision-makers but complex bureaucracies with routines, incentives, capacities and cultures that shape whether information is absorbed. Research on organizational learning, drawn heavily from the private sector, suggests that learning requires deliberate structures: feedback loops, protected time for reflection, and leadership that rewards experimentation. Public sector studies indicate that involvement in performance management routines can increase the use of performance information, but that public managers may be more risk-averse than private counterparts, and that status quo bias systematically dampens appetite for reform. The authors call for research that treats evidence adoption as an organizational capability to be built, not merely a message to be delivered.

What emerges from the synthesis is a concrete research agenda rather than a lament. The authors identify gaps at every stage: we know little about how policymakers search for evidence in the first place, how they trade off competing studies, how intermediary organizations should be designed, how to predict transportability, and how organizational contexts amplify or suppress uptake. They argue that closing these gaps is not an academic luxury. Billions of dollars in social spending ride on decisions made with thin or stale evidence, and the promise of the evidence-based policymaking movement, improving societal outcomes through better decisions, cannot be kept if the last mile between journal and legislation remains unmapped. The Perspective, which grew out of collaboration between academics and practitioners inside government, models the kind of co-production it recommends: research questions shaped by the people who must actually use the answers.

Subject of Research: A research agenda for closing the gap between the production of policy-relevant research and its adoption in government policymaking.

Article Title: A research agenda for making evidence useful, usable and used in policymaking

Article References: A research agenda for making evidence useful, usable and used in policymaking. (n.d.). https://doi.org/10.1038/s41562-026-02593-3

Image Credits: AI Generated

DOI: 10.1038/s41562-026-02593-3

Keywords: evidence-based policymaking, Nature Human Behaviour, production-adoption gap, behavioural public policy, knowledge brokers, organizational learning, policy diffusion, implementation science, What Works Centres, randomized evaluations, policymaker decision-making, research translation