Previous
This short series titled Evidence as Politics in Policy explores a simple but often overlooked point: evidence does matter in policymaking, but never on neutral ground. In all political systems, evidence is used strategically, selectively ignored, unevenly protected, and shaped by democratic and non-democratic pressures, institutional design, and political incentives. Across these articles, I argue that the challenge is not simply to produce more evidence or communicate it better, but to understand the political, economic and social conditions under which evidence is used, contested, or sidelined — and what that means for anyone interested in evidence-informed policy.
The reflections in these articles have been drawn from a series of interviews with political scientists in Latin America, Africa and Asia as part of our ongoing effort to better understand the political economy of evidence use.
Read until the end for summaries of the interviews.
So far in this series, I have argued that evidence is political, that it is often ignored for reasons that have little to do with its availability or quality, and that only some institutions retain the room to use it consistently.
So what should we do about it?
For sure, we should not avoid going down the usual paths: produce better evidence, communicate it more clearly, strengthen capacity, train policymakers, improve access to data.
None of that is wrong, but it is incomplete.
In the previous posts in this series, I have argued that the problem is not simply a lack of evidence or a deficit of technical skills. The deeper problem is that evidence has to operate in complex and imperfect political systems shaped by competing interests, unequal power relations, electoral incentives, institutional weakness, professional hierarchies, and very different ideas of what counts as credible and legitimate knowledge.
That means the challenge is not to build a more technocratic version of policymaking, but to create better conditions for evidence to be used within politics as it is actually practised.
This is easier said than done.
The answer is not more technocracy
It should be clear by now that the answer is not simply to double down on evidence. We must give up the hope that if only governments listened more carefully to the data, trusted specialists more consistently, and protected policymaking from public pressure, then better decisions would follow.
That view is attractive because it is tidy and easy to turn into fundable projects. It imagines evidence and evidence use as clean, rational and external to politics. It imagines politics as the messy – and separate – space that needs to be disciplined by expertise.
But this is not how policymaking works, and it is not necessarily how it should work either.
Our conversations with political scientists for this series point in another direction. Certainly, in democratic systems — however imperfect — public opinion, organised constituencies, teachers, unions, parents, community actors and social movements also generate forms of knowledge that cannot simply be dismissed as noise.
They shape the political legitimacy of decisions and influence whether reforms are accepted, resisted, implemented or reversed. They often matter more, politically, than formal research findings.
This creates a real tension for evidence-informed policy advocates and practitioners.
On one side is the promise of technical evidence: studies, datasets, evaluations, modelling, statistical analysis and institutional expertise.
On the other side is what one might call the evidence of political life: social demands, lived experience, collective memory, public expectations, and the knowledge held by those who have to live with a policy once it is introduced.
The mistake is to imagine that one of these should replace the other – or that they can be treated as equal and compared like-for-like.
Just as I have argued that a purely anti-intellectual model would lead to huge economic, social and political costs, a purely technocratic model is not only politically unrealistic; it may also be normatively undesirable. If policymaking becomes little more than rule by experts, insulated from public participation and dismissive of social claims, it risks hardening into a narrow and elitist view of public policy.
And that, too, can produce bad decisions. In their book on policymaking in the United Kingdom, Anthony King and Ivor Crewe explain policy blunders through accurate accounts of what happens when decisions are made in small, homogeneous groups with little connection to the real world – regardless of their access to data, evidence, expertise, or resources.
Public knowledge matters
One of the strongest lessons emerging from the interviews is that what counts as evidence is itself politically contested.
For some actors, evidence means statistics, impact evaluations, administrative data and economic modelling. For others, it means what teachers see in classrooms, what parents experience when services fail, what unions know about working conditions, or what local officials understand about implementation constraints.
These are not interchangeable forms of knowledge. I think they do different things.
Large-scale data may reveal patterns that no individual can see. It can show whether learning outcomes are improving, if resources are distributed efficiently, whether a reform is reaching the intended population, or whether a system is reproducing or reducing inequality.
But local and experiential knowledge can reveal why those patterns persist, how policies are interpreted, where administrative rules break down, why frontline workers resist reforms, or why citizens do not trust a programme even when the data says it should work.
This is a crucial distinction for policymakers.
A minister may have national data showing that a reform is technically justified. But if teachers believe it is unworkable, parents see it as unfair, unions interpret it as an attack, or communities experience it as externally imposed, then the policy may fail.
No RCT will make them change their mind. In fact, RCTs and impact evaluations rarely change policymakers’ minds!
From evidence pyramids to evidence webs
This suggests that we may need to rethink how we imagine evidence itself.
Too many debates and projects still rely, implicitly or explicitly, on a rigid hierarchy of evidence: scientific evidence at the top, experiential knowledge at the bottom; quantitative data treated as serious, while local experience, teacher testimony, union perspectives or citizen feedback are treated as secondary or anecdotal.
Rather than a narrow evidence pyramid, we may need something closer to a web: multiple types of evidence connected to different questions, audiences and decisions.
A minister may need national data to understand system-level trends, but also local stories to understand why those trends persist. A reform may look strong in theory but encounter resistance because teachers, school leaders, or frontline administrators can already see implementation problems that central data do not capture. A policy may satisfy technical criteria yet fail politically because it has been designed without regard to how citizens understand fairness, risk or public responsibility.
If evidence is defined too narrowly, many of these dynamics will remain invisible until it is too late.
Democratising evidence use does not mean lowering standards
In many contexts, evidence is still treated as the domain of economists, statisticians, evaluation specialists, policy analysts and international consultants. These actors matter, but if they become the only legitimate users and interpreters of evidence, the system becomes fragile.
Teachers, district officials, local administrators, health workers and other street-level bureaucrats are often expected to act on evidence without ever being given the tools, time or institutional support to do so. Citizens are invited to participate after the main decisions have already been made. Professional associations and unions are treated as obstacles rather than as possible sources of knowledge.
I think this is a weak foundation for evidence-informed policy.
A more legitimate (in a democracy or otherwise) approach would aim to broaden the community of evidence users and producers whose expertise is grounded in implementation rather than formal research methods.