The Experiment Societyexperimentsociety.orgGet involved

The Field

Evidence-based policy is a movement. ES is one part of it.

A growing ecosystem of researchers, consultants, funders, and capacity-builders is working to make public decisions more evidence-informed. The Experiment Society occupies a specific niche within that ecosystem — not duplicating what others do well, but filling a gap that the existing organizations are not designed to fill.

How ES fits

The gap between research and institutional practice.

Existing organizations do excellent work producing evidence and advocating for its use. What most do not do is build the capacity of public institutions to generate their own evidence, continuously, without external researchers or consultants.

That gap — between knowing what the evidence says and being the kind of institution that learns from evidence — is where ES works.

QuestionMost of the fieldThe Experiment Society
Who runs the experiments?External researchers or consultantsThe public institution itself
Primary geographyGlobal development (J-PAL/IPA) or federal/state (RfA)Municipal and local government, any country
ModelResearch production or policy advocacyCapacity building — institutions learn to experiment
OutputPublished findings or certified standardsLearning institutions + public registry
RelationshipProject-based or arm's lengthOngoing institutional partnership
Theory of changeDecision-makers with good evidence will make better choicesInstitutions that learn will govern better over time

Organizations

Eight organizations worth knowing.

These organizations collectively define the evidence-based policy movement. Each does important work. Understanding how they differ helps clarify what ES is — and is not — trying to do.

Research network

J-PAL

Abdul Latif Jameel Poverty Action Lab

MIT, Cambridge MA

Founded 2003

Academic research network

→Methodological rigor

→Global scale

→Policy translation work

→Researcher training network

J-PAL is the world's leading network for randomized evaluations of social programs, primarily in low- and middle-income countries. Its affiliated researchers have run more than 1,500 RCTs covering 400 million people. J-PAL synthesizes findings, trains researchers, and advocates for policies with strong evidence bases.

How ES relates

J-PAL produces research about what works. ES builds the capacity for public institutions to find out themselves. J-PAL's researchers evaluate programs; ES's model has practitioners running experiments. The two are complementary, not competing — J-PAL's findings populate ES's registry; ES's pilots generate findings that belong in J-PAL's databases.

Research network

IPA

Innovations for Poverty Action

New Haven, CT

Founded 2002

Research and implementation

→Implementation partnerships

→Randomized evaluation at scale

→Financial inclusion research

→Policy uptake focus

IPA conducts and scales rigorous research to address global poverty, with a strong focus on financial inclusion, health, and agriculture in sub-Saharan Africa and South Asia. IPA works closely with implementing partners and governments to ensure research findings translate into policy.

How ES relates

IPA is strongest in low-income country contexts and international development. ES focuses on municipal and local government in higher-income settings. IPA evaluates programs run by implementing partners; ES trains institutions to evaluate their own programs.

Government consultancy

Behavioural Insights Team

The Behavioural Insights Team (BIT)

London, UK (with offices worldwide)

Founded 2010

Government consultancy / social purpose company

→Scale of experimentation

→Government access

→Practical behavioral science

→EAST framework

Originally the UK Cabinet Office 'Nudge Unit,' BIT applies behavioral science to improve government services and public outcomes. BIT runs thousands of randomized trials annually across health, employment, education, and tax compliance. It operates as a social purpose company with government and private sector clients.

How ES relates

BIT is a consultancy — it brings behavioral science expertise to client governments. ES is an infrastructure builder — it trains institutions to apply behavioral science themselves, without needing external experts for every experiment. BIT's model produces better individual interventions; ES's model produces learning institutions.

Behavioral design

ideas42

ideas42

New York, NY

Founded 2008

Behavioral design nonprofit

→Behavioral design methodology

→Equity focus

→US domestic programs

→Practitioner-facing tools

ideas42 applies behavioral science to social challenges — financial health, criminal justice, education, and public health. It designs and tests interventions with government and nonprofit partners, with a strong emphasis on equity and ensuring that behavioral tools benefit people who need them most.

How ES relates

ideas42 designs behavioral interventions for partner organizations. ES builds the systems for organizations to design, run, and learn from experiments themselves. ideas42 is project-based; ES is capacity-based. The two organizations share a commitment to behavioral science as a tool for equity.

Evidence synthesis

What Works Network

What Works Network (UK)

United Kingdom

Founded 2013

Evidence synthesis and standards

→Sector-specific synthesis

→Government backing

→Teaching Schools / Evidence Champions model

→Tiered evidence standards

The UK's What Works Network comprises ten sector-specific 'What Works Centres' covering areas from education (Education Endowment Foundation) to crime reduction, local economic growth, wellbeing, and early intervention. Each centre synthesizes existing evidence and supports its use in practice.

How ES relates

The What Works Network synthesizes and grades existing evidence. ES generates new evidence through local pilots. The two functions are complementary: ES's experiments can enter the What Works pipeline; What Works findings can populate ES's registry. The UK model has also demonstrated that sector-specific evidence infrastructure produces higher adoption than generic evidence advocacy.

Policy advocacy

Results for America

Results for America

Washington, D.C.

Founded 2012

Policy advocacy and government practice

→Federal and state policy influence

→Budget advocacy for evidence

→Standard-setting

→Political access

Results for America works to ensure US government decision-makers have access to and use the best available evidence. It produces the Invest in What Works Federal Standard of Excellence and the State Standard of Excellence, tracking which governments are building evidence capacity into their policy processes.

How ES relates

Results for America advocates for evidence-based policy at the federal and state levels. ES focuses on municipal and local government, below the radar of most federal policy work. RfA works top-down through policy mandates; ES works bottom-up through institutional capacity. RfA creates the standards; ES helps institutions meet them.

Municipal capacity

Bloomberg What Works Cities

Bloomberg Philanthropies — What Works Cities

New York, NY

Founded 2015

Municipal capacity building

→Direct municipal engagement

→Certification framework

→Technical assistance network

→Peer learning communities

What Works Cities helps US cities use data and evidence to improve services and make better decisions. It offers certification (the What Works Cities Certification) and connects cities to a network of technical assistance partners. As of 2023, more than 130 cities have participated.

How ES relates

What Works Cities focuses on data infrastructure and performance management — helping cities measure what they're already doing. ES focuses on experimentation — helping cities test new approaches. The two complement each other: data infrastructure is a precondition for experimentation, and experimentation is what data infrastructure should ultimately enable.

Evidence synthesis

Campbell Collaboration

Campbell Collaboration

Oslo, Norway

Founded 2000

Systematic review network

→Systematic review methodology

→International scope

→Open-access evidence

→Named after ES's patron saint

The Campbell Collaboration produces systematic reviews of the effectiveness of social interventions — synthesizing evidence on crime and justice, education, international development, and social welfare. Named after Donald Campbell, whose work on the experimenting society directly inspired The Experiment Society.

How ES relates

The Campbell Collaboration synthesizes and rates existing evidence. ES generates new evidence and builds institutional capacity to generate more. Campbell provides the accumulated evidence base that practitioners should consult before designing a new experiment; ES produces the experiments that eventually enter Campbell's databases.

What we share

A common set of commitments — despite different models.

The organizations in this ecosystem disagree about methods, scale, and where to focus. They share a more fundamental agreement that public decisions should be informed by evidence and that institutions should get better at learning.

The Experiment Society aims to work alongside these organizations, not in competition with them. Our registry draws on research produced by J-PAL, IPA, and Campbell. Our methodology is informed by BIT and ideas42. Our municipal partnerships align with What Works Cities. Our standards are consistent with Results for America's frameworks.

Randomization is the gold standard, not the only standard

RCTs are the best way to establish causal relationships, but quasi-experimental designs — regression discontinuity, difference-in-differences, interrupted time series — can produce credible evidence when randomization isn't feasible. The field broadly agrees that rigor is non-negotiable; the exact method is context-dependent.

Null results are as important as positive ones

Every major organization in this ecosystem publishes null and negative results. This norm — still contested in academic publishing generally — is now standard practice in evidence-based policy. What works and what doesn't are equally important inputs to policy decisions.

Evidence doesn't implement itself

Producing evidence is necessary but not sufficient. The field has learned — through painful experience — that rigorous findings that are not connected to implementation often have no policy impact. The uptake and use of evidence is an active design problem, not an automatic consequence of publication.

Equity is a design requirement

Average effects can obscure distributional harms. The best organizations in this ecosystem pre-specify subgroup analyses, publish heterogeneous effects, and treat equity as a central evaluation criterion rather than an afterthought.

Get involved

The field is growing. So is the gap it has not yet filled.

Most of the organizations listed here work at national or international scale, with governments that already have significant research capacity. The municipal level — libraries, parks departments, permit offices, school districts — is largely outside their scope. That is where ES works.

Get involved →Browse the registryRead the Dispatch