Decision Science in Service to All: Making Behavioural Science Work in Government
A GAABS Conversation with Baruch Fischhoff, PhD and Alycia Chin, PhD
Behavioural science doesn't simply drop into government ready for deployment. It must be translated, embedded, and ultimately measured by whether it changes how agencies serve the public—not by academic outputs.
This is the perspective shared during a GAABS conversation featuring Baruch Fischhoff, PhD, Howard Heinz University Professor at Carnegie Mellon, and Alycia Chin, PhD, Senior Financial Economist at the U.S. Securities and Exchange Commission. Together, they co-edited the PNAS (Proceedings of the National Academy of Sciences) special feature Decision Science in Service to All, which brought together practical applications across five federal agencies. Their conversation offered both a roadmap and a reality check for practitioners seeking to embed behavioural science in organisations where it isn't the primary mission.
TL;DR
In public service activities, translation matters more than theory.
Behavioural science works in government when researchers learn to speak the language of the domain.
Build standing institutional capacity, not episodic projects.
Webinar Video
Available here for 30 days.
The Core Insight
Behavioural science doesn't work in government because the research is weak—it fails because researchers and policymakers speak different languages.
The Key Problem
Behavioural science is typically a supporting function, not the organisation's main mission. This creates a translation gap: "recommendation" means something different at the SEC than in academic research. Opening each domain's "black box" is essential.
Getting Behavioural Science Into Government
The PNAS special feature spans five federal agencies—the Federal Trade Commission, Securities and Exchange Commission, Office of Evaluation Sciences, Congressional Budget Office, and Food and Drug Administration. The breadth is intentional. As Alycia Chin explained, whether readers work in healthcare, banking, insurance, or another field, "they should find either relevant subject matter or methods they can adapt to applied settings."
But embedding behavioural science requires foundational conditions that go beyond hiring a specialist. Chin identified what's needed: clear legal authority and understanding of the agency's mandate, sufficient staff capacity, access to outside expertise, interdisciplinary collaboration among behavioural scientists, domain experts, and specialists, and—critically—leadership, trust, and institutional support.
The core challenge is structural: behavioural science is typically a supporting function, not the organisation's primary purpose. This creates what Chin calls the translation problem. "Effective collaboration requires learning those perspectives and translating between them. Sometimes translation turns on a single word."
At the SEC, for example, the word "recommendation" carries specific legal meaning. A research design must either meet that legal standard or deliberately study everyday advice instead.
At the FDA, regulatory processes structure how change happens.
At the Congressional Budget Office, economic models dominate.
Opening each domain's "black box" is essential to making behavioural science relevant to policymakers operating under different conceptual frameworks.
“Effective collaboration [with government agencies] requires learning those perspectives and translating between them. Sometimes translation turns on a single word.”
Building Institutional Capacity
Rather than running episodic projects, Chin's team at the SEC built standing capacity: a multidisciplinary innovation lab serving as an internal consulting hub, and Thrive, a nationally representative survey panel of 7,500 respondents who participate at least quarterly.
This infrastructure does more than deliver research; it normalises research conversations within the agency. "A standing capacity lowers the barrier to starting new projects and builds interest over time," Chin noted. What might otherwise feel like a special request becomes part of how the organisation works. "A standing capacity lowers the barrier to starting new projects and builds interest over time," - Alycia Chin, PhD
The longitudinal nature of Thrive proves particularly valuable. Cross-sectional studies answer "what is true now?" Longitudinal data answers "what changed, and why?"—a question closer to how policymakers actually think about effectiveness. But Chin also emphasised the importance of qualitative research, often underappreciated in policy contexts. In one overdraft-explanation project, her team spent considerable time refining legally accurate wording only to find that users were not reading the explanatory text at all.
“A standing capacity lowers the barrier to starting new projects and builds interest over time. ”
During qualitative testing, "one participant made a decision before finishing the sentence. Seeing that behaviour directly was more persuasive to stakeholders than simply being told that people might not read closely."
This distinction matters profoundly: showing stakeholders real behaviour is more convincing than reporting abstract findings.
From Concept to Practice: The FDA Journey
Baruch Fischhoff offered a deeper examination of how behavioural science becomes embedded institutional practice. His case study: the FDA's benefit–risk framework, which reveals what institutional change actually looks like. In government efforts, the timeline is instructive—and sobering when you understand the timescales involved.
The FDA employed psychologists who recognised communication problems, but institutional change proved difficult. In 2005, the Institute of Medicine criticised the agency's communication practices. The FDA responded with a strategic plan. Congress created a statutory Risk Communication Advisory Committee (which Fischhoff chaired), meeting quarterly and holding "show me the science" sessions that summarised relevant behavioural-science evidence in concise, accessible form.
Pressure from industry and patient advocates for greater transparency revealed limitations in lengthy, difficult-to-navigate FDA reports. Over years of consultation, pilots with 12 products, and senior oversight, the FDA developed a benefit–risk framework: a comprehensible table distinguishing evidence from uncertainty, separating factual judgements from value judgements, and making clear that the FDA weighs benefits as well as risks.
A crucial innovation emerged: quarterly "Voice of the Patient" sessions on difficult topics—sickle-cell disease, ME/CFS. These provided a structured, behaviourally informed way for patients, clinicians, regulators, and technical experts to work together.
"The approach proved successful and was subsequently replicated many times. It expanded from pharmaceuticals to biologics, including work related to what later became the mpox vaccine." - Baruch Fischhoff, PhD
The FDA eventually adopted the benefit–risk framework as standard practice. This represents what Fischhoff called "a substantial and durable contribution from the social, behavioural and decision sciences." But it only happened because people stayed committed through years of incremental progress, political changes, and institutional resistance.
Six Lessons for Academic–Agency Collaboration
When asked what academic researchers benefit from knowing or having before embarking on government work, Fischhoff outlined six principles that are worth examining because they reveal what persistence and institutional commitment look like:
Build personal connections.
Relationships with people inside the agency matter more than credentials or published papers.
Learn the domain.
Understand how laws, regulations, and technical processes actually work—not the simplified version.
Stay committed.
Agency partners need to know researchers will remain engaged when projects become difficult. Some of Fischhoff's FDA work took years to show results.
Be theory-informed, not theory-driven.
Use relevant theory, but don't treat organisations merely as laboratories for academic hypotheses.
Publish practical lessons early.
Share actionable messages before the slow peer-review process completes. Conduct follow-up analyses and studies later.
Support publication.
Agencies should resource rigorous follow-up research; published evidence strengthens both the agency and the field.
These are apt suggestions for work that actually lasts.
What Works: Patience and Opportunism
When asked—"What do you do when you work on something for a long time and it doesn't move forward?"—Fischhoff acknowledged both the frustration and the reality of applied work in governmental settings. Academic work provides teaching and advising responsibilities, which help with patience, but stalled work in the world of government agencies can be genuinely frustrating.
Opportunism matters too. He shared an example: during the Clinton administration, his team developed a method for setting risk priorities within and across agencies. By the time it was ready, the administration changed and interest disappeared. "Later, the Blair government in the UK had a risk initiative, and HM Treasury adopted and adapted the work. Embedding it in formal requirements made it last longer than many staff-led initiatives."
The lesson: good work doesn't disappear. It may find its moment elsewhere, in different contexts, sometimes years later.
For practitioners trying to win attention and sustain research within organisations where science isn't the core mission, Chin offered a practical model that contrasts sharply with the frustration of stalled projects: the "repeat customer" approach.
"When an internal team shows that it is collaborative, understands the client's needs and respects existing practice, partners are more likely to return for a second and third project. That is how science can grow organically within an organisation."
What's Still Missing?
Despite these successes, both speakers identified structural gaps that remain. Chin pointed to training and career pathways. Graduate and master's programmes prepare researchers for academia, not for organisational realities. Future policy researchers don't typically learn legal perspectives before entering practice. They don't learn how regulatory processes actually work. They don't learn to translate between academic and practical languages.
The result: applied researchers repeatedly rediscover lessons others already learned.
Fischhoff offered two broad aspirations:
"Make the case to academia that applied government work is valuable and should be a high aspiration for talented students."
"I would also like to bridge the gap with engineering fields that still doubt the value of social science, even though behavioural science can be useful to them."
Moving Forward
What emerges from this conversation is not a blueprint—each agency's conditions differ—but a set of principles for making behavioural science work in government and other organisations where it's not the primary mission:
Translation matters more than theory.
Understanding how legal, accounting, or clinical perspectives frame problems is essential to making behavioural insights relevant.
Institutional capacity beats episodic projects.
Standing research infrastructure—panels, labs, regular meetings—signals commitment and lowers barriers to new work.
Patience and opportunism co-exist.
Some projects take years to show results. Others must be shelved and revived under different political circumstances. Both require commitment and flexibility.
Qualitative insight often convinces better than quantitative findings.
Showing stakeholders real behaviour is more persuasive than abstract data.
Patient voice and lived experience matter.
Structured engagement with those affected by policy decisions brings reality that experts and regulators don't possess.
The measure is public impact, not academic credit.
As Chin noted, successful systems often go unnoticed: people are more likely to notice a confusing form or a failed payment than a process that works smoothly. The work that matters most is the work people don't even see—precisely because it made something that should be simple actually work.
For GAABS members working in government, corporate settings, or seeking to embed behavioural science in organisations where it's not the primary mission, this conversation offers both encouragement and realism.
Behavioural science can fundamentally change how institutions serve people. But it requires building relationships, learning domains, staying committed through frustration, accepting that your best work might be invisible, and recognising that transformative impact is measured not in publications but in the millions of people whose lives are made slightly easier when something that should be simple actually is.
The PNAS special feature Decision Science in Service to All is available open access.
A Next Step Worth Taking
Members of GAABS attend webinars like this one to keep up with Behavioural Science. Our members receive early notification and offers to register for live events like this one with Professor Fischhoff and Dr. Alycia Chin. To receive your early notifications, JOIN US.
