Is giving people a sense of agency better than giving them cash? (with Richard Sedlmayr)
Spencer Greenberg speaks with Richard Sadelmeier about human agency. Richard, a funder, explains how empowering individuals with tailored information and shaping mental models can effectively reduce poverty, advocating for iterative, research-driven implementation.
Deep Dive Analysis
16 Topic Outline
Defining Human Agency and its Importance
Evolution from Evidence-Based Development to Agency Focus
Challenges of Applying Biomedical Analogy to Poverty Interventions
Information as an Agency-Oriented Intervention
Distinguishing Effect Size from Cost-Effectiveness in Interventions
Social Psychology's View: Information, Meaning, and Mental Models
Role Model Interventions and Shaping Narratives
The Grand Theory: Mental Models and Opportunity Sets
Importance of Timing in Agency Interventions
Operationalizing Agency: Examples in Education and Pregnancy
Interpreting Research Findings Amidst the Replication Crisis
Tightening the Loop: Research and Action in Intervention Design
Iterative Learning and Building Robust Interventions
Concerns with Non-Profit Iteration and Misaligned Metrics
Agency Interventions and Effective Altruism
The Agency Fund: Cultivating an Ecosystem of Research and Action
6 Key Concepts
Human Agency
Human agency refers to an individual's capability to live with a sense of purpose, self-determination, autonomy, and dignity. It's about supporting people in navigating their own decisions rather than dictating a course of action.
Evidence-Based Movement in Development
This movement, prevalent for two decades, applies randomized evaluations (like clinical trials) to development interventions to determine what works. However, its analogy to biomedical trials isn't perfectly clean due to the complex, heterogeneous nature of economic and social issues.
Heterogeneity in Interventions
Unlike biomedical interventions with clear necessary conditions (e.g., mosquito nets for parasites), economic and life decisions are highly personal and context-dependent. What works for one person (e.g., a specific training or loan) may not work for another, making it hard to find a universally 'right' course of action.
Information as an Intervention
This approach involves providing people with specific, actionable data they might not otherwise encounter, enabling them to make better decisions. An example is informing students about the returns to finishing high school, which can significantly impact educational attainment.
Mental Models
Mental models are the internal schemas, beliefs, and social constructs that shape how individuals perceive and navigate the world and their opportunities. Agency-type interventions work by updating these mental models, especially those that are particularly constraining or relate to important, non-obvious opportunities.
Research and Action Loop
This concept advocates for a tight, continuous feedback loop between research and implementation, moving beyond the traditional model where research 'finds' what works and then implementation 'rolls it out.' It involves constant iteration, testing, and learning within the implementation process itself, similar to A/B testing in product development.
8 Questions Answered
Human agency is defined as the human capability to live with a sense of purpose, self-determination, autonomy, and dignity, allowing individuals to navigate their own decisions.
Biomedical interventions often have clear necessary conditions, but economic and life decisions in poverty are highly personal and context-dependent, involving a longer list of factors (heterogeneity) that make it difficult to identify a universally 'right' course of action.
Information can be an effective intervention by providing individuals with specific, actionable data they might not otherwise have, enabling them to make more optimal decisions, such as informing students about the returns to finishing high school.
While there's a perception that information interventions have low impact, many of the most cost-effective interventions demonstrated in research have been information or intangible types, especially when they address critical missing data or misconceptions.
To an economist, information is data, but to a social psychologist, receiving information involves making meaning of it, which is influenced by its source, how it's interpreted, how it files into existing mental models, and how it shapes perspectives and beliefs about oneself and the world.
The replication crisis highlights that findings from social science may not always generalize due to complex, hidden contextual factors (hidden moderators). For agency interventions, this suggests a need for a tighter loop between research and action, focusing on continuous learning and iteration rather than seeking universal, 'final' truths.
Robustness in iterative learning can be achieved by integrating research and implementation, continuously testing and improving interventions (like A/B testing in tech), and focusing on outcomes that are measurable and directly linked to the program's goals, rather than just easily collected metrics or donor satisfaction.
Research, including trials with GiveDirectly, indicates that agency-based interventions can achieve higher poverty reduction dollar-for-dollar compared to direct cash transfers, especially because the means of reaching people with intangibles can become increasingly cheaper over time.
15 Actionable Insights
1. Update Constraining Mental Models
Design interventions to update individuals’ mental models about themselves, their surroundings, and their future, particularly focusing on those that are most constraining. This approach aims to unlock potential by shifting fundamental beliefs and perspectives.
2. Leverage Role Models and Narratives
Introduce individuals to compelling role models and narratives that challenge existing limiting beliefs and help them reimagine their potential. This approach, exemplified by girls watching an inspirational movie, can significantly impact self-conception and motivation.
3. Provide Actionable, Customized Data
Offer individuals specific data relevant to their decisions, especially when they might have misconceptions, to help them optimize their choices and improve outcomes. For example, informing students about the actual income returns to finishing high school can significantly reduce dropout rates.
4. Understand Context for Interventions
Before implementing information-based interventions, conduct thorough research to understand the target audience’s existing beliefs, motivations, and capabilities. This ensures the information provided addresses a genuine need or misconception, increasing its effectiveness.
5. Time Interventions to Key Decisions
Identify critical decision points in individuals’ lives and deliver targeted information or support when it is most relevant and actionable. This ensures that insights address immediate needs and influence high-stakes choices, such as preparing for college or making birth plans.
6. Implement Lean Learning Iteration
For social interventions, adopt a continuous “lean learning” approach, integrating research and action in a tight loop. This means constantly iterating, testing, and improving interventions based on real-time feedback, similar to product development.
7. Integrate Randomization in Implementation
Embed randomization and A/B testing directly into the rollout of products or interventions to continuously gather data and learn what works best. This allows for ongoing optimization and ensures that improvements are applied to benefit all users.
8. Beyond Trivial A/B Testing
Utilize A/B testing and embedded experiments to gain substantial knowledge beyond simple optimizations like button colors. Focus on learning how to effectively design and improve programs to achieve specific social outcomes, generating powerful, actionable insights.
9. Align Iteration with True Impact
When iterating on non-profit programs, actively guard against optimizing for easily measurable metrics or donor preferences instead of genuine social impact. Strive to connect feedback loops to ultimate outcomes, even if they are more challenging to quantify.
10. Improve Existing Cash Transfers
Instead of only starting new cash transfer initiatives, prioritize improving the design and framing of existing large-scale government cash transfer programs. Consider how narratives and meaning associated with these transfers can significantly amplify their impact.
11. Fund Capable Learning Organizations
Prioritize funding organizations that have proven their capability to implement and continuously improve agency-based interventions, rather than just funding specific, discrete solutions. This approach supports the development of “learning muscles” for sustained and evolving impact.
12. Bayesian Approach to Funding Ideas
When assessing new intervention ideas, use a Bayesian approach by extrapolating from existing knowledge and similar successful interventions, even if direct evidence is scarce. This helps determine which ideas are promising enough to warrant further testing and investment.
13. Answer Life-Changing Questions
Visit clearerthinking.org to answer free, scientifically-backed “life changing questions” designed to provide valuable self-insights. These questions can also be used as a physical card deck to foster deeper connections with friends and family.
14. Use MindEase for Stress/Anxiety
If experiencing stress or anxiety, use the free MindEase app (mindease.io) for scientifically proven, interactive exercises that help calm your mind in under 10 minutes. The app tailors to your situation, measures progress, and provides insights into effective techniques.
15. Manage Social Media Post Anxiety
To reduce anxiety when posting on social media, be extra careful with critiques and proofread for mistakes. If a mistake is found, fix it promptly and acknowledge the correction to ease your mind.
5 Key Quotes
When you look at the world of kind of, you know, there's thousands of social science, including just development RCTs now, and you sort of look at the ones that are most surprising. You mean randomized detail trials, right? Right, like randomized evaluations. When you look and sort of squint and you try to make sense of where are there really exciting things happening? Where is a lot happening with little? So it's often the case that you're kind of, the intervention engaged people's consciousness. It didn't decide things for them or see them as, you know, some instrument to get impact out of, but try to just add the margin, support them in the navigation of, of their own decisions.
Richard Sedlmayr
All humans try to navigate the world with some deliberation, but they don't navigate the actual physical world. They navigate the world that's in their head. And that world is kind of comprised of certain schemas and beliefs and social constructs that it's like a stock of who you are, the stock that's kind of shaped by your experiences. And that determines basically how you engage with your opportunity sets. And that determines what you see and what you don't see.
Richard Sedlmayr
The replication crisis, part of the source of the replication crisis is sort of an unrealistic expectation that people have about the capabilities of social science to answer things that with a sort of degree of certainty that isn't warranted when you work in these kinds of, you know, complex environments where there's all this heterogeneity and all these moving parts.
Richard Sedlmayr
I think the idea that there's humans might be so far from their true sort of inherent potential, speaking for everybody, right, but certainly also for people who grow up in low socioeconomic status context.
Richard Sedlmayr
I think ideally you would want to scale the organizations that are capable of doing these things and doing them in an ever better and better way. So the end game, I think, again, this is all kind of work in progress, but the end game after some successful pilots, I think are more likely to be like organizational investments that don't just invest in a discrete solution or a discrete bot or whatever it might be, right? That invest in an organization that's demonstrated the capability to make a dent in something.
Richard Sedlmayr