Before we get involved.
We’re a small operating foundation — we run programs rather than make grants. Each program is a multi-year commitment, so the questions we ask before getting involved are narrow. Below: how we think about the work, the five questions that screen new commitments, and the reading that’s shaped both.
Four commitments that come first.
The five-question screen below is operational. These four commitments are upstream of it — the stance from which we even start asking those questions. They’re the part of the work that doesn’t change when the program does.
Stay in one place long enough to be useful.
We default to multi-year partnerships with a small number of local educators and institutions, not one-off visits across many. The cost of being wrong about a place is borne mostly by the community, so we move slowly into a new geography and stay once we’re in.
Read what others have already learned.
Most of the questions we face have been studied carefully by someone else. Before we run a new program or back a new model, we read the evaluations, the meta-analyses, and the field notes — and we ask collaborators inside the community what we’re still missing.
Pair with the local organization. Don’t replace it.
In every place we operate, the program is anchored by an organization that lived there before we arrived and will be there after. Our job is to add capacity and learning, not to substitute for the relationships, trust, and continuity we couldn’t build from outside.
Measure honestly, including the disappointments.
We track survey responses, program completion, follow-on enrollment, and longer-run trajectories where we can. When a program underperforms its premise, we say so and adjust — in writing, on this site, and in how we deploy the next dollar.
These aren’t aspirational. If a prospective program would push us off any one of them, that’s a signal we’re the wrong funder — even if the five-question screen looks promising.
Five questions we ask.
None of these is a checkbox. A clear “yes” to four out of five usually moves us toward a deeper conversation; a clear “no” on any one is usually disqualifying.
- 01
Has the program model actually been tested?
Has someone — ideally not us — already run a credible field study, randomized evaluation, or long-run cohort follow-up on a program of this shape? If the underlying mechanism has never survived rigorous evaluation, we’re rarely the right first funder.
→ The evidence base for what we fund - 02
Does the pedagogy line up with how children and adults actually learn?
Programs aimed at young people need to be honest about cognitive and socio-emotional development. The mechanisms are well-studied; good programs lean on them deliberately.
→ How we think about education and opportunity - 03
Is CS or entrepreneurship the right vehicle for this community?
CS and entrepreneurship education compound — but only when the surrounding capital, mentorship, and labor-market infrastructure can turn skill into opportunity. We try to be specific about where this case is strong and where it isn’t.
→ CS and entrepreneurship as paths out of poverty - 04
Can we measure whether it’s working — in a way that lets us learn?
Measurement-for-reporting and measurement-for-learning are different design problems. We design ours around the second — short feedback cycles, beneficiary voice, and metrics we’d act on even when they tell us something uncomfortable.
→ How we measure impact - 05
Do we have research, relationships, or advisory trust that ground this work?
We operate in places we don’t live. We rely on Chuck’s peer-reviewed research, long-term partnerships with practitioners on the ground, and advisors whose judgment keeps us honest about what we can’t see from a distance.
The evidence base for what we fund
Field experiments and randomized trials we lean on when deciding which kinds of programs are worth backing.
- Abhijit V. Banerjee & Esther Duflo · 2011
Poor Economics: A Radical Rethinking of the Way to Fight Global Poverty ↗
The book that made it normal for serious philanthropists to ask whether their programs actually work. Banerjee and Duflo's case for randomized evaluation — and the long humility about what does and doesn't move the needle for people living in poverty — sits underneath almost every diligence question we ask before funding a program.
- Abdul Latif Jameel Poverty Action Lab
What Works in Education: J-PAL Education Bulletins ↗
J-PAL's running synthesis of the randomized evidence on what improves learning in low- and middle-income settings — pedagogy, teacher incentives, technology, structured curriculum, conditional cash transfers. When we evaluate an education program, this is the first place we check to see whether the proposed mechanism has been tested at scale.
- Banerjee, Duflo, Goldberg, Karlan, Osei, Parienté, Shapiro, Thuysbaert & Udry · 2015
A Multifaceted Program Causes Lasting Progress for the Very Poor: Evidence from Six Countries ↗
The six-country randomized evaluation of BRAC's Ultra-Poor Graduation model — asset transfers, training, consumption support, and coaching delivered as a bundle. Effects persisted years after the program ended. This is the cleanest evidence we have that a structured, intensive program can durably lift people out of extreme poverty, and it shapes how we think about the Uganda refugee work.
- Innovations for Poverty Action
Small and Medium Enterprises Program Brief ↗
IPA's synthesis of what the rigorous evidence says about SME and microenterprise interventions — capital, training, mentoring, and combinations of the three. A useful corrective to the assumption that any training program for entrepreneurs will produce business growth. Helps us be honest about which parts of our own entrepreneurship programs are evidence-backed and which are still hypotheses.
How we think about education and opportunity
How children actually learn — academically, socially, in character — and what that means for the programs we evaluate.
- Paul Tough · 2012
How Children Succeed: Grit, Curiosity, and the Hidden Power of Character ↗
Tough's synthesis of the research on non-cognitive skills — persistence, self-control, conscientiousness — and how they get built (or fail to get built) in children from different backgrounds. It's why we care about character formation alongside CS and entrepreneurship skill-building, and why we look for programs that take socio-emotional development seriously rather than treating it as a soft add-on.
- Carol S. Dweck · 2006
Mindset: The New Psychology of Success
Dweck's growth-mindset research — that students' beliefs about whether ability is fixed or malleable affects how they engage with hard problems — has been replicated, contested, and refined. Even after the 2010s replication debates, it holds up well enough that we use it as a design constraint: programs we fund should leave students more confident that effort matters, not less.
- Daniel L. Schwartz, Jessica M. Tsang & Kristen P. Blair · 2016
The ABCs of How We Learn: 26 Scientifically Proven Approaches, How They Work, and When to Use Them ↗
Schwartz (Stanford GSE Dean) and colleagues walk through 26 learning mechanisms — from analogy and elaboration to spacing, generation, and worked examples — and when each one is actually appropriate. We use it as a sanity check when we hear about a new pedagogical model: is it leaning on a real mechanism, or is it a brand?
- Collaborative for Academic, Social, and Emotional Learning (CASEL)
The CASEL Framework for Social and Emotional Learning ↗
The CASEL five-competency framework — self-awareness, self-management, social awareness, relationship skills, responsible decision-making — has become the common vocabulary for taking character and socio-emotional development seriously inside schools. We use it to evaluate whether a K–12 program treats these competencies as deliberate instructional goals or as a happy by-product.
- William Damon · 2008
The Path to Purpose: How Young People Find Their Calling in Life ↗
Damon (Stanford GSE, Center on Adolescence, Chuck's colleague) argues from hundreds of interviews that the missing ingredient in youth development is purpose — sustained intention that's meaningful to the self *and* consequential beyond the self. The second half is the part most education programs leave out. We treat it as a design requirement: a program that builds skill without building purpose is half a program.
CS and entrepreneurship as paths out of poverty
Why CS and entrepreneurial skill-building can change the trajectory of underserved communities — and what good programs look like.
- Jane Margolis (with Rachel Estrella, Joanna Goode, Jennifer Jellison Holme & Kim Nao) · 2017
Stuck in the Shallow End: Education, Race, and Computing ↗
Margolis's multi-year study of why Black and Latino students in LA high schools were systematically locked out of meaningful CS instruction — and what it took to change that. The clearest single account of why equitable CS access is a structural problem, not a motivational one. Our baseline text for any K–12 CS work we fund.
- Code.org, CSTA & Expanding Computing Education Pathways Alliance
State of Computer Science Education ↗
The annual U.S. landscape report on K–12 CS access — which states require CS, which school systems offer it, who is actually enrolled, and where the equity gaps are widest. We use it to ground-truth where our K–12 work fits in the larger movement and to spot which places still have no functioning pathway.
- Salman Khan · 2024
Brave New Words: How AI Will Revolutionize Education (and Why That's a Good Thing) ↗
Khan's case for AI tutoring in classrooms — and the practical experience Khan Academy has accumulated with Khanmigo. We don't agree with every prediction, but this is the most thoughtful single account from someone who has actually shipped AI tools into schools at scale. It's directly relevant to our AI-literacy work in Molokai and Penang.
- AnnaLee Saxenian · 2006
The New Argonauts: Regional Advantage in a Global Economy ↗
Saxenian's account of how immigrant engineers trained in Silicon Valley returned to Taiwan, India, China, and Israel and built technology entrepreneurship ecosystems back home. It's the structural argument for why investing in CS and entrepreneurship education in underserved regions can compound — talent moves, knowledge moves with it, and the diaspora becomes part of the institution-building.
- World Bank Group
Building Digital Skills for the New Economy ↗
The World Bank's running synthesis of digital-skills programming across developing economies — what works for basic digital literacy, intermediate technical skills, and advanced CS pathways. Useful for thinking about the broader pipeline our partners in Uganda, Tanzania, and Malaysia are operating inside, and which rungs of that pipeline are most under-supplied.
How we measure impact
Frameworks and field guides that shape our internal measurement practice.
- Acumen / 60 Decibels
Lean Data Field Guide ↗
Acumen's practical playbook for collecting customer voice data from low-income beneficiaries cheaply and well — short mobile surveys, sensible benchmarks, comparable across programs. We use the underlying approach to keep our own outcome measurement honest without pretending we can run randomized trials on every program.
- Matthew Forti, Tracy Foster & Bridgespan colleagues
Measurement as Learning: What Nonprofit CEOs, Board Members, and Philanthropists Need to Know to Keep Improving ↗
Bridgespan's series on measurement-for-learning rather than measurement-for-reporting. The distinction matters: most foundation measurement systems exist to satisfy a board, not to help a program improve. We try to design ours around the second goal.
- Global Impact Investing Network (GIIN)
IRIS+ System: Generally Accepted Impact Measurement Standards ↗
GIIN's catalog of standardized impact metrics, organized by theme and SDG. Useful as a vocabulary — when we describe outcomes for a program, IRIS+ is usually where we go first to see if there is already a widely-used definition rather than inventing our own.
- Mulago Foundation
Design Iteration Format (DIF) ↗
Mulago's one-page program design tool — what is the impact you're after, what is the unit of behavior change, what is the actual measurable indicator, how does that line up with what the organization can actually deliver. We have not yet met a foundation tool that forces clearer thinking in less space.
From our own research
Peer-reviewed work by Chuck that informs the Foundation's program design — Stanford alumni studies, refugee entrepreneurship, immigrant founder economics, and online education at scale.
- Charles E. Eesley & Yong Suk Lee · 2021
Do university entrepreneurship programs promote entrepreneurship? ↗
Stanford alumni data testing whether university entrepreneurship programs actually cause the entrepreneurship rates of their participants. After correcting for selection, effects on raw founding rates are weaker than naive OLS suggests — but program participation meaningfully lowers startup failure and raises revenue conditional on founding. Disciplines us against assuming a well-funded E-program is automatically producing founders, and toward programs whose alumni build durable companies.
- Yong Suk Lee & Charles E. Eesley · 2018
The persistence of entrepreneurship and innovative immigrants ↗
Drawn from the 2011 Stanford Innovation Survey (140,000+ alumni), this paper examines how entrepreneurship rates differ by ethnicity and nationality. Asian American alumni are more entrepreneurial than white American alumni; non-American Asian alumni substantially less so; parental entrepreneurship predicts strongly across groups. Anchors why we keep funding international entrepreneurship programs even when domestic immigration policy is unfriendly to founders.
- Stanford Technology Ventures Program (Zahra Hejrati & Charles Eesley) · 2024
Refugee entrepreneurship in action: A Stanford Impact Labs fellowship update ↗
Field writeup from the Uganda refugee entrepreneurship work the Foundation co-funded with the King Center for Global Development. The piece that the Uganda program page draws from — what the partnership with Challenges Uganda and MUBS actually looks like in practice, what the 2024 cohort looked like, and what the research design is testing.
- Charles E. Eesley & Lynn Wu · 2020
For Startups, Adaptability and Mentor Network Diversity Can be Pivotal: Evidence from a Randomized Experiment on a MOOC Platform ↗
Randomized experiment on the NovoEd entrepreneurship platform testing which mentor connections and which kinds of adaptability predict venture outcomes. Sticking rigidly to a founding vision wins short-term and loses long-term; adaptability paired with mentors who access structurally diverse networks produces the best long-run revenue, funding, and pivot outcomes. Informs our skepticism of accelerators that pile on mentor density without thinking about network structure.
From our advisors and collaborators
Work by the researchers and practitioners who advise the Foundation and shape our thinking.
- Hadiyah Mujhid
Black Founders Built That: Lessons from Building HBCUvc ↗
Hadiyah Mujhid is the founder of HBCUvc and a Foundation advisor. Her writing on what it took to build a venture investing pipeline for Black students at HBCUs — the structural barriers, the deliberate program design, and the institutional pushback — directly shaped how we think about pipeline programs in CS and entrepreneurship.
- Brynjolfsson, Eesley, and co-authors · 2024
Generative AI as a tool for misleading information at scale ↗
Nature paper coming out of a multi-year collaboration with Erik Brynjolfsson and colleagues on how generative AI changes the cost structure of producing misleading information — and what that means for institutions that have to defend against it. Co-authoring with Erik continues to shape how we think about the AI-literacy components of our K–12 work.
- Sharon Yixuan Li
Out-of-Distribution Detection and the Reliability of Open-World AI ↗
Foundation advisor and leading AI reliability researcher at UW–Madison. Her work on out-of-distribution detection — making AI systems aware of what they don't know — is foundational to whether AI tools can be safely deployed in classrooms and small businesses, where confident-but-wrong answers carry real cost. We point K–12 partners to her talks for the reliability story.
- Yanbo Wang
Research on entrepreneurship, innovation, and institutions ↗
Yanbo Wang is a Foundation advisor and professor at the HKU Business School. His research on how institutional context shapes entrepreneurship — across Chinese, Hong Kong, and U.S. settings — informs our cross-border program design. We rely on his published work and his judgment when we think about how a program will travel from one institutional environment to another.
Last refreshed May 2026. We update this list as the underlying evidence base evolves, and as new work from our own research and our collaborators comes out.