Why We Need a Science of Scaling

Poverty is a large-scale problem: even at a very low threshold of $2.15, around 8.5% of the world or almost 700 million people live in extreme global poverty (World Bank 2024). At any given time, we need many rigorously tested solutions to tackle multiple issues arising out of, or related to extreme poverty: physical and mental health, education, livelihoods, gender violence, climate resilience. Further, we need to scale successful solutions rapidly and well. Randomized controlled trials solve the first of these problems (testing many different solutions) but to scale successfully, we need to complement the science of testing with the science of scaling. 

Consider this: a migration study in which a subsidy of $11.50 (enough to pay for a round trip bus fare to a nearby city and some food) given to landless agricultural workers during the lean season seems very successful. The percentage of agricultural workers heading to the city in search of jobs increases from 36% to 58% and their families back home consume an average of 600 calories per person per day more. But when the program is scaled up to reach hundreds of thousands of households, results are disappointing. The program is withdrawn (Mobarak 2022). What went wrong and can we identify these threats to scaling earlier on in the program? 

Scaling a program is not about simply expanding a successful initial pilot like the one above. For one, questions around “design for scale” have often not been considered when testing a theory or mechanism in tightly controlled conditions: who will pay for the program at scale over many years (financial sustainability),what are threats to implementation fidelity and how could we counter those? Other questions arise. Will a program that successfully delivers a bundle of specific health services to rural populations in Sierra Leone work as effectively in India (external validity). What are the effects of a school-based gender violence prevention program on non-beneficiaries? (spillovers and network effects). If all entrepreneurs in a given area receive a particularly successful training to increase sales and profits, does that reduce competitiveness and therefore profits of the entrepreneurs themselves? (general equilibrium effects). 

In this talk, I will discuss these considerations using case studies and examples from work that the Yale Research Initiative on Innovation and Scale (Y-RISE) has been doing. In particular, given my own background as a behavioral scientist now working with researchers on large-scale field trials, I will discuss the role that interested psychology researchers could play in helping to develop a science of scaling.

Sponsorship of an event does not constitute institutional endorsement of external speakers or views presented.