MIT's Abdul Latif Jameel Poverty Action Lab (J-PAL) is technically located in Building E60 on the edge of east campus. But J-PAL's real laboratory is a primary school in a sub-Saharan African town, a household kitchen in a home in rural India, an unemployment line in a suburb of Paris-anywhere antipoverty programs are necessary to improve a population's health and well-being.
J-PAL is dedicated to fighting poverty by ensuring that policy decisions are based on scientific evidence. As part of that effort, J-PAL undertakes, promotes the use of and disseminates the results of randomized evaluations of poverty-alleviating programs.
Randomized evaluations provide clean, simple and reliable estimates of impact that can help policymakers determine what works, and what doesn't work, so that resources can be directed towards effective projects that improve people's lives. For example, J-PAL researchers discovered that a very inexpensive de-worming program in Kenya substantially improved children's health and primary school attendance.
To spread its research and promote the use of randomized evaluations, J-PAL organizes courses at MIT and around the world. Last summer, 40 development practitioners from eight African countries traveled to Abuja, Nigeria, to participate in one such course led by economists and graduate students from J-PAL.
"An important weapon in the fight against poverty is good evidence about what works," said Esther Duflo, J-PAL co-founder and co-director and the Abdul Latif Jameel Professor of Poverty Alleviation and Development Economics.
"The aim of this course was to build the capacity of others--particularly researchers in Africa--to answer questions about what works and to be able to interpret the often complex evidence about the effectiveness of alternative approaches to reducing poverty."
Like previous courses in Cambridge and Chennai, India, the course in Nigeria was designed to provide a thorough understanding of randomized evaluations and pragmatic step-by-step training for conducting one's own evaluation.
Through lectures, group work and case studies, course leaders presented topics such as why and how to randomize evaluations, measurements and outcomes, sample size and data management, managing threats and analyzing data.
Participants came from Ghana, Kenya, Malawi, Niger, Nigeria, Rwanda, Sierra Leone and Uganda. Some were officials from health and education ministries, while others were managers and researchers from development organizations working in Africa.
Participants were expected to apply the knowledge and skills acquired during the training course to evaluate social and development programs in which they are directly involved.
"The workshop opened us up to whole new experiences in conducting evaluations," said Colina Kpayagula of the Institutional Reform and Capacity Building Project in Freetown, Sierra Leone. "It also exposed us to different ways of analyzing issues before, during and after the evaluations, to ensure that the results are accurate. It's interesting to note the importance of preparing for evaluations at the very start of the project and not at the end, as we are used to doing. I believe the workshop is what one will call a once-in-a-lifetime experience, giving you a fresh, new outlook."
The J-PAL course in Nigeria was followed by a World Bank workshop that brought teams from the education ministries of nine different countries together to discuss how education policy can use randomized evaluation to work towards evidence-based policy and programs, and how their countries' education strategies can incorporate randomized evaluations.  Â
The J-PAL course was organized in cooperation with the MIT International Science and Technology Initiatives (MISTI) and sponsored by Total. A sponsor of MISTI's MIT-France Program, Total has contributed graduate fellowships to MIT for students working in energy and social and economic development and for students from African countries, and is actively involved in MIT's Joint Program on the Science and Policy of Global Change.
A version of this article appeared in MIT Tech Talk on November 7, 2007 (download PDF).
