Esther Duflo hopes to take the measure of a wide range of anti-poverty programs. Applying scientific methodology, her colleagues and students at the MIT Poverty Action Lab are approaching the projects of well-intended governments and NGO’s (non-government organizations) with a fresh eye. “We have a spotty and scattered idea of the most effective ways to deliver social impact,” says Duflo, so evaluating what works is important.

She describes the U.N. goal of ensuring that all children worldwide attend school. Many programs aimed at achieving this goal simply don’t deliver the results intended. Some approaches that gained credibility and support involve giving away school uniforms and providing free meals. But, says Duflo, “Sometimes ideas that become conventional wisdom are erroneous and need to be rethought,” especially since the “budget for fighting poverty is extremely limited and will remain limited.”

Researchers compared a program that aimed to improve children’s school attendance through a program of deworming, with a program that paid kids to go to school. Testing these projects “the way we do drugs, with treatment and control groups chosen randomly,” Duflo found that the $3 per year deworming program resulted in a dramatically higher increase in school years attended than did the $6,000 per year program paying kids to attend school.

Duflo insists on “being pragmatic about what works and what doesn’t,” and attempts to evaluate not just the effectiveness of programs but the auditing of corruption often found in social programs in the developing world. If the groups implementing a program partner early with Duflo, and embrace a rigorous evaluation of their work, they can often abort ineffective approaches and expand successful ones, maximizing their anti-poverty investment, says Duflo. “The best quality research must form the basis of good policy,” she concludes.