Rigorous randomized evaluations of programs to fight poverty can go a long way in proving what works and what doesn't, researchers Annie Duflo and Dean Karlan write in The New York Times.
Studies on poverty intervention programs have demonstrated, for example, that while microloans are helpful to the poor, they don't actually increase borrowers' incomes on average. These same types of studies can help program designers retool their efforts to be more effective. "Hope and rhetoric are great for motivation, but not for figuring out what to do. There you need data," the authors say.