In the movie 21, a fictionalized recounting of how several former MIT students used mathematical skill to win big at blackjack, being "pretty good with numbers" looks like a quick way to get rich in Las Vegas. Real life at MIT may seem less glamorous, but it's actually more exciting, because our students and faculty are using their gift for numbers and analysis to change the world.

Take, for example, the problem of global poverty. Over five decades, the world has spent upwards of $2 trillion on development aid, without many lasting results. One reason is that, to a striking degree, aid money is spent without an understanding of which interventions really work. It's as if a new drug could enter the market simply because some patients who took it got better. We've long understood that without a control group for comparison, there's no way to tell whether symptoms improved because of the drug or for some unrelated reason.

Today, 2.6 billion people struggle to survive on less than $2 a day. Given the magnitude of the problem, it's imperative to identify which antipoverty efforts work best. That is exactly the aim of MIT's Abdul Latif Jameel Poverty Action Lab, which is headed by MIT economists Abhijit ­Banerjee and Esther Duflo and includes a growing network of international researchers. The lab is leading a quiet revolution. The idea is simple: to identify the most effective ways to alleviate poverty by using the same kind of rigorous, scientific, randomized trials routinely used to test new drugs.

For instance, if you wanted to prevent the spread of HIV to a new generation in rural Kenya but had limited funds, would you teach schoolgirls about HIV? Or would you help girls stay in school by covering costs like required uniforms? Or inform girls that in Kenya, older men (particularly those 20 to 45) are more likely than younger men to carry HIV? Researchers at the Jameel Poverty Action Lab test the value of interventions like these by comparing a group that participates in a program with a similar group that does not. By evaluating the outcomes, they measure how well an intervention works. By comparing results from different interventions, they can determine cost effectiveness, too.

Sometimes they reach surprising conclusions. In Kenya, Duflo, with colleagues ­Pascaline Dupas and Michael Kremer, found that keeping girls in school was more effective in reducing girls' risky behavior than teaching the standard HIV curriculum. Moreover, alerting girls to the higher HIV rate among older men dropped the rate at which children were born to teen mothers and older fathers by a stunning 65 percent. By identifying the best ways to reduce the spread of HIV to a new generation, these findings could help change the course of the epidemic.