The Need for More Randomized Controlled Trials in U.S. Social Policy Interventions?

Feb 17, 2010, 3:45 AM, Posted by

Esther Duflo, a development economist at MIT, gave a thought-provoking talk at TED on using randomized controlled trials to study the impact of anti-poverty interventions in developing countries. Instead of trying to answer the big, controversial question, “Does (international monetary) aid work?” Duflo tries to answer smaller, local questions that provide insight to the big question.  For example, mosquito nets are highly cost-effective for preventing malaria, but they’re not being used widely.  Duflo wanted to know why and whether cost had something to do with it.  Are poor villagers more likely to use mosquito nets if they have to buy them (at a low, subsidized price) versus getting them for free?  Her research showed it’s more effective to give the nets away for free. 

What’s most appealing about Duflo’s research is that she’s able to show what works (or doesn’t) and she can back it up with data.  Randomized controlled trials are the gold standard- for example, the 1971 Rand Health Insurance Experiment is still cited today, yet, they’re not extensively conducted on U.S. social policy interventions (with the exception of education) primarily due to cost, ethical issues, and complexity of the research design.  Knowing what works would ensure that scarce funds are directed towards those policies that have meaningful, lasting impact to improve the lives of millions of people.

Duflo’s talk raises some important questions: When is appropriate to conduct randomized controlled trials in U.S. social policy?  Should we conduct more randomized controlled trials in health and health care to understand which interventions work and to invest in? Or are randomized controlled trials too controlled, localized, unrealistic and infeasible?

This commentary originally appeared on the RWJF Pioneering Ideas blog.