|
To shape good policy, we must understand what causes the problems we are trying to cure and which cures work. Randomized trials are central to generating this knowledge. Suppose we would like to see whether one thing (e.g. "schooling") really improves a life outcome (e.g. "health"). The natural instinct is to compare the health of those who have schooling to those who don't. But this would be like comparing apples and oranges: People who have been to school are different in so many ways from those who haven't. Perhaps they have more advantaged social backgrounds, greater access to government services and so better access to schools. And only a few of these factors can be measured and accounted for in a standard statistical analysis. So this simple comparison--those with schooling to those without--may tell us little about the effect of schooling: It may instead be the effect of any of these numerous other differences like social background. If policy are set on the basis of such apples and oranges comparisons, quite a bit of disappointment may result. On the other hand, it is widely accepted that randomized trials provide a clean and simple way of figuring out what causes what. The basic idea is very simple. Suppose one group of children is randomly allocated an opportunity to go to school that they would not have otherwise had, while another group did not get that chance. This may occur, for example, if a school construction program begins by building schools in a few areas randomly. Now, we can compare apples to apples. The students who got the schooling opportunity are the same as those that did not, except for the flip of a coin that determined whether a school would be built near them. This means that any health outcomes could only be due to the effect of the school. Are randomized trials practical? Our experience shows that by being creative, the methodology of randomized trials can be applied to answer a wide variety of important questions. For example, even though race is obviously something that cannot be randomly assigned to people, Lab founder Sendhil Mullainathan (along with Lab affiliate Marianne Bertrand) used a randomized trial to look at how the perceived race of a job applicant affects the chances of getting an interview. Numerous studies such as these have now shown that randomized trials can improve our understanding of social situations and thereby help shape better policies to fight poverty.
Links to relevant news and papers:
|
