While randomized control trial (RCT) have long been seen as the epitome of evidence based research, having served to modernize the field of medicine, this article questions its use to test social policy. The argument is that where RCT’s come into their own when testing a discrete treatment, measuring over time, in a controlled environment, it does not suit when the environments differ, there are many different possible solutions and there is a need for iterative learning, such as often found in schools. A teaching method that works for one teacher in a class, may not be the same for a teacher with a different set of skills and pupils. further, assuming one has found the optimal solution that can be tested over time, does not allow for honing and continuous improvement of the solution. Hence, other methods should be used to test and implement policies.
Or put differently, RCT’s have their strengths, but as with all methods, is not optimal for all settings. (see also Nisbett’s crusade on multivariate regression analysis)
The Problem With Evidence-Based Policies
Ricardo Hausmann, a former minister of planning of Venezuela and former Chief Economist of the Inter-American Development Bank, is Professor of the Practice of Economic Development at Harvard University, where he is also Director of the Center for International Development. He is Chair of the World Eco…