...That we haven't tried.
Ian Parker has an article, "The Poverty Lab", in this week's(May 17, 2010) issue of The New Yorker. This article is a profile of Esther Duflo, this year's winner of the John Bates Clark award and a co-founder of the Abdul Latif Jameel Poverty Action Lab (J-PAL). J-PAL conducts controlled experiments in the field in developing countries.
This is neat work. As I understand it, it can be extended. Does J-PAL conduct experiments designed to implement more than one treatment at once? In such experiments, one would analyze the results with Analysis Of Variance (ANOVA), instead of a T test (or the Mann-Whitney-Wilcoxon test, which is the corresponding nonparametric test). This would probably be challenging on the scale of their experiments, but part of the point of the design of experiments is to allocate resources efficiently.
Field experiments, in some sense, are an extension of the methodology of laboratory experiments, seen in the work of, for example, Daniel Kahneman and Amos Tversky. The design of experiments could also be extended back into theory by applying it to simulations and computer programs implementing theoretical models. Duflo's work tells us about the world, while this would tell us more about economic theory. I'm thinking of Stefano Zambelli's work on aggregate production functions.
6 months ago