Lies, damned lies, and RCT : a J-PAL RCT on rural microcredit in Morocco

How can we explain the academic success of a randomized study whose validity, both internal and external, is very problematic ? Drawing on a study conducted on Moroccan rural microcredit by J-PAL, this article uses analytical tools from statistics, political economy and sociology of science to answer this question. It describes the entire study production chain, from sampling, data collection, data entry and recoding, estimates and interpretations to publication and dissemination of results. It highlights a particularly aggressive strategy carried out throughout the study process and in the field of research. This allows J-PAL researchers to put the past behind them, including by freeing themselves from a "data culture", rejecting criticism and bypassing the basic rules of scientific exercise throughout the research process. Well beyond J-PAL, our analyses question the supposed superiority of randomized methods while reflecting a growing unease within the academic field, which is less and less successful in enforcing the basic rules of ethics and scientific deontology