Behind the scenes of science in action: tinkering with a randomized control trial in Morocco

International audience This article is a ‘replication in context’ of a flagship randomised control trial (RCT) conducted in Morocco on microcredit. ‘Replication in context’ consists in combining the quantitative replication of an RCT with a contextualised analysis of its implementation and its political economy, in the sense of the interplay between different stakeholders with divergent and potentially conflicting interests, constraints and powers. ‘Replication in context’ draws on quantitative and qualitative data and uses the tools of statistics, political economy and sociology of science. This method allows us to describe the entire RCT production chain, from sampling, data collection, data entry and recoding, estimates and interpretations to publication and dissemination of the results. We find that this particular RCT does not respect the key principles of randomisation (imbalanced sampling and contamination) nor those of statistics (coding and measurement problems, poor-quality data and arbitrary trimming procedures). The qualitative analysis highlights the difficulties of implementing a randomised protocol in the real world. Beyond this particular case study, our analyses call into question the supposed superiority of randomised methods, echoing the growing unease in an academic field increasingly struggling to enforce the basic rules of ethics and scientific deontology.