La randomisation à l'épreuve du terrain. L'exemple du projet de microassurance sky au cambodge

</titre> Drawing on a case study ? the randomized evaluation of the impact of Sky, a micro-insurance program in Cambodia ? this article examines the practical arrangements of collaboration and compromise between the constraints of academic ?rigor? and those of practitioners and between various forms of knowledge production (quantitative/qualitative). The analysis examines the sets of protagonists in order to highlight the multiple challenges of these two forms of articulation in the case of randomized control trials. Compared with other quantitative methods, this double articulation is unavoidable but also particularly complex, for the following reasons : the narrowness of the research questions, the very strong constraints of the survey protocol and the necessary collaboration of field practitioners in its implementation.