Measuring impacts of different policy tools and trying to evaluate them with the help of average treatment effects is a widespread method. Since the sample in these studies usually consists of many different individuals, the within variation becomes very large. Therefore, it becomes difficult to differentiate, which individuals get really affected and how strong these effects are. To reduce this variation, I will investigate different sub-groups of an already conducted experiment in Benin. The goal is to examine whether the effects differ across sub-groups and consequently find opportunities to improve the experiment from the subsequent insights.