Epidemiological investigations very often allow saying with certainty that there is a relation between a macrophenomenon F and a certain value of increase or decrease of a certain pathology P. The abductive inference which leads to such a conclusion, however, does not allow establishing which cases of the pathology P are actually caused by cases of F and which are not. Given that in order to establish penal responsibility in most Western countries the law requires that there is a causal relation among token - events (which here we will identify with so-called Kim-events) it is frequently argued that in such cases no causal relation, and a fortiori no penal responsibility, can be properly established. The problem will be examined with the tools of quantified conditional logic. The aim of the paper is to argue that identifying a causal relation in which causes and effects are at a different level of determination does not prevent establishing penal responsibilities.