For government and policy-makers educational research fails both to provide the answers they are looking for. Worse, much research fails to enhance professional practice. Crucially, for the position developed here, it is also criticized for being fragmented into myriad of incommensurable case studies which merely revel in their own uniqueness. On the one hand policy-makers are looking for a science of education, on the other hand it is argued that such an aspiration is based on false beliefs about what research can deliver. We will therefore attempt seriously to take the claim that we can look for causes in order to understand the reality we live (in), and we will focus primarily on "the natural world". We will argue that even if we would fully endorse the program of looking for antecedents, a dominant driver for many educational researchers, this would still not solve the problems they set out commonly to address. The paper will then illustrate the position it argued for by an analysis of "Class-size research" and some other examples of large scale population studies. Finally it will offer some concluding remarks concerning the nature of inference between evidence and recommendations for policy and practice in such studies.