In studies in psychology and education it is essential to think clearly about causal mechanisms. I will outline the use of path diagrams as tools for representing, reasoning about, and testing causal models. The examples I will use come from studies of children’s reading and language disorders. In studies of such disorders we can probably never practically or ethically manipulate the ultimate causes (genes and environments) of a disorder. I will argue, however, that identifying causes at the Cognitive Level of explanation is crucial for planning effective treatments/interventions for these disorders. Furthermore, if interventions are successful we may, using mediation analyses, get close to identifying the proximal causes for different disorders.