Traditionally, research has been seen as a process in which particular cases are studied in order to produce generalisations that can later be applied to other situations. This is arguably the case, for instance, of plain statistical generalisation from samples to populations, but also of grounded theory, local theory and democratic theory. Other research approaches, such as case study research and action research, have challenged this conception and have formulated a process in which transfer takes place directly from particular cases to other particular cases, thus bypassing generalisations. Nevertheless, I argue that, even in research on single cases, and regardless of whether it is descriptive, explanatory or normative, any piece of research unavoidably produces, supports, modifies, qualifies or refines generalisations in the course of a research project. These generalisations are constitutive of the very descriptions, explanations and normative justifications used to talk about the particular. Nevertheless, they vary in their degree of explicitness, certainty and complexity, as well as in the substantive dimensions they generalise on. The neglect of this characteristic in the educational literature may stem from the inductive assumption that knowledge in research is produced when one first finds something about one or more cases or situations, and then generalises (or does not generalise) the results to other contexts. But generalisation exists all along the process.