Data modelling is viewed as a way of perceiving, organising and describing data. This shares important issues with the epistemological and ontological questions: how is valid knowledge acquired, and what is the nature of the world to which it refers? The philosophical positions taken on these questions correspond to antithetical paradigms: the realist-positivist and the nominalist-interpretivist. These are combined into a framework through which the various existing data modelling approaches are related. The most common data modelling paradigm, the realist-positivist, is explored in some detail. The result sheds doubt on the efficacy of this paradigm as a basis for data modelling approaches.