Most published research in information systems is underpinned by a positivist or empiricist philosophy. This generally involves the collection of quantitative data and its subsequent analysis using some form of statistical modelling, often multivariate such as regression. Alternative paradigms, such as interpretivism, often critique such statistical analysis on the grounds that the social world is inherently different to the material world. However, this often leads to a strongly anti-realist position which denies the existence of any forms of external social structures. The purpose of this paper is also to put forward a critique of traditional statistical modelling and analysis but from a different direction - critical realism. This maintains the ontological reality of social systems whilst recognising the cultural and historical epistemological limits to our knowledge of them.