Realistic Models? Critical Realism and Statistical Models in the Social Scieces

The aim of this paper is to question the scepticism of critical realist philosophers of science in relation to the use of statistical methods in social science research. By arguing that statistical analysis is inevitably 'deductivist' in nature (Bhaskar, 1998a; Lawson, 1997, 1998, 2001; Pratten, 1999), I believe that critical realists merely reinforce the influence of empiricism. Moreover, by confining their criticism of statistics to the social sciences, these writers adopt an unwarranted anti-naturalist stance. In contrast, I will argue that critical realism can help to resolve a number of philosophical problems in relation to the specification, assessment and interpretation of statistical models. Social scientists are increasingly aware of these issues, and it is therefore timely to reconsider how their concerns might be addressed from within the framework of critical realism. I am in agreement with the principal tenets of critical realism, which I believe to be one of the most promising forms of scientific realism to have emerged in recent years, and the criticisms that I will make in this paper are therefore intended to strengthen rather than undermine this philosophical approach.