CR is a approach to the philosophy of science that seeks an alternative to two dominant positions (for cogent outlines, see Sayer 1992 and 2000). These are positivism and an approach variously labelled as relativism and social constructionism. Positivism seeks explanation in terms of law-like empirical regularities. It is faulted for addressing only empirical regularities rather than the underlying mechanisms producing these regularities; its basis in deductive- nomological approaches prevents it from asking why things occur as they do. Social constructionism focuses on the social processes through which people create meaning. It tends to ignore the influences of structures that lie outside the processes of social construction.
There are many variants of constructionism, and indeed of CR. The present approach follows those CR writers who recognize the value of a constructionist position but wish to add to it. Consider the example of a performance appraisal scheme. A constructionist approach would argue that ‘performance’ is not an objective characteristic but is defined through political processes (see Edwards and Wajcman, 2005: 96-101). It would go on to address how different definitions of performance are created and sustained. This is very valuable; yet there are two linked weaknesses. The first is a tendency to focus on construction to the neglect of the structural context in which it takes place. Second, revealing generic processes is stressed over causal explanation. It is true that performance is politically defined, but the ways in which the relevant processes occur will vary, and it is important in addition to ask why construction takes a particular form under given conditions.
CR argues that there are real, if unobservable, forces with ‘causal powers’ and that it is the task of science to understand the relevant mechanisms. The social world is seen as being different from the natural because it requires human intervention, but it does not follow that society is wholly the product of human design or discourse: rules, norms and institutions develop with logics independent of the choices of individual actors. CR stresses that causal powers are not necessarily activated and is thus very sensitive to the importance of institutional context. It aims to move beyond the discovery of empirical regularities to understand the mechanisms that not only produce these regularities but also determine when they will occur and when they do not.
Sayer (2000) gives examples of realist research in practice. One with resonances with IR debates is drawn from Morgan and Sayer (1988).
Conventional, ‘taxonomic’, approaches use large data sets to seek invariate relationships between independent factors and performance. But such relationships rarely exist because of the ‘openness of systems’. Morgan and Sayer switched to an intensive methodology, treating firms in causal rather than taxonomic categories. . . . [E]xplanations as to why firms behaved as they did were in fact easier to come by than would have been possible through seeking determinate statistical relationships (Sayer, 2000: 24). A further IR example would be the link between unemployment and union membership. Much UK and US research finds an inverse link, but studies in other countries found no such link or even a positive one, for reasons to do with their systems of unemployment insurance; an alternative approach to the links between unions and the development of capitalist economies then addresses sets of
Parts of this section are drawn from Edwards (2005).