Realism and Social Science

But does this realist account of science have any bearing on the social sciences? Is it any more acceptable as a model for the social sciences than the positivist programme? Here, there is quite a lot of disagreement among realists.

Roy Bhaskar advocates yet another position, which he terms ‘critical naturalism’ (Bhaskar 1998). This approach accepts the main burden of the interpretivist, or hermeneutic critique of the positivist version of ‘naturalism’ (that is, in this context, the proposal to study society on the model of the natural sciences). However, he argues that despite, and even because of, the fundamental differences between natural and social objects of knowledge, it is still possible to have a science of society in the same sense as the sciences of nature, but not necessarily of the same form as them and not employing the same methods.

The sort of knowledge that can be achieved, and the methods for achieving it, will vary according to the subject-matter of a discipline. An implication of realism, then, is that the answer to the question ‘Can there be a science of society?' will turn on what sort of thing ‘society' is.

Bhaskar's (1979, 1998) starting point was one of scepticism about existing strategies in the social sciences. For him, indeed, what has to be established was, precisely, whether a scientific study of society was possible. To begin with, existing explanatory work in social science would beg the question at issue (in a way that using this strategy in relation to the natural sciences would not). Bhaskar's approach to the question ‘Is society the sort of thing that can be studied scientifically?' is, as before, to rely on transcendental arguments.

Social Ontology: Structure and Agency

The examples we used earlier, of someone cashing a cheque or being a student, will do as a starting point. Both actions, as described, are possible only on condition that the agent is situated in a set of institutional relations, which exist prior to and independently of their actions. For someone to cash a cheque there has to be a money economy, a banking system and so on, but also they have to be located in that system as an account holder (not to mention having money in their account or an overdraft facility!). This is the sort of point that a structuralist might make. However, it is also true that without individual people and their activities there could be no such things as accounts, cheques, banks and economies. Institutions do not exist independently of the activities of people, but, on the contrary, are nothing but regularities in the aggregate patterning of those activities. This is the sort of point a methodological individualist might make. A third possibility might be a synthesis of the two rival views. Bhaskar refers to the dialectical view of Peter Berger and his associates (see especially Berger and Luckmann 1967). On this account, society is an outcome of individual agency, which then reacts back upon individuals. But Bhaskar rejects this account, too, since it does not sustain the persistence of social structures as both conditions and outcomes of human agency, or of people as both products of and conditions of possibility of social structures.

This solution to the structure/agency problem, then, involves a commitment to the reality of social structures, conceived as relations between social agents in virtue of their occupancy of social positions. Structures are causally efficacious, in that they both enable actions which would otherwise not be possible (cashing cheques, getting degrees and so on), and constrain actions (bouncing cheques, imposing structural adjustment policies on Third World governments, enforcing essay deadlines and so on). To complement this account of social structure, Bhaskar develops what he calls his ‘transformational model of social action’. On this account, it is only through the activities of social agents that social structures are kept in being (reproduced), but individual or collective agency may also modify or transform social structures. The outcome of social action in either reproducing or transforming social structures may be unintended, as when employees go to work in order to earn a living, but in doing so also help reproduce capitalist relations of production without (usually) particularly wanting to do so. Or it may be intended, as in the case of successful social movement or political mobilization. It is central to this account of the relation between social structures and human agents that they are ontologically distinct from each other. This distinguishes Bhaskar's account (and that of most critical realists) from what Margaret Archer calls ‘elisionist' approaches (such as Anthony Giddens's ‘structuration') which collapse structure and agency together (see Archer 1995; Archer et al. 1998: ch. 14; Giddens in Bryant and Jarry 1997; see also Stones 1996: ch. 4; Craib 1992).

These arguments and the concepts they give rise to are very effective against both methodological individualist and empiricist tendencies to dismiss the reality of (unobservable) social structures.

Naturalism and Its Limits

In this context the term ‘naturalism’ generally means the view that there can be a scientific study of social life, in the same sense of science as in the natural sciences. Bhaskar is, as we have seen, committed to naturalism, but his account of social structure and agency implies some radical ontological and other differences between nature and society, with implications for the possibility of our knowledge of them. It is for this reason that his position is termed ‘critical naturalism’. So, first, what are the relevant differences, and how is it still possible to assert the possibility of a social science?

Bhaskar lists three ontological, one relational and one epistemological limit to naturalism. The ontological limits have to do with supposed differences between social and natural structures. Social structures are maintained in existence only through the activities of agents (activity dependence), whereas this is not true of structures in nature. Social structures are concept-dependent, in the sense that they are reproduced by actors in virtue of the beliefs actors have about what they are doing (but, as we saw above, the reproduction of the structures may not, and usually will not, form part of the pattern of beliefs which are the actor’s reasons for acting). Finally, social structures are only relatively enduring (are ‘space-time-dependent’), unlike structures in nature.

Though the terminology is different, most of these limits to naturalism are fairly familiar items in the anti-naturalistic arguments of the hermeneutic tradition (already encountered in Chapters 5 to 7). Both Collier (1994a: ch. 8) and Benton (1981) have argued that Bhaskar makes too strong a contrast between natural and social reality. Benton has argued, for example, that Bhaskar’s commitment to the activity­dependence of social structures comes close to undermining his own ontological distinction between structure and agency. Social scientific explanation often makes use of the notion of unexercised powers. Modern nation states, for example, have at their disposal an immense capacity for the use of violence to maintain order. This power is, however, rarely used - in part, at least, because dissident subjects know it is available. Benton also argues that space-time dependency of social structures is not peculiar to them. There are historical natural sciences such as geology and evolutionary biology, as well as developmental sciences which deal with often quite ephemeral but naturally occurring structures. The thesis of concept dependence, too, can be misleading. Much of social life is habitual and routine, involving bodily activity rather than conscious thought or symbolic meaning. Some important sociological knowledge - such as the well-established links between social class and occupation, on the one hand, and likelihood of premature death and chronic illness, on the other - points to causal mechanisms in society which operate independently of the conscious awareness of human agents.

Benton's view was that Bhaskar's strong contrast between social and natural ontology was based in part on his taking ‘basic' sciences such as physics and chemistry as the paradigms of natural science. Natural sciences such as meteorology, evolutionary biology and developmental biology share many features with the social sciences, and the social sciences themselves have very diverse subject-matters. It seems clear that a strongly anti-naturalistic social ontology will be a serious obstacle to the development of research programmes to address such questions as the relation between socio-economic processes and ecological change, where collaboration across the social/natural divide is essential (see Benton 1991).

However, for Bhaskar, his largely anti-naturalistic social ontology and related arguments can still be harnessed to a naturalistic defence of a scientific approach to social life. In some respects, this is because there are substitutes or compensations for the absence in the social world of features which enable the scientific study of nature. In some respects, he argues that social science is possible just because of the differences between the social and the natural.

The concept and activity dependence of social structures enable social scientific work, rather than limit it. This is because the beliefs actors have about their social life are available as a resource for social scientific thinking. This hermeneutic dimension of social life is, for Bhaskar, the necessary starting point for social science. However, the common­sense ideas of social actors are not treated as the final authority. Theoretical argument (including, notably, the use of transcendental arguments) and empirical evidence can lead to accounts of the social structures which differ from, or even contradict, those of the lay actors. The fact that social structures are only relatively enduring does not prevent their being real, and nor does it prevent their being objects of scientific investigation either during the time period, or within the spatial limits of their occurrence. That social science can take itself as its own object of study does not obliterate the transitive­intransitive dimension. Social science is only a part, and not the whole, of the subject­matter of the social sciences, and when it is being studied it is still possible to distinguish what is being studied from the process of studying it. It could even be argued that the self-referential character of sociology encourages a beneficial methodological reflexivity which is less evident in the natural sciences (this has been emphasized most strongly in feminist approaches to social scientific knowledge - see Chapter 9).

The epistemological limit, deriving from the necessary occurrence of social phenomena in open systems, and the consequent impossibility of experimental closure is a more tricky issue. As we saw, not all critical realists think it can be overcome. Bhaskar's response is to look for analogues or substitutes in social science for the role of experiments in the natural sciences. One analogue is the occurrence of crises in the social order, during which structures which are concealed in normal times become transparent. An example might be the 1984 miners' strike in the UK. The widespread acceptance of the police as a neutral force for maintaining public order was called into question by the use of extensive physical force by the police against the miners' pickets.

However, while this was certainly manifest to the miners, who were at the receiving end of police action, and their supporters, it was by no means a consensus. In general, social and political crises tend to polarize opposed interpretations of the social world, rather than settle differences of opinion.

Another alternative to experiment is the use of transcendental arguments. Everyday social practices under agreed descriptions can be analysed in terms of their conditions of possibility, and accounts of the underlying social structures built up in that way. Bhaskar considers it a plausible interpretation of Marx's Capital that it consists largely of such arguments from the experiences and understandings of those involved in economic activity to an account of the underlying structures and dynamics of capitalism (Bhaskar 1979, 1998: 65). However, this would seem to imply a very limited role for empirical research in the social sciences. Similarly, Collier's pessimism about the possibility of genuinely scientific social science is based on his view that measurement and statistical analysis are not a substitute for the impossibility of experiment. However, it is not clear why he believes this, and a good case can be made for non-positivist and critical realist uses of statistical analysis in the social sciences (see, for example, Levitas and Guy 1996).