Unlike the natural/physical sciences, the social sciences show no settling of method. By this I mean that within the social sciences, scientists work from multiple and fundamentally disparate approaches to research and enquiry. This varies somewhat across disciplines; anthropologists almost entirely employ ethnography while economists almost all use formal models, though there are differences even within these fields as to what the ontological and epistemic status of data and theory actually is. But on the whole, there doesn’t seem to be any single way to define what social science is and how it is distinguishable from other sciences. Rather, there are several dominant views as to what it means to commit a social science, and there is no obvious way to select one as superior to the others in any categorical sense.
In this post I will provide a bit of an introduction to these different approaches. I’ll go over some of the main philosophical problems that confront philosophers of social science, and trace how different solutions to those problems have led to different social sciences (in the methodological rather than the disciplinary sense). I’ll offer a few critiques of each of those approaches, and finally I’ll situate my own views and preferences, providing a brief defence for them.
What does it mean to commit a social science?
There are three key questions that social scientists must, at least implicitly, answer in order to get on with the business of doing social science.
- What is the domain of social science? In other words, ‘what is the social world?’ This is a deceptively complex question in that multiple answers seem plausible. Is social science about theorising how humans behave around other humans? Is it about studying meaning – about situating people within ‘webs of significance that [they themselves] have spun‘? Is it about studying all action? Is the domain of science continuous with the domain of language? Is it the study of rules? Or is there no rigorous ontological way to demarcate the social from the non-social, such that the domain of social science is really no more than whatever social scientists want it to be?
- What is an explanation of social things? Again, there are multiple plausible answers to this. Some answers should be familiar to anyone who has read even a little into the philosophy of science? Is a good explanation one which subsumes events under general or predictive laws? Or is it one that specifies the entities of an objective reality and describes their causal interactions? Or is it that, in social science, a good explanation is simply one that provides us with the reasons that motivated people to have done what they did?
- What counts as evidence in the social sciences? Simply put, what does it mean to observe or document the social world and how do these data allow us to build whatever it is that an explanation of social things should be? This is not just a question of methodology, but requires us to take a stand on the relationship between observer and observed, and on the status of knowledge claims in the sciences. Depending on our answer to the preceding two questions, we might decide that data are limited to recordings of events, or we might say that descriptions of real structures count, or perhaps only that reports about motives or beliefs tell us something about anything.
There are huge differences in the methodological implications that follow from how we answer these; that is, social science research will proceed in very different ways, based on how the researcher solves these philosophical puzzles. And these questions remain contested within the social sciences. Of course, philosophers of science still contest similar kinds of questions when it comes to the natural sciences, but actual practicing physicists, chemists, biologists, and the like rarely need to wonder about them. For whatever reason (more on that later), natural scientists have been able to avoid major fractures on these matters, and their disciplines tend to be more or less unified when it comes to matters of basic methodology.
Though ‘positivism’ as a mode of enquiry in the social sciences dates, at least nominally, back to Auguste Comte, what typically receives the label today in the social sciences is significantly different from what Comte proposed. In discussing its roots, it makes more sense to relate it back to the tradition of logical positivism and to critical responses to that tradition made by Popper and Hempel. For positivist social scientists – who comprise the bulk of scholars in political science and sociology – the social world is constituted by the behaviours characteristic of interpersonal interaction. In other words, positivists are studying (a) people engaging in actions that (b) engage with other people and with things like institutions or ideologies. If this seems like a vague or under-specified definition of what social is, I suggest setting aside this concern. Positivists typically have. Positivist explanations typically seek to subsume social behaviour under general, predictive, though typically probablistic laws that specify how a change in one variable ’causes’ a change in another variable, ideally though not always through hypothetico-deduction. True to its Humean roots, evidence for positivists consists of recordings of events, sorted into types and coded to indicate occurrence or non-occurrence at a given value. And, true to their roots in logical positivism, positivist scholars in the social sciences have advocated for a unity of method that entails a similar ‘logic of enquiry’ for both social and natural scientists.
Though positivist social science for a while truly embraced the search for laws of behaviour, and did not aspire to anything more (on the grounds that such aspirations were doomed), it has since expanded in horizon to incorporate the search for causal mechanisms, typically defined as ‘intervening variables’. It has always quite frequently referred to the behaviour of entities such as institutions, states, or firms, rather than to only that of individuals. These broader horizons, though they reflect the practical necessity of referring to collectives and to processes when talking about social reality, bring with them certain philosophical tensions. To refer to causal mechanisms as ‘intervening variables’ appears to dissolve the ontological implications of a mechanistic theory of causality; it turns talk about the role that certain kinds of interactions play in bridging cause and effect into observations of nothing more than finer-grained sequences of events. This is in keeping with Humean empiricism, but it doesn’t seem to preserve what makes mechanistic theorising helpful in the first place – namely, that it allows us to step beyond ‘sense events’. Meanwhile, important questions about the relationship between individuals and the larger social entities in which those individuals live – on the relationship between structure and agency – become answerable only in terms of behavioural patterns, effectively shoving into a ‘black box’ questions of meaning, subjectivity, and normativity that are quite central to other modes of social enquiry.
Hermeneutics and Interpretivism
From the mid-19th century onwards, philosophers have sought to bifurcate the natural sciences and the social sciences according to a fundamental difference in method. While the natural sciences involve the search for the laws of nature, the ‘sciences of man’ (geisteswissenschaften as those in this almost-exclusively-German tradition called them, taking after Dilthey, who coined the term) involve uncovering the reasons according to which actions are rational. The bifurcation of ‘explanation’ and ‘understanding’ defines this methodological difference; the social sciences, according to this tradition, require us to interpret social life and reconstruct the cultural or normative backdrop of a given set of events. In other words, the social world is one of meaning-laden actions taken by subjects moved by moral and instrumental reasons, and an explanation of those actions is an interpretation of them that shows why they made sense.
While this approach is historical, in that it involves rooting explanations in particular points of social time and space, it still leaves plenty of room for theorising. However, theory in this tradition is very different from theory in the positivist tradition; finding laws are not the goal (nor even held to be possible), but taxonomy and ideal-typification are common. The prototypical examples of theory in this tradition would be Weber’s taxonomy of authority and his attempt to explain the apparent affinity between capitalism and certain Christian communities in his book The Protestant Ethic and the Spirit of Capitalism. But in the past several decades, the interpretive tradition has taken a sharp post-modern turn, in which social scientists are less interested in trying to explain big social processes spanning tens or even hundreds of years and more interested in limiting their investigations to local and often overlooked discourses or forms of life , without trying to abstract from them. In other words, the interpretive tradition has taken on a distinctly anthropological sensitivity. This is not inconsistent with the roots of the hermeneutic or interpretivist tradition, for a consistent thread throughout is to treat theory more as a means of organising our data and less as means of uncovering deeper relationships, but the critical and post-modern shade of much scholarship entails a lack of interest in truth or objectivity that many scholars find off-putting. It is also difficult to talk of anything other than discourses and meanings in this tradition, which makes theorising social structures related to class-relations or institutional orderings like states or bureaucracies difficult if not impossible.
Structural Realisms: Historical Materialism, Structural Functionalism and Critical Realism
A third stream of social science, which sometimes gets conflated with positivism but which really, really isn’t, is Critical Realism and its antecedent in the social sciences, historical materialism. Historical materialism is the Marxist approach to theorising society, and classically involves explaining explaining all social phenomena as the product of underlying economic structures – of ‘basic’ relations of exploitation and exchange – and explaining variation across social time and space in terms of differences in the broader form (or ‘mode’) that those underlying structures take. This traditionally takes the form of mainly economic analyses of particular historical cases; knowledge is produced not through empirical cross-case comparison but through explicating the causal processes and relationships specific to those cases. In some ways similar to historical materialism, though not coming from the Marxist tradition, is structural functionalism. Structural functionalists hold that society is made up of interlocking sets of self-regulating systems, and thus that actions can be understood according to their role (or function) in maintaining the stability of those systems. Thus like historical materialism, structural functionalism involves explaining specific cases by making reference to a delimited set of social structures which determine who people are, what their interests are, and why they do what they do. The ‘economic determinism’ of historical materialism has fallen out of favour, and many aspects of the Marxist approach to thinking about the social world have been taken up by postmodern discourse theorists, while structural functionalism has fallen out of favour for all sorts of reasons, largely being replaced by historically-minded interpretivists or by rational choice theorists. But there remains an interest in ‘real structures’ and their properties. This interest has, over the past three decades, been increasingly taken up by Critical Realists, working from the philosophical foundations set forth by Roy Bhaskar.
Critical Realists retain the historical materialist view that society is constituted by structured relations between actors whose roles are defined by those relations, but drops the economic determinism; now structures of ideas can be just as causally important, and the structures salient to explaining society can comprise a nexus of interacting ideas, institutions, even physiological attributes. For example, gender can be as important as class. What really distinguishes Critical Realism here is the addition of a dispositional theory of causality to that structural ontology. Critical Realists hold that structures have ‘causal powers’ – the inherent disposition to generate a discrete set of effects under the right conditions. Explanation thus consists of the identification of the structures, causal powers, and conditions that generate social events – to go beyond correlations of outcomes and to find the underlying social objects and relationships that make society what it is. Thus while it is historical like the Weberian or hermeneutical tradition, it is not about trying to understand how situated social actors make sense of their worlds, but about describing the world as it really is, with theories serving not merely as ideal types but as possibly correct representations of objective reality.
While Critical Realism is considerably more sensitive to the power of ideas than historical materialism, it depends heavily upon highly contentious ontological and epistemological presuppositions, including strong emergence and the possibility of isomorphically representing the social world through our theories. While a Critical Realist philosophical framework makes it possible to use the language of causation and constitution to discuss a wide range of social phenomena, it ceases to be much more than a very complex set of ideal types if it can’t be realist. At this point, it becomes nothing more than a curiously blithe form of interpretivism, in which relatively little explanatory role is given to the uncovered reasons that people had for doing what they did, and most attention is paid to arranging the data into coherent structural patterns – in other words, hermeneutics that ignore the very thing that would make them explanatorily adequate, out of the vain desire to transcend the perspectives of individual subjects. Suffice it to say, this would also deprive Critical Realism of much of what makes it ‘critical’ to begin with.
Some Final Words, and My Take
I broadly locate myself in the hermeutical or interpretivist tradition, but I find a great deal of value in Critical Realism as well. Ultimately, I find many of the criticisms of realist philosophy of social science to be compelling: I do not think that it is possible to describe some ‘really real reality’ through the use of language, and even if there is enough stability to the natural world for some sort of quasi-realism, I do not thing that the social world exhibits that level of stability. I join with the American Pragmatists in viewing the entire project of theorising as an attempt to make sense of the world, and thus I ultimately see theories as ideal-types. That said, it is common for hermeneutic and interpretivist approaches to focus so much on individual minds or, alternatively, on mind-less discourses containing no agents but only symbols, and these approaches make it hard to think about the big picture in social life. It may be that the idea of a social structure, like class or gender, is nothing more than a convenient way for us to organise our experiences of the world as we live through it, but it is a very helpful idea nonetheless. It is a challenge to link the concept of a structure, which is basically ‘objective’ in character and which sits at a ‘macro-level’ position in our social ontologies, and the phenomenal bedrock of life and action that underpins the hermeneutical approach.  But that challenge need not prevent us from maintaining our broadly hermeuticist philosophical foundations and still building theories that make reference to macro- or meso-level social things like structures, processes, causal mechanisms, and the like. It merely means that we must ultimately recognise that our theories about such things are about solving conceptual problems arising out of our need to understand, and not about painting a picture of the social world as it really is.
Finally, one might notice that I’ve devoted little time to my position on positivism. This is ultimately because I find positivism an uninteresting and unhelpful form of theorising. Its fundamental mode is to locate conjunctions of events, but this provides me with no satisfying account of causal relationships or of actors’ reasons. In other words, it provides neither explanation nor understanding.
 This is not to say that the natural sciences don’t feature many significant methodological debates, nor that natural scientists never consider questions of philosophical relevance. Rather, I mean that the basic questions about being and knowledge that must be given at least implicit answers in order for a science to proceed don’t really come into the picture like they do in the social science.
 It is important to note that Critical Realism is not supposed to be specific to the social sciences, even if philosophers of the natural sciences have not given it much attention.
 This would be contentious for a few reasons. First, many physicalists – and Critical Realism is a kind of physicalism – would deny that mental states could be anything other or more than token ‘brain’ states, meaning that agency, which is an essential feature of the social world, is a sort of ‘illusion’ resulting from the vast complexity of the mind. Were this true, Critical Realist theories would be guilty of reducing actors to ‘structural dopes‘, and therefore would suffer from the same undesirable determinism as their historical materialist and structural functionalist predecessors. Second, all of the criticisms that philosophers have made of the notion that we can ‘step outside of language’ or in any way represent an ‘external reality’ through language apply to Critical Realism in a way that they do not to hermeneutical/interpretivist or even to positivist alternatives.
 A challenge I hope to take up in some of my own work.