Paths between Positivism and Interpretivism: An Appraisal of Hay's Via Media



Hay's Political Analysis raises foundational issues for all social scientists, not least in its outline for a via media, or middle way, between positivist and interpretivist social science. In this view, social science should be firmly grounded in empirical study but take seriously the notion that there is no privileged vantage point from which to generate dispassionate knowledge claims about the social world. This article asks whether this apparent via media is coherent and meaningfully captures what it means to be doing positivist and interpretivist social science without, so to speak, conceding too much ground to the other approach.

Colin Hay's Political Analysis hopes to make social research more conscious and explicit about the underlying assumptions upon which its choice of analytical strategies is premised (2002, p. 1). This choice requires necessary trade-offs that ought to be rendered explicit given the way in which, he claims, social research cannot achieve absolute objectivity. Indeed, although they often remain unacknowledged, implicit meta-theoretical positions and assumptions profoundly shape our approach to theory and method (Marsh and Furlong, 2002, p. 17). This article considers whether Hay's argument that such assumptions ought to be rendered explicit – based on a via media, or middle way, between positivist and interpretivist social science – maintains strength and coherence in the wider explanation and understanding debate. As such, while many critical contributions have focused on his epistemology, this article takes a broader perspective on Political Analysis's contribution to the philosophy of social science.

Clearly, Hay's intervention in this literature explicitly draws on a number of previous contributions dating back at least to Roy Bhaskar's foundational The Possibility of Naturalism and, as such, holds some affinities with critical realism more widely. In particular, his arguments about the political and contested character of social research, given the social scientist's inability to achieve disinterested objectivity, certainly mirror Bhaskar's distinction between the ‘intransitive’ nature of those objects of inquiry that exist and act independently of their identification and the ‘transitive’ nature of the knowledge that we might possess about these objects, which ultimately consists in historically specific social forms (Bhaskar, 1998, pp. 11–12). What is more, in its criticisms of positivism and interpretivism, Hay's position is similar to appraisals of positivism dating back to Weber and parallels other previous attempts to combine both approaches (for example Giddens, 1979). However, this is not to say that his intervention is lacking in consequence. In fact, given the ongoing task of developing a critical political analysis in a social science context in which the long-discovered limits of positivism are widely acknowledged but often not fully acted upon through fear of the spectre of radical post-modernism, Hay's reinvigoration of a via media approach is timely. From this perspective, such approaches ought to be welcomed because they put forward a necessary degree of epistemological caution with respect to knowledge while avoiding the full descent into self-defeating relativist scepticism (Potter and López, 2001, p. 9; Sayer, 2000, p. 102).

The philosophy of social science literature generally holds that ‘positivism’ places an emphasis on experience in general, and observation and testing in particular. A key example is the work of Emile Durkheim who aims to ‘extend the scope of scientific rationalism’ to human behaviour based on the belief that, in light of the past, it is capable of being reduced to relationships of cause and effect (Durkheim, 1982, p. 33). By contrast, ‘interpretivism’ appeals to ‘subjective meaning’, Verstehen and concepts such as empathy and interpretation. In the strong version, based on an anthropological or sceptical relativism, it holds that social differentiation gives rise to distinct ‘ways of knowing’ between which there is no way of according common standards (Williams, 2001, p. 10). In all, this article will attempt to show that Hay's via media approach successfully navigates a path between these positivist and interpretivist approaches. To this end, we must evaluate his claims about the subject-matter of the social sciences, the limits of positivism and interpretivism and, finally, the via media approach itself.

The subject-matter of the social sciences

A key theme of Political Analysis is the qualitative difference in subject-matter between the natural and the social sciences. Following Bhaskar, Hay maintains that social structures, unlike natural structures: (1) do not exist independently of the activities they govern; (2) do not exist independently of an agent's conception of their activity; and (3) may only be relatively enduring (2002, p. 85).

Taking each in turn, distinction (1) holds that active, conscious and reflexive subjects inhabit the social world whereas the units of analysis in the natural sciences can be assumed to be inanimate and un-reflexive. This brings into question the notion of a predictive science of the social world and thus explanations based on the probabilistic symmetry between explanation and prediction. If social systems are ‘intrinsically open’, decisive test situations for our theories cannot be created, which means there can be no rational confirmation or rejection of predictions (Bhaskar, 1989, p. 5). This argument can be criticised for overemphasising the difference between natural and social structures based on human agency because it rests on the idea that ‘nothing happens in society save in or in virtue of something human beings do or have done’ (Bhaskar, cited in Collier, 1994, p. 246). Arguably, this statement could be reformulated by substituting the relevant kinds of entity for ‘society’ and ‘human beings’ in each case, which would allow similar principles to hold in the natural sciences (Collier, 1994, p. 246). Nevertheless, for many there is an intuitive accuracy to the idea that social systems are intrinsically open due to the reflexive nature of social action. Steven Bernstein et al. argue that human intervention in society is striking because the more individuals think they understand the environment in which they operate, the more they attempt to manipulate it to their advantage (2000, p. 51). In this sense, the ‘laws’ of social science are open to the environment to which they refer (Giddens, 1979, p. 258). Moreover, even when social scientists attempt to emulate experimental conditions, the research is hindered because a subject's behaviour is potentially influenced by learning about its hypotheses and methods (Rosenberg, 1988, p. 174). Therefore, distinction (1) is useful.

Building on the former, distinction (2) holds that once propositions or ‘laws’ of social behaviour enter public discourse they may lead actors to modify their behaviour. In other words, the ideas we hold about the social world as theorists, commentators or social subjects can alter its composition (Hay, 2002, pp. 79, 86). Contrastingly, natural scientists do not have to deal with intentionality and reflexivity because their subject-matter does not modify itself in the light of claims made about it. What is more, many propositions in the social sciences are formulated with the explicit hope that the regularities on which they are based may eventually be transformed or disrupted; again in contrast to natural scientific inquiry which is thought to forswear evaluative judgement. Now, it is possible to accept this line of argument without suggesting that social theorists or actors ultimately ‘construct’ reality. Instead, more plausibly, social theory and reality are ‘causally interdependent’– theory is both conditioned by society as well as having consequences in it (Bhaskar, 1989, p. 5). Crucially, this entails that social scientists face questions about their own position with regard to research, a view typically criticised by those more inclined to naturalism (Trigg, 2001, p. 252). Indeed, Durkheim argues that we fail to penetrate the ‘inmost essence’ of phenomena if the social scientist's preconceptions dominate the research (1982, p. 380). However, although certainly problematic for those wishing to produce entirely value-free analysis, distinction (2) cannot be disregarded on these grounds alone because it may merely require a different form of ‘science’ for the social. In all, this distinction remains robust because social theory has the potential to alter its own object of analysis in a way that has ‘no parallel in the natural sciences’ (Collier, 1994, p. 247).

Finally, distinction (3) maintains that social processes tend to be culturally, spatially and historically specific in contrast to universal natural laws. Hay compares the study of the global political economy and physics to highlight this difference (2002, p. 86). In the former, the use of simplifying assumptions to generate testable propositions is made extremely problematic by incessant change. In the latter, by contrast, the generalised laws of physics can be assumed to pertain in all situations across time. However, in one theoretical sense, social structures can be held to be space-time invariant as when certain conditions are met, for example economies with certain features, particular tendencies will operate (Collier, 1994, p. 244). In this way, we can formulate social laws in terms which are ‘universal’ by virtue of being conditional. Yet, in a more compelling sense, when considered in historical perspective social structures do appear to be only relatively enduring. Indeed, a devastating problem for covering-law explanations of social phenomena, which attempt to identify generalised laws based on observable regularities, is that they prove either impossible or trivial because we are forced to add so many subsidiary details to the ‘initial conditions’ that in some cases only the name of the event or process is missing from the given ‘explanation’ (Outhwaite, 1996, p. 94). Thus, again, distinction (3) is appropriate.

Overall, by highlighting the basic divisions between the subject-matter of the natural and the social, Hay is successful in laying the foundation for the call to recognise the essential differences in their methods while still allowing both to hold some claim to knowledge production. We must now consider how social scientists engage with their unique subject-matter.

Positivism, interpretivism and the via media

According to Hay, and building on these three distinctions above, social scientists find it extremely difficult to generate dispassionate and objective knowledge claims for two key reasons (2002, pp. 86–87). First, based on (1) and (2), is the inevitable location of the social scientist within that which forms their subject-matter. From this embedded position, the social scientist cannot simply escape their complex and densely structured environment in order to carry out dispassionate study of the social world. Second, based on (2) and (3), there are ethical dilemmas associated with the social scientist's privileged position owing to their potential to shape the social environment. Again, from an embedded position, the social scientist may come to redefine and alter that which is socially possible. In reply, we can explore three distinct approaches, each having their own response to these ethical considerations associated with the problem of objectivity: positivism, interpretivism and the via media.

Although most positivist approaches would not require all theoretical terms and assumptions to refer directly to observables, there is a commitment to the notion that substantive hypotheses must be able to be falsified using an appeal to empirical evidence. For Hay, positivism is thus simply unable to confront the ethical responsibility he identifies as inherent to social science. This is because to achieve ‘epistemic security and conceptual clarity’ it tightly demarcates what can count as knowledge – a statement is not meaningful unless it is possible to explain what would count as falsifying it (Williams, 2001, p. 91). Rather than a damaging criticism, at this point Hay's critique constitutes a description of the aims of positivism: to delineate precisely what status we can give to our knowledge claims is a key strength of the approach, even if it means we have to exclude certain types of knowledge.

However, arguably this exclusion is unacceptable because positivism can only superficially eschew ethical responsibility and normative judgement. Unlike the natural sciences, the social sciences do not enjoy consensus on what the questions are that each has the power to address or agreement about the methods to employ. As such, these choices rely on unverifiable starting assumptions (Rosenberg, 1988, pp. 4–5). Problematically, if the scientific method is prioritised at the expense of normative judgement, then researchers may only choose problems to investigate because they are thought tractable, not because they are of importance (Bernstein et al., 2000, p. 44). To overcome this problem, those who subscribe to a positivist approach require extraordinary abilities: Durkheim claims researchers must become ‘free’ by removing those ‘fallacious notions which hold sway over the mind of the ordinary person’ (1982, p. 73). In truth, positivism's appeal to objectivity is undermined by the very act of selecting a problem on which to work because this is to attach meaning and significance to expected explanations. If positivists are forced to concede that social scientists have the capacity to influence the course of social change through these choices, they may have to abandon any pretensions for a science of the social that is entirely objective.1 We must now consider the interpretive answer to this problem of objectivity.

Although it is problematic to identify, ‘interpretivism’ can be regarded as a form of social analysis which represents profound scepticism towards claims to objectivity and a privileged access to knowledge. Thus, in these terms, it would appear that interpretivism suitably accounts for Hay's problem of objectivity. To take seriously the challenge presented by post-modernist critics to social science is to acknowledge the value-laden and normative content of many of its assumptions and much of its language (Rosenberg, 1988, pp. 187–188). However, for Hay, interpretivism cannot accurately be thought of as making a substantive contribution to social analysis (2002, p. 217). Its ontological conviction that the world can be viewed from a multiplicity of perspectives, along with its normative commitment to according them equal respect, means that interpretivism is drawn inevitably towards a suspicion of all epistemological foundations. This degree of relativism is just as harmful as positivism to the social scientist's ethical responsibilities. The interpretivist approach does not restore the role of normative inquiry because it rejects the claim that there are ultimate metaphysical truths beyond appearances. In short, positivism deliberately confuses reality with knowledge and interpretivism reacts by removing the possibility of knowledge (Trigg, 2001, p. 250).

Thus, we are in search of an alternative to positivism's ‘blindness to ethical considerations’ and interpretivism's ‘nihilism and fatalism’ (Hay, 2002, pp. 87–88). In response, the via media takes seriously the ethical responsibilities that come with an acknowledgement that epistemology cannot adjudicate social knowledge claims while still accepting the possibility of making them. For Hay, this means social scientists must acknowledge the necessarily normative content of their work and render their normative assumptions as explicit as possible. We should not dismiss empirical study altogether because it provides the grounding for our descriptive analysis, but as soon as we move from this to explanation we move from the realm of science to that of interpretation. In this latter realm there is conflict between competing narratives premised on different meta-theoretical assumptions, which necessitates that we make our ‘normative assumptions explicit’ (Hay, 2002, pp. 87–88). In this way, a via media is developed that appeals to a form of qualified empirical study that acknowledges the interpretivist critique. The question remains as to whether this via media constitutes a coherent approach.

Explanation and understanding?

Alexander Rosenberg claims that with many philosophy of social science questions there is rarely a ‘happy medium that splits the difference’ because accounts are often logically incompatible and attempts made to combine parts of each usually result in incoherence rather than genuine compromise (1988, p. 18). At the very least, theories that purport to reconcile the tension between positivism and interpretivism – or ‘explanation’ and ‘understanding’– tend to be ‘fragile’ because they are ‘always telling two stories’ (Hollis and Smith, 1990, pp. 5–7). So, does Hay's via media provide a successful combination?

According to Martin Smith, Hay is plotting a careful path between explanation and understanding comparable to Odysseus navigating between Scylla and Charybdis (2006, p. 285). He makes two interesting claims about Hay's analysis: first, that Hay wholly rejects ‘attempts by positivists to develop predictive models’ due to the ‘inherent uncertainty of social life’ and says nothing more about his approach to positivism. Second, although Hay also ‘questions the ultimate nihilism of postmodernism’, he still ‘appreciates the way it sensitizes us to the need to question assumptions’. Here, arguably, Smith's interpretation of Hay presents him as developing an outright rejection of positivism, whereas interpretivism, although flawed, has more value for social science. Perhaps Hay is sailing too close to Charybdis, understanding, rather than navigating a successful middle course.

Yet equally, one could argue, Hay prioritises explanation. He claims that whereas social scientists are likely to remain divided over the interpretation of particular events or processes, there is greater chance of reaching agreement on their description through the accumulation of empirical evidence (2002, p. 252). This echoes ‘post-positivist’ positions which accept interpretation and meaning as important, but tend to treat them as intervening variables without fully acknowledging the subjectivity of the observer (Marsh and Furlong, 2002, pp. 25–26). The emphasis still lies with explanation rather than understanding; experience rather than meaning. This is problematic because an appeal to pure experience is ambiguous: ‘experience’ can refer to both ‘what is presented to us’ and the actual act of ‘experiencing’, which is something Hay would surely acknowledge given his appeal to the interpretivist critique (Hollis, 2002, p. 71).

Thus, Hay is open to criticism from both sides of the debate for being too close to the other. It is difficult to find a way out of this tension. Interestingly – even though they are co-authoring the same book – Martin Hollis opts for understanding as an approach to social scientific inquiry whereas Steve Smith chooses explanation (Hollis and Smith, 1990, pp. 14–15). Smith sees actors' understandings as conditioned by factors external to them; Hollis sees them as the very constituent parts of the world which they seek to understand. For them, this tension is irresolvable, leading them to call for a plurality of acceptable positions. In the final analysis, they are forced to concede that the area that represents a middle ground between explanation and understanding should be thought of as a ‘movable counter’ that can be repositioned to ‘whatever place on the chart the reader finds most satisfactory’ (Hollis and Smith, 1990, p. 214). Likewise, David Marsh and Paul Furlong accept the contested nature of epistemological positions and welcome a plurality of voices (2002, p. 20). After a call for plurality, it would appear that the foundational assumptions on which this choice between explanation and understanding is made ought to be rendered explicit, as Hay argues, if readers are left to ‘make up their own minds’. Thus, at least the via media is convincing.

In this spirit, perhaps it is telling to search in Hay's substantive work for indications as to whether the via media is coherent. The analysis found in Demystifying Globalisation provides a useful example of how a combination of explanation and understanding can inform our knowledge of social phenomena (Hay and Marsh, 2000). Here, ‘real’ (empirically verifiable) processes are identified, such as the spread of financial markets and increased levels of trading, which constitute observable patterned forms of interaction. Nevertheless, simultaneously, care is taken to acknowledge that the way in which these processes affect outcomes is mediated by the very discursive construction of these processes. In other words, the ideas commonly held about these processes actually give the patterned forms of interaction greater causal effectiveness. Indeed, British governments have in the past argued that the pursuit of neoliberal policies is inevitable in light of (their beliefs about) globalisation, even though there is only limited empirical evidence that a globalised political economy must necessarily determine economic policy. As such, causal mechanisms can be revealed empirically (explanation) but only fully accounted for through recognition of their social construction (understanding). At once ‘there is an appeal to the real world, but the emphasis is on the discursive construction of that world’ (Marsh and Furlong, 2002, p. 35). Thus, the via media approach effectively illustrates how dominant ideas about globalisation are causally efficacious, in the sense of making a real difference to social outcomes through the shaping of economic policy, but it also acknowledges that the ‘real’ processes of globalisation constrain the resonance of different discourses.

Other areas of social study that might benefit from the via media approach include, for example, the analysis of global poverty in the international political economy literature. Indeed, Branwen Gruffydd Jones demonstrates that although orthodox approaches to global poverty provide accurate explanations at the descriptive level of surface appearances – for example, that the global poor are unable to satisfy their basic needs because they lack access to income-earning opportunities – they remain ‘blind to the real, non-empirical relations’ that generate these empirical appearances, such as the social relations that govern the activities of producing the objects of basic human needs (2003, p. 228). As such, a via media approach to the analysis of global poverty might acknowledge the importance of observable characteristics, inasmuch as they are revealed as genuine patterned forms of interaction, but might further recognise that ideas about these characteristics themselves have causal outcomes, including the reproduction of the social relations that help bring them about.

Although only briefly sketched here, these examples suggest that Hay's via media can overcome the idea that explanation and understanding are ‘mutually incompatible’, in Rosenberg's terms (1988, p. 205). Overall, it is perhaps most appropriate to choose a unique position with regard to explanation and understanding while allowing others to identify theirs – the call for an acceptance of a plurality of epistemological positions given above. Yet, in accepting the possibility that we are wrong, we are not resigning the hope of rationally justifying our choices. For in the process of making reasoned judgements about empirical evidence, the social scientist is appealing to the ‘public character’ that has served experimentation so well in the natural sciences. Such judgements are based on the choices of the social scientist, which is why, to make the via media coherent, Hay is right to argue that normative and meta-theoretical assumptions ought to be made explicit.


Hay's via media approach remains a robust and coherent position to take with regard to the philosophy of social science for a number of reasons. First, the central tenet on which his argument is built – the distinction between the subject-matters of the natural and the social sciences – is useful and accurate. Second, given that this subject-matter brings certain ethical responsibilities to bear on the social scientist, we can accept the limitations of both the positivist and the interpretivist approach, and the call for a via media that makes normative and meta-theoretical assumptions explicit. Finally, we have some grounds on which to argue that Hay's approach remains coherent in the wider explanation and understanding debate. Rather than assuming mutual incompatibility, by recognising that the debate requires a trade-off that is ultimately the reasoned choice of the social scientist and, crucially, also of the audience which has to accept it, we can use the via media approach to take seriously both positivist explanation and interpretivist understanding.


  • I would like to thank Edward Page, Matthew Watson and two anonymous referees for POLITICS for their helpful comments on an earlier version of this article. This article was written with the assistance of an Economic and Social Research Council ‘1 + 3’ Studentship. I gratefully acknowledge the ESRC's support of my research.

  • 1

    This holds true, as Weber pointed out, even when social scientists attempt to be objective in the application of descriptive procedures (value-freedom) because the actual objects of their study are connected to prevailing societal values in a number of ways (value-relevance).