Modernism, politics and science, and the nation-state

Political science is the product of the modern era, specifically the nineteenth and twentieth centuries, and these account for the affinity with the natural sciences, with their laws and regularities. This conception is challenged periodically by those who locate it among the softer disciplines of the humanities (Grofman, Reference Grofman2007). The positivist mainstream insist that political science should (a) be paradigmatic, with a unified ontology, epistemology, and methodology; (b) adhere to methods and standards akin to those of the natural sciences; (c) be cumulative. This is unambiguously articulated in King et al. (Reference King, Keohane and Verba1994),Footnote 1 who indicate that research should aim at descriptive knowledge of the real world, or at causal inference. The ontology and epistemology are realist and positivistic, in that the social world exists and we can have direct knowledge of it. The world is characterized by regularities and rules such that we can establish causal relations between events. Values can and should be strictly distinguished from facts. Theories should be generated deductively but then tested empirically. The basic unit of analysis is the individual. Within this overall approach, there is room for methodological pluralism. Case studies, small n. comparisons, historical approaches all have their place, but subject to the basic logic of seeking universal explanations based on causal inference (Brady and Collier, Reference Brady and Collier2004).

Another modernist legacy is the search for universalism, reinforced during the behavioural revolution of the 1940s and 1950s. The basic tenet is that human nature is essentially the same everywhere and that the same causes will produce the same effects in any location. While this is essentially a scientific tenet, it was underpinned by a normative liberalism and a desire to combat older stereotypes about ‘national characters’ or racial and ethnic differences.

Less often noticed was another modernist legacy, in the form of the nation-state. It is the state that largely defines the subject of political science, yet the concept was rarely subject to critical analysis. It has two quite distinct meanings. In one sense it means an order in which the state is aligned with an underlying nation. The other sense, in both comparative politics and international relations, refers to the sovereign state as the basis of domestic and international order. The only link between the two is provided by the doctrine of nationalism, which holds that the sovereign state should be congruent with the nation. Yet nationalism was neglected as a field of study after the Second World War and few people seemed to notice the discrepancy. In so far as it was analysed, the nation-state was widely seen as the path not only to modernity but also to universalism, as polities integrated and overcame internal differences, first in the West and then in the rest of the world.

These trends were particularly pronounced in the United States, often seen as the harbinger of modernity. In the study of American domestic politics, where the interest-group pluralist perspective prevailed, the state tended to be defined away or disaggregated into a set of agencies interacting with groups and each other. This contrasts with continental European or even British traditions, where the state has a being and role distinct from society. In the study of international relations, by contrast, American scholars tended to give states a supreme importance, treating them, in the realist perspective, as unitary actors. Yet, different though they seem in their ontological assumptions, these internal and external perspectives on the state led in the same direction, to a view of politics as the interaction of self-interested and rational actors, whether these be individual citizens or states. Culture and tradition were important in neither case, and motivation was usually assumed to be unproblematic.

Political science now has to face the loss of the certainties that were scientific realism, the modernization paradigm, and the nation-state. The affinities with the natural sciences have been difficult to sustain as the natural sciences have moved on. More sophisticated social scientists incorporate post-Einsteinian physics to account for indeterminacy, but this merely reinforces the affinity to the natural sciences. Others have looked to the biological sciences, emphasizing evolution, which retains the idea that behaviour can be explained but abandons the hope of prediction. This might be criticized as more of the same thing, as political science latches onto whatever the leading-edge natural science is at any given time. Indeed, social scientists have dabbled with evolution before, in the form of social Darwinism.

Classical modernization theory, with its optimistic predictions about convergence and integration, enjoyed a brief revival at the end of the Cold War with the ‘end of history’ (Fukuyama, Reference Fukuyama1992) debate, but soon succumbed to the evidence that the world was a diverse place, that social and political evolution can follow different tracks, and that individual actors are socially embedded. Normative universalism has been questioned as the imposition of western values on other societies and the progressive side of politics is more likely to champion diversity, while it is often the political right that insists on universality.

Challenges have also emerged to both meanings of the nation-state. The state has experienced a loss of ontological distinctiveness as it is penetrated and limited from above, below, and laterally. Spatial rescaling means that functional systems no longer correspond to boundaries of states (to the extent that they ever did). Boundaries between the state and market and between state and civil society are shifting. The autonomy of governments is curtailed. The easy linking of state and nation is challenged by new nationalisms, and the legitimacy of the state order called into question (Keating, Reference Keating2001). These changes pose a challenge to all the social sciences but particularly to political science, since they call into question the distinctiveness of the political, its focus on the state and what it does, and previously unquestioned assumptions about legitimacy.

The erosion of the certainties underlying positivist political science has encouraged scholars from alternative perspectives, by definition more difficult to label since they dispute the very certainties on which mainstream political science relies, but who generally argue the following. The nature of the world is problematic and it is unobservable directly; thus, we are dependent on our own interpretations. Human behaviour is not caused in the sense that physical events are, but is unpredictable, since human beings exercise free will, and have the ability to reflect on themselves and their relationship with the world. Normativity is built into the very concepts that we use so that the fact–value distinction disintegrates. Approaches in the social sciences are incommensurate; thus we can never prove that one theory is right and conclude that its rivals are therefore wrong. Modernity is just one model of society, containing its own hidden normativity and the state is not a neutral arena for group competition. Work in this tradition emphasizes contingency, reflexivity, and interpretation in a double sense, in that actors interpret the world and social scientists interpret their interpretations.Footnote 2 Exponents of these approaches may call themselves constructivists, emphasizing the way in which our understanding of the world is built on concepts not realities; or intepretativists, who emphasize that our knowledge of the world is built from interpretation and that we should seek understanding rather than causal explanation. More radical approaches, which might be brought under the broad umbrella of post-modernism, question the scientific basis of knowledge, universalism, and the notion of progress.

There are five distinct issues, or levels, in this argument. First is ontology, or what we know and how we conceptualize it. Second is epistemology, or how we know it. Third is methodology, or how we approach the study of it. Fourth is methods, which are merely ways of gathering data. Fifth is theory, or our working assumptions and frames of analysis about human behaviour. There is no necessary one-to-one correspondence among these levels. Realist ontology takes the existence of the world as axiomatic but does not entail the epistemological proposition that we can know it directly. Neither of these entails the positivist conception of the world as ordered. All of these, in turn, are independent of theoretical propositions about causes and linkages. Methods for their part are merely ways of acquiring data and are independent again. These are all connected only through methodology, which takes a research question and a theory, and seeks to convert it into something researchable.

However, in much of the debate, options are reduced to a dualist choice between ‘quantitative’ and ‘qualitative’ ‘methods’, in which questions of ontology, epistemology, methodology, and methods are all subsumed. The vocabulary could hardly be more misleading, since the ‘qualitativists’ often use statistical techniques to measure texts, while the quantitativists accept interview material and case studies, as long as they are used to sustain their central positivist-causal model. The term ‘qualitative’ itself is used in two quite different senses. One has to do with ontology and epistemology, and equates with interpretivist or constructivist approaches. The other refers to methods in the narrowest sense (as the gathering of data) and is used by positivists to embrace the use of non-quantitative analysis as a source of data about a social world that is given a real existence. The Manichean division between ‘quantitative’ and ‘qualitative’ approaches is, however, self-sustaining intellectually and politically. The resulting schools (or factions) do not always coincide with the real choices to be made in political research. Rather, like political parties, the schools include a diverse constituency of support, sustained by common identity and community. Indeed, they include some rather illogical matchings.

For example, rational choice theorists tend to make common cause with ontological realists and epistemological positivists, despite the fact that the former depend on a stylized representation of the world based on a conceptual scheme that has no necessary relationship to the world revealed by observation and their arguments. Based as they usually are on restricted assumptions, they are typically impossible to falsify empirically. Yet, Marsh and Stoker (Reference Marsh and Stoker1995: 290) write uncritically that ‘…within the discipline there are authors utilising perspectives as diverse as rational choice theory and discourse analysis. The former operates from a positivist epistemological position and emphasises quantitative analysis; the latter operates from a relativist epistemological position and concentrates on qualitative analysis.’

Another anomaly arises in rational choice approaches, where analysts emphasize the actor as central and focus on choice but then demonstrate how his/her actions can be predicted from theory, thus lining up with the determinists (Hay, Reference Hay2002). ‘Realists’ in international relations create a stylized world in which the state is reified and endowed with interests, yet they are often allied with positivists, who are normally methodological individualists (see below). It seems to be an outcome of disciplinary politics that their principal opponents have been ‘constructivists’. The factional division is further reinforced by the tendency to caricature other approaches and to simplify the work of the classical scholars in order to demonstrate the distinctiveness of one’s own approach.Footnote 3 This is happening at a time when political science is already losing its distinctiveness in subject matter and methodology.

One result is a misleading division within the discipline and a series of arguments where there could be cross-fertilization and complementarity. Another one, especially within the positivist tradition, is an emphasis on patrolling the boundaries with adjacent disciplines so that political science can maintain its professional status, and, consequently, solutions to new problems are sought within itself. There is a strong reductionist tendency stemming from the effort to model the political process and find parsimonious theories. The inadequacy of these theories then leads to their extension by the rediscovery of old concepts. So the state, or history, or ideas are ‘brought back in’.

Another is the coining of new concepts, which seem to be born without a shared meaning (and not just stretched afterwards). Often it is not clear, when a new concept is used, whether it refers to a new state of the world (after the demise of the nation-state), to something that was always there but we have just discovered, or a new way of describing old ideas. The term ‘governance’ has numerous meanings (Pierre and Peters, Reference Pierre and Peters2000);Footnote 4 indeed, it is used widely both by interpretivists and positivists, on the basis of very different ontological foundations. Meanings include an overarching one (of which government is part), a narrow one (as a sub-branch of government), and one in which governance is replacing government, but all seem to emphasize a lack of hierarchy and the old structural configurations of power. All seem to point to a conceptual gap where politics, government, and the state used to be. ‘Stakeholders’ is a concept that lies somewhere between citizens (of a polity) and interest groups (which do not require a polity). ‘Globalization’ was another master concept that has often seemed to mean everything and nothing, although it has recently begun to go out of fashion as social scientists have realized its lack of precision or explanatory value. ‘Social capital’, borrowed from sociology, has also been stretched a long way, to fill the void where traditional institutions and behaviour used to be. The ideas of ‘construction’ or ‘social construction’ are stretched to cover almost everything (Hacking, Reference Hacking1999).

One effect of extending reductionist models to ‘bring back in’ other elements or coin new concepts is a re-invention of concepts that are already developed in other disciplines. Political scientists could profit from engaging with these disciplines more seriously. Crossing boundaries can also, paradoxically, allow political scientists to gain some internal coherence and find more middle ground in shared concepts. Too often, however, when political science borrows from adjacent disciplines, the ideas are simplified, reduced, or reformulated so as to fit the dominant positivist and individualist approach, and maintain the boundaries of the discipline. Below are three directions in which political science could go to enrich its conceptual and methodological tools, accepting the limits of scientific positivism without falling into post-modern scepticism, and find some middle ground. The first concerns the relationship of the individual to the collectivity, a core concern of sociology and anthropology; the second brings in time and space; and third points to the incorporation of critical realist epistemology and of normativity, from philosophy into empirical research. Many of the same debates are occurring within the adjacent disciplines and, while they may not have reached more definitive conclusions than political science, they have avoided some of the theoretical dead ends into which political science has run.

The individual and the social: into Sociology and Anthropology

Positivistic political science rests on three postulates, which are open to serious question and not quite consistent among themselves. The first is individualism, holding that only individuals really exist (ontological individualism) or that only individuals can act and, therefore, social science is the study of what individuals do (methodological individualism). Most versions of rational choice theory start from the individual and explain broader processes as the aggregation of individual acts. However, the socially disembedded or unsituated individual, far from being the obvious building block for social theory, is a difficult concept to grasp, and the self-interested individual is as much a conceptual construction as is the group.Footnote 5

Second, methodological individualism is combined with a search for decontextualized knowledge of universal application, which finds expression in the widespread insistence that political science is about the study of variables and not of cases. In the words of Przeworski and Teune (Reference Przeworski and Teune1970), we must ‘eliminate proper names’. Action is then explained by the impact of variables, which means that, paradoxically, the individual actually disappears from the explanation. Positivists have, of course, taken into account the complexity of cases, in which factors affect each other, but again reduce it to the same logic as single-variable analysis by modelling their explanation as the interaction effect of variables (Franzese, Reference Franzese2003).

Third, positivists seek explanation based on causal mechanisms. Yet, this just raises further problems as causation itself can take different forms, such as Aristotle’s four (efficient, material, final, and formal). For example, in a murder case, the forensic scientist may say that cause was a gunshot, the police will look for motive, and the criminologist will examine the social conditions that are linked to the incidence of murder. In the positivist tradition, the strongest proof of causation is given by correlation of variables, using large numbers of cases. This form of causal theory is, however, probabilistic, showing only the likelihood that one factor will cause a particular outcome, not the certainty (as with the criminologist above). The uncertainty may for some positivists merely stem from the lack of data and could be resolved with more information, allowing us to control for all relevant variables; but even many positivists would concede that there is at least an undetermined element. In any case, correlation is not causation in a strict sense as found in Newtonian physics, merely an indicator that there is likely to be a causal mechanism, albeit hidden in a ‘black box’.Footnote 6 Sometimes rational choice theory is invoked here to show that, in given circumstances, a rational actor would act only in one way. This violates the rational choice model in two ways. It brings in determinism to choice (see above) and it uses a formal model, which is by definition non-falsifiable, to back an empirical claim. Indeed, as Jackson (Reference Jackson2006) notes, it involves an act of interpretation.

What is missing is a theoretical synthesis that would place the individual in context; account for actions in general as well as particular actions; and incorporate theories of motivation and reasons into explanations of behaviour, a central concern of Weber (2004) and, more recently, Pizzorno (Reference Pizzorno2008). Neo-institutionalism is pushing at the boundaries here, particularly its sociological and normative variants, which show that actors may act not merely according to a logic of consequences but according to a logic of appropriateness (March and Olsen, Reference March and Olsen1984, Reference March and Olsen1989; Hall and Taylor, Reference Hall and Taylor1996).Footnote 7 Political science has also returned to the study of ideas and values, although in the consistent search for positivist consistency it tends to treat these as variables on the same analytical level as other ‘causal’ factors.

It has struggled, however, to re-embrace the concept of culture, which was widely neglected in the course of the discipline’s becoming more scientific. Culture has come back in four ways. Political culture, an idea pioneered in the 1960s, has been revived from the 1980s (Almond and Verba, Reference Almond and Verba1980; Inglehart, Reference Inglehart1988; Lane and Ersson, Reference Lane and Ersson2005). Social capital is a concept that has come into prominence to bridge the gap between individual and collective action (Coleman, Reference Coleman1988; Portes, Reference Portes2001). It is sometimes expressed as trust, an understanding that good behaviour will be reciprocated. Civil society is now often seen as a domain of collective action and interaction outside the institutions of the state, although in its original meaning it covered the state as well (Ferguson, Reference Ferguson1966), doing some of the current work of ‘governance’. Trust is used as a broad concept to explain actions not reducible to the logic of simultaneous reciprocity.

The problem is that these concepts are then usually operationalized according to the individualist/positivist logic and treated as variables, not ways of appreciating whole cases. Political culture is reduced to the social and political values held by individuals and then measured by surveys of individuals. Trust is similarly measured by surveys asking individuals whether they trust people in general or particular types of people.Footnote 8 Social capital is measured either by counting numbers of associations, or by survey questions directed at individuals. The quality of civil society is measured by counting groups. All of these help to fill the gap where conventional conceptions of politics and the state have been found wanting, but they do not take on board the real significance of the underlying concepts. Surveys are prone to level of analysis fallacies (individual or ecological) by inferring from the individual to the collective or vice versa. More fundamentally, the four related concepts are not fundamentally about attitudes or institutions, but about relationships. They are essentially contextual and, rather than being subjective (as in attitudes) or objective (as in associations) they are inter-subjective (Delanty, Reference Delanty1999; Bevir, Reference Bevir2000).

Rather than try to subject culture to the same treatment as other variables in positivist analysis, political scientists should return to a Weberian (2004) understanding and recognize that it has several dimensions (Lichbach, Reference Lichbach1997; Ross, Reference Ross1997). One is as a means of defining the reference group, whether this be an ethnicity, a social class or a social or political movement. Identity has come to new prominence in social science, as the old categories of modern or industrial society seem to lose their power. In its worst form, this becomes a form of primordialism or essentialism, in which individuals are credited with ascriptive identities, which guide and explain behaviour. More sophisticated approaches see social and political identities as constructed, contested, open to change, and often ambivalent. Individuals may have more than one identity, often corresponding to different social roles – say, as a parent, a member of a national group, a member of a class – but even competing as influences within a single role, as when people have more than one ethnic identity available. Identity in this sense is forged by socialization into a culture, which consists of an elaborate series of codes, including shared knowledge and interpretations.

A second element is a framework for interpretation and constructing visions of the world. The scientific endeavour has historically been to establish one set of meanings and interpretation of the physical world, and positivist social science has since the nineteenth century had similar ambitions. Yet, human beings make their own interpretations of themselves, their situation and other humans, and social scientists, in turn, need to interpret the interpretations. An obvious example is religious beliefs, which contain their own cosmologies, including visions of both the physical and spiritual worlds not reducible to instrumental calculation.

A third element concerns the value put on particular actions and attitudes. Here cultural approaches may complement rational choice ones (Lane and Ersson, Reference Lane and Ersson2005). If rational choice analysis assumes that people will maximize their own utility function, cultural analysis helps explain what that utility function is.

None of these three elements implies that societies are homogeneous and monolithic or unchanging. Societal cultures are almost always contested as valuations of behaviour and achievement evolve, and it is this very quality that leads to their more explicit articulation. Interpretations of the world shift and are never more than partially shared. Definitions of group membership are contested at the boundary, which is where much of the most interesting work on culture is done. Individuals normally belong to more than one cultural milieu, receiving multiple and often conflicting signals. Cultural communities are rarely sealed but overlap and link at many points. It is precisely this form of contestation and debate that allows evolution and change so that any society will contain within it the seeds of its own transformation.

Anthropology takes us deeper into the relationship between subjective perceptions of the world and the social context (Geertz, Reference Geertz1973). Ethnographic research starts with cases, looking at their internal logic and the way in which actors themselves read their situation. This does not, as some critics argue, mean that there is no method. On the contrary, cases have to be framed, since they do not exist in a concrete way, a rigorous methodology is needed and specific methods employed (Bray, Reference Bray2008). Studies are ‘holistic’ in examining many aspects of a single case, rather than selected aspects of several cases. They are in-depth, in learning more about a few cases rather than less about more. For many years, ethnographic work and anthropology in general was thought appropriate for ‘pre-modern’ societies, while modern society as the domain of sociology and political science, with their rationalist assumptions.Footnote 9 Since this prejudicial way of dividing the world has been abandoned, ethnographers have moved into the study of European society and politics, with some very interesting results (Abélès, Reference Abélès1989), by no means in general contradiction to those found using other social science methods but often complementing them. Ethnographic findings about how individuals perceive issues can also be used to improve the quality of questions in survey analysis and to interpret their results. Even some branches of economics have overtaken political science here, as institutional economics have moved from narrow transaction costs models for explaining co-operation into richer cultural ones (North, Reference North2005). Bevir and Rhodes (2006), however, while re-introducing tradition (another Weberian concept), eschew ‘cultural schemes’ in an apparent effort to maintain their distance from mainstream political science.

Time and place: into history and geography

At the height of the behavioural revolution, political science was criticized for, if not eliminating time, then at least neglecting history. Since then, historical institutionalists have brought history back in (Steinmo et al., Reference Steinmo, Thelen and Longstreth1992). Some political scientists have used the past merely to generate additional points of observation, as a set of atomized incidents and timeless variables, although Bevir (Reference Bevir2008) is unfair in criticizing historical institutionalists on these grounds. On the contrary, they take cases in their historical context and show how events at one time will influence the future. They do not reduce cases to variables but take them as wholes. Comparative historical analysis similarly puts things in their historical context and examines change over time (Mahoney and Rueschemeyer, Reference Mahoney and Rueschemeyer2003; Pierson, Reference Pierson2004). The impact has been greater in the United States than in Europe, because in most European countries political science has never become so detached from history. Yet, historical institutionalism has come into Europe, often brought by American scholars working on Europe, where its relationship to history is highly problematic. There seems to be remarkably little cross-over or exchange between political scientists using history and historians themselves and little recognition that the discipline of history is racked by many of the same epistemological and methodological disputes as the social sciences.

Historians have too often been dismissed by political scientists as atheoretical or lacking in method, and indeed historical institutionalism is sometimes seen as a way to make history more ‘scientific’. Yet history is theoretical in the sense that it seeks explanations of events, although it does not generally seek universal laws. It proceeds by selection of events from a mass of data and circumstance, and by interpretation of motives, causes, and outcomes. In this, it resembles the more interpretive end of the social sciences, although the connection is rarely made. Historians are also acutely aware of the way in which the selection and writing of history is influenced by events at the time of writing. There has, as in political science, been a reaction away from teleology and the modernist paradigm. Historians are aware that their conceptual and normative lenses are conditioned by present-day concerns. Historiography is prone to constant revision and counter-revision and historians do not expect to arrive at a final consensus on the past.

This insight is often lost on historical institutionalists and practitioners of comparative historical analysis, who tend to treat the past as a set of objective facts and to rely on secondary sources, where the selection of relevant events has already been made. They get away from a pure variable approach and look at whole cases but still depend on a traditional positivist epistemology in order to demonstrate causal effects. This is equally true of the ‘process tracing’ approach, which seeks to fill in the gaps in causal accounts based on correlation (George and Bennet, Reference George and Bennet2005).Footnote 10 The mechanisms may be the closing or opening of options to individuals (as in rational choice institutionalism) or more sociological or normative. This is not to suggest that they are unaware of the problems of historical sources, but Mahoney and Rueschemeyer (Reference Mahoney and Rueschemeyer2003: 12, n. 28), for example, address only the problem of bias (which can be corrected by cross-checking) rather than the essentially contested nature of historical interpretation. Indeed they suggest that their approach is based on ontological realism. Lustick (Reference Lustick1996) similarly problematizes historical knowledge but in the pursuit of a single agreed account, drawing on multiple sources to check bias. A similar problem affects normative theorists who seek to use history as a justification for granting rights. Kymlicka (Reference Kymlicka1995), like others, would grant rights of self-government to peoples who were self-governing at some time in the past. Nevertheless, the debate about which groups existed and whether they were self-governing is so informed by considerations of the present that it merely displaces the normative debate about rights, projected into the past.

The alternative, however, is not to adopt a position of relativism or hyper-constructivism, in which anyone’s view of the past is as valid as anyone else’s or, with Jenkins (Reference Jenkins2006), to abandon history altogether on the grounds that beyond post-modernism there is nothing. History, rather, needs to be approached critically but with a realization that not all accounts are equally valid (Fulbrook, Reference Fulbrook2006). There are two requirements here. One is to accept the inescapable importance of interpretation, not in the post-modern sense in which the act of interpretation is the text, but in the more traditional one in which judgement must be made about the material available and its significance, a judgement that can never wholly be replaced by methodological devices.Footnote 11 Mahoney and Rueschemeyer’s (Reference Mahoney and Rueschemeyer2003) dismissal of interpretation as linked to postmodernism thus leaves historical institutionalism with a large whole at its centre.Footnote 12 The second is to realize that the present may determine our understanding of the past as much as the past determines the present. Indeed objectively bad history, in the form of myths, may be more powerful again. Myths in this sense are not necessarily false, rather their power is independent of their truth or falsehood. They are, rather, an ingredient in the societal culture, a way of making sense of the world, establishing community, and asserting values. Again, there is no reason why political science should remain separated from history or why it should not include critical historiography as one of the factors that influence social and political life.

Political (and other social) scientists are now widely agreed that both state and nation are products of modernity. It is the challenge to these, I have argued, that has provoked the wave of new concepts among positivists as well as the post-modernist challenge. Yet a longer historical perspective shows that politics is perennial and forms of polity can be traced back to antiquity (Ferguson and Mansbach, Reference Ferguson and Mansbach1996). Political science therefore needs concepts that travel in time. If we keep insisting both that the world is new and that we need new concepts to understand it, then historical knowledge becomes deeply problematic. To take a well-worn example, the exhausted debate between primodialist/perennialist and modernist theories of nationalism can better be addressed by a historically informed analysis of the relationship between group construction and boundaries, and institutions over time.

Positivist political scientists have also sought to eliminate space in the search for universalization. For them, territory itself can explain nothing, only the interaction of variables that happen to impinge on a particular place. In so far as it matters, it tends to be defined by the nation-state as a coincident set of social, economic, and political boundaries. For social geographers, on the other hand, territory is a frame that shapes the usual social and economic variables and structures their interaction. We cannot eliminate territorial effects by controlling for other variables, until territory disappears; nor can we estimate territorial effects as a discrete variable, since this would involve comparing territories with a non-territorial world that does not exist. Territories can thus only be compared with each other. Modern understandings of territory go further, by presenting territory not as sharply bounded space, but as something that is constructed socially in multiple, subtle, and complex ways (Paasi, Reference Paasi2002). Different territorial imaginations may overlap and are often contested.Footnote 13 Such understandings have been influential in the study of culture and of economic development, but much less so in political science. Multilevel governance, which comes out of organization theory, for example, does not appear to include any conceptualization of territory at all. This again implies the use of case study methods and ethnographic approaches to complement statistical and institutional analysis.

Epistemology and the normative turn: into philosophy

Commenting on the long division in epistemology, (Hacking, Reference Hacking1999: 84) cites the aphorism that everyone is born either an Aristotelian or a Platonist. Bevir (Reference Bevir2008) criticizes political scientists for their ignorance of recent developments in philosophy as the basis for his radical interpretivism. Yet on another reading, the debate between positivists and constructivists has exhausted its substance. It is difficult to find naïve positivists, who believe that the social world is real in the way the natural world is, or that there is a one-to-one correspondence between that world and their own concepts.Footnote 14 Bhaskar’s (2002) critical realism unpacks the issues, observing that one can be an ontological realist (accepting the existence of the world) but an epistemological relativist. Constructivists do not deny the existence of a material world or the possibility of truth but insist that it is not to be confused with our conceptual understanding of it (Kratochwil, Reference Kratochwil2008).

Similarly, embracing normative questions does not entail abandoning realist ontology or even positivist epistemology. Originating in nineteenth-century conceptions of science, the attempt to exclude norms and values reached its apogee in post-war American social science. The overwhelming value consensus concealed the very presence of values, although the whole modernist paradigm was value-laden and assumptions about liberal pluralism underpinned much of the research effort. Questions of the good were consigned to political philosophy, which tended to operate with hypothetical cases rather than dealing with social reality in its complexity. Cultures not conforming to the modernist paradigm could be assigned to other disciplines, notably anthropology, whose techniques in turn were excluded from the study of western polities.

Yet many of the concepts of traditional political science are normatively loaded. This was always true of ‘development’ (political and economic) or of the Weberian concept of the state. Now that the state itself has been demystified and transformed, its implicit legitimacy is exposed to questioning. Faced with challenges from above, below, and laterally, its defenders need an explicit justification.Footnote 15 Normativity is inescapable when political scientists have to grapple with concepts like democratization, multiculturalism, or self-determination.

The old concepts of state and government provided an apparently clear focus for students of politics, with its demarcated political institutions. With their invocation of governors and governed and its institutionalization in forms of domination, participation, and representation, they also lend themselves to normative questions about representation and accountability. We can talk of democratic government, participative government, tyrannical government and so on, while disagreeing on the practical applications. States can be legitimate or not according to varied criteria, although they were usually taken for granted except in extreme cases or civil strife or ethnic disaffection. With the old certainties of state and government under challenge as analytical concepts, political science is faced with the need to think about the legitimation of power and authority. The emergence of the European Union has sparked another debate about legitimacy and it is striking how often this debate falls back on concepts that are essentially derived from the nation-state frame such as the existence or not of a unitary demos, parliamentarism, and federalism.

The governance debate again exposes this. It raises normative questions rather quickly (partly by exposing implicit normative assumptions that were not questioned in the world of government bounded by the nation-state) but, treating the concept as a neutral or positivist one, its advocates lack the concepts and vocabulary to address them.Footnote 16 The suggestion (e.g. Bache and Flinders, Reference Bache and Flinders2004) that the next stage is to complete the concept by expanding it and then endowing it with a theory of legitimacy involves an inversion of theory (coining a concept and then trying to define it afterwards); normativity is something that is inherent in the design of concepts, not added on as afterthought. The same might be said of stakeholders. The concept carries with it the suggestion of legitimate participants in social and political processes, but is linked to no rigorous theory of how legitimacy might be judged. Yet, at least we can say that the debate about governance has put normative questions back on the agenda of empirical political science.

There is a widespread belief that normative social science is necessarily prescriptive and that it is ideologically skewed. However, most of the debate in normative political theory focuses on broadly shared values and how they can be operationalized and realized in practice, bearing in mind that they may often conflict. This is a task for rigorous logical analysis and empirical social enquiry. Indeed, it is precisely in this combination of normative and empirical work that the social sciences may have the most to offer (Flyvbjerg, Reference Flyvbjerg2001; Van Langenhoven, Reference Van Langenhoven2007). Nor is the potential influence all one way. Normative theorists have been criticized for sociological naïveté including a tendency to reify or essentialize concepts and categories such as ethnicity, nationality, culture, or identity. Ideas such as deliberative democracy are divorced from the social contexts in which they occur and fail to take into account much of what we already know about political behaviour. The combination of normative and empirical theory is one of the most promising developments in modern political science, although it also takes us back to the classical period before the parting of the social science disciplines (Bauböck, Reference Bauböck2008).

Methodological pluralism and middle ways

The modernist project to unite political sciences in a single positivist perspective based on the natural sciences has not succeeded. It was internally inconsistent and, as the modernist assumptions underlying both the natural science analogue and the social world have come into question, it has faced increasing challenge. The effort to try and incorporate new ideas within the same rubric has inhibited learning and led to the creation of new concepts that seem to be stretched at birth. Its more radical interpretivist opponents have been, in some respects, its objective allies, in insisting on a dualist approach and the need to choose between them (Bevir, Reference Bevir2008). We are thus presented with two paradigms, with no commensurability or possibility of combining them.

Those who insist on radically distinct epistemologies will continue to do so and it would be futile to try to dissuade them. However, the alternative to modernist positivism is not post-modernism, where that takes the form of radical indeterminacy or relativism. We can work from a modified form of positivism and take in a great deal more than the strict positivist tradition would usually allow. The positivist vision is always incomplete, as indeed is any paradigm in scientific research. It can incorporate interpretation, context, time and place, judgement, and norms without ceasing to be political science. Indeed, before the middle of the twentieth century this was normal.

An obvious alternative both to the effort to encompass everything within the positivist perspective and the sharp ‘dualist’ one is methodological pluralism. At its extreme, this takes the form of an ontological and epistemological pluralism, in which different approaches would merely co-exist with mutual tolerance. At the other extreme, it refers to the combination of methods in the narrow-sense research techniques, as recommended by those wishing to include qualitative data in the positivist tool-kit. A third approach, however, is to encourage a pluralism of approaches that can inform and learn from each other at the level of methodology, lying between broad world-views and detailed methods. In this sense, it is analogous to social and political pluralism, which is the basis neither for rigid segregation nor for assimilation but rather for mutual influence and development. It is not a search for the ultimate truth but recognizes the inherently uncertain nature of the social sciences. We can probably never arrive at a shared understanding of the social world, any more than we have yet a shared understanding of the physical world; but different perspectives may help us understand diverse aspects and sustain a debate about the whole.

There is no necessary alignment of specific ontologies, epistemologies, methodologies, and techniques. These are separate levels and social scientists can move across them, depending on the question at issue. So rational choice approaches need not be linked to positivist epistemology (as already noted, this is a rather strange combination) but can be linked to culture, where the latter provides motivation. A broadly positivist epistemology is not necessarily incompatible with ethnographic methods, where these are seen as ways of getting more information. Interpretation is a necessity in all social science, since data do not ‘speak for themselves’ and does not necessarily imply a normative relativism or an anti-realist ontology. Looking at cases in a holistic way or placing them in historical or spatial contexts does not necessarily violate the canons of comparative research, since both cases and variables are artefacts of our research design. Analysis of discourse can be a way of exploring motivation and ideas without linking it to radical forms of interpretivist epistemology. The alternative to traditional positivism is not, therefore, merely scepticism in which commensurate knowledge and comparison are impossible. Rather, by reaching across its internal divisions and into adjacent disciplines, political science can find common ground between its various streams. Debate among political scientists can take place on the basis of reasoned argument invoking various forms of evidence and interpretation.

Most of what I have written applies with particular force to the United States where it is may be linked to broader struggles in society. During the heyday of the behavioural revolution, there was a value consensus in the USA in which it was easy to confound the American model of society with modernity itself. More recently, the value consensus that underpinned American interest-group pluralism has given way to a conflict over cultural values, but rather than fostering pluralism this seems to have unleashed a new struggle over universal truth, which finds its counterpart in the struggle over the one-best-way to do social and political research. It is not that Europe does not have the same arguments about the same issues. Political science is a global enterprise and people on both sides of the Atlantic read the same English-language journals. The domination of political science by US scholars and publications in English encouraged their wide diffusion. The universalizing pretensions of this approach also, of course, made it exportable in the way that national European traditions were not. Rational-choice Europeans are much like rational-choice Americans and the same can be said of positivism and constructivism. The difference is in the impossibility of ever imposing a single vision, given the diversity of national traditions, often encapsulated in language. Disciplinary boundaries are drawn differently in different European countries; thus, political science never parted company with history in the United Kingdom and Ireland, is close to sociology in Italy and France, and touches on philosophy and law in many countries, especially in southern Europe. There was no need for a debate in Europe about ‘bringing the state back in’ because in most countries it never went away. There is a deeper concern with culture as tradition, although it is too often contrasted with modernity. The study of the European Union itself has raised questions about the nature of political authority and is necessarily interdisciplinary. Problem-driven rather than methods-driven research, which still, to a large degree, distinguishes European from US doctoral programmes, is another way of opening boundaries. Working within existing frameworks and starting a research project with hypotheses derived from established theories can merely reinforce the paradigms and close off innovation.Footnote 17 Indeed, some of the most innovative work in social sciences is done by people writing outside their own disciplines altogether.

The division between positivists and interpretivists, however, is reinforced by the organization of the discipline and its organization in the form of a profession. Professions have their own territory to defend and do so by claims to possession of specialized knowledge and expertise. Within this there are journals, departments, and promotion systems to ensure socialization into one or another approach and rewards for those who conform. It is also driven by the emphasis on research methods in doctoral programmes leading to methods-driven rather than problem-driven research. The professionalization of the discipline on American lines or the establishment of a single paradigm for research, on the other hand, would be a retrograde step.Footnote 18 So is the mania for competitive citation rankings and the pressure to recognize a hierarchy of journals, which necessarily entails a bias towards particular types of research (Erne, Reference Erne2007). Politics is more than, and older than, political science and the modern state and can be released from the assumptions that those bring to it without risking the nihilism feared by Mahoney and Rueschemeyer (Reference Mahoney and Rueschemeyer2003).

There is always a risk in writing a paper like this of being accused of attacking straw men, or ignoring the complexities of the discipline. There are indeed middle ground approaches out there already. The most crowded middle ground in political science is currently that of neo-institutionalism (Hay, Reference Hay2002), which in its various forms draws fruitfully on neighbouring disciplines and within the social sciences although the cost is often a stretching of concepts to the point where one might question whether there is a central body of theory.Footnote 19 Constructivism, predominantly in international relations, also provides a comparable middle ground, stimulated by hyper-positivist claims of the realists. The perestroika movement in the United States has challenged the orthodoxies of the positivist approach. Modernity itself is being re-appraised in the social sciences (Adams et al., Reference Adams, Clemens and Orloff2004) and, as the radical disjuncture among pre-modern, modern, and post-modern societies comes into question, so are the corresponding social science paradigms. New approaches in evolutionary politics break with the scientistic paradigm and acknowledge that the mechanisms for change in social life must be different from those in the natural world (Lewis and Steinmo, Reference Lewis and Steinmo2008). Work on multiculturalism (Banting and Kymlicka, Reference Banting and Kymlicka2007) and European constitutionalism matches empirical with normative concerns. Most academics writing about politics are, quite happily, able to mix and match methods and probably few consciously adhere to the rigidly opposed paradigms. Yet, in justifying our work and in teaching graduate students we seem to feel a need to adhere to standards that we do not practice ourselves. A more self-reflective discipline, more open to the other social sciences and humanities and more conscious of the classics, would have the confidence to innovate more and to accept that absolute truth will remain elusive.