Scientific Approaches to the Study of International Relations

Essay, 2006

14 Pages



1 Introduction: Some Basic Concepts in the Philosophy of Science, and What it Means to Be ‘Scientific’ in the Academic Study of International Relations

2 From Traditionalism to Behaviouralism: the ‘Second Great Debate’ of the 1960s and its Impact on the Contested Epistemology of IR
2.1 The Success of Positivist Methodology in the Natural Sciences
2.2 Traditional Approaches to the Study of IR and the Positivist Challenge
2.3 Positivism and its Shortcomings: Why ‘Scientific Realism’ Faces Significant Limits in the Social Realm

3 Assessing the Conceptual Adequacy and Empirical Potential of a Synthetic Perspective on IR Research Methodology: Two Illustrations, One Suggestion 7
3.1 Game Theory and its IR Applications: Reasons for and Against a New Orthodoxy of Rationalist Social Analysis
3.2 The Democratic Peace Thesis as an Example of Combined Positivist and Normativist Thought in IR
3.3 Bridging the Gap between ‘Hard Core’ Positivism and ‘Radical’ Interpretivism: the Study of World Politics and the Differing Foci of Causal and Constitutive Theories

4 Conclusion: Is Positivism (Still) an Appropriate Toolkit for Systematic Inquiry in the Social Sciences?

Appendix: The Four Humean Conditions of Causality



‘In the first place, in International Relations, as in other branches of Political Science, we are dealing with ideas and concepts which are “essentially contestable” because they have political implications. […] In the natural sciences it is often possible to “stipulate” a definition: that is, to employ a definition of a concept which will be accepted because it is clearly set out in advance. In politics, this is much more difficult—some would say actually impossible’ (Brown, 2005: 11).

1 Introduction: Some Basic Concepts in the Philosophy of Science,and What it Means to Be ‘Scientific’ in the Academic Study of International Relations

Since the behaviourist[1] turn of the 1960s, questions concerning the appropriateness and desirability of a positivist[2] research agenda have been at the forefront of meta-methodological debate within and among the social sciences. The evolving ‘science wars’ between positivists and normativists have also presented enormous challenges to the epistemological identities and professional self-images of scholars working in the academic field of International Relations[3] (IR).

Whereas positivists maintain that the overarching aim of science is the experimentally guided explanation of empirical phenomena under ‘covering laws’ (Brown, 2005: 31), normativists and traditionalists[4] hold that social scientists cannot—and, in fact, should not—emulate the causal models of the natural sciences. According to this view, it is virtually impossible to study the influences of distinct variables in complex social interactions (Nicholson, 1996: 131), and statistical aggregation merely obscures the fact that the true ‘causes’ of events are rarely obvious in the social world.[5] Hence, the purpose of political and social research ought to be a desire to understand processes ‘from within’ rather than to explain them ‘from outside’ (Wendt, 1998: 102).

Yet the traditionalist critique of social scientific positivism did not imply that positivists would be entirely oblivious to the importance of norms in international life. Problem-solving by means of intervention has always been a central motivation of positivist analysis from its very inception.[6] Various authors held that, despite—or rather because of—their different views concerning the function of norms in the process of research, both positivists and normativists should acknowledge the need for different kinds of theory in the social sciences: explanatory, normative and interpretative ones.[7] IR does not only deal with descriptive, but with political (and, ultimately, prescriptive) aspects of the social world.[8] Thus, it might appear worthwhile to ask: how scientific are so-called ‘scientific’ (positivist) approaches to the study of IR—if their theoretical premises and empirical achievements are taken at face value and judged by their own standards of ‘scientific’ neutrality and precision?

To answer this question, I will first describe the spread of positivist thought in IR (Section 2). Secondly, I will outline in how far two research programmes which have been heavily influenced by positivist method—game theory and the democratic peace thesis—have challenged traditionalist approaches, and whether they can be regarded as ‘truly scientific’ (Section 3). Drawing on these insights, I will conclude that, on closer inspection, positivist research methods in IR do lack a perfectly ‘scientific’ status, although they have made important contributions to the academic field (Section 4).

2 From Traditionalism to Behaviouralism: the ‘Second Great Debate’ of the 1960s and its Impact on the Contested Epistemology of IR

2.1 The Success of Positivist Methodology in the Natural Sciences

Positivism as a specific form of empiricism first took hold in the natural sciences. Its basic idea was that ‘observation and experience were the central criteria by which one judged scientific theories’ (Nicholson, 1996: 130). Originally rooted in Newtonian physics and the philosophical tradition of J.S. Mill, this notion of science posited that (1) causes and effects are usually identifiable as distinct factors; (2) it is possible to create experimental situations with study and control groups; (3) independent, dependent and control variables can be specified; and (4), perhaps most importantly, all surrounding parameters can be kept constant[9] for a certain period of time.

Although this framework of experimental or quasi-experimental research has proved to be remarkably successful, philosophers have frequently argued that even in the natural world, the ideas underlying ‘brute facts’ and ‘objective observations’ are often not accounted for by positivists (Wendt, 1998: 107-8): reality—however it is conceptualised[10] —is not identical with its appearances. There appears to be more to observing and explaining things than just sensual experience, and measurements are always subject to (pre)theoretical reasoning and intersubjective (re)interpretation.[11]

2.2 Traditional Approaches to the Study of IR and the Positivist Challenge

Scholars from other fields—notably physicists, systems analysts and mathematicians (Nicholson, 1996: 135)—introduced positivist paradigms into the social sciences from the end of the 1950s onwards. Most IR researchers had hitherto used hermeneutic methods borrowed from the disciplines of history, law and historical sociology (Keohane, 1988: 382). These approaches were actor-centric insofar as they focused on perceptions, ideas, norms, values and ideologies as independent variables.

After the behaviourists had overcome the initial scepticism voiced by many social scientists working in economics and cognitive psychology, their attempts ‘to replace the anecdotalism of traditional diplomatic history’ (Brown, 2005: 32) became more prominent in IR as well (Nicholson, 1996: 129). Rational-choice methods and systematic hypotheses-testing (Lijphart, 1971: 684) were being routinely practised by a growing positivist community. Later on, influential positivists used these research techniques to perform structural macro-analyses within neorealism (Waltz, 1979) or world-system theory (Wallerstein, 1979).[12]

2.3 Positivism and its Shortcomings: Why ‘Scientific Realism’ Faces Significant Limits in the Social Realm

Some proponents of constructivism argue that a rationalist research programme can be justified by the concept of ‘scientific realism’[13]i.e., the assumption that an external reality exists, yet independently of the cognitive experience of human observers (Wendt, 1999). Though this relaxed form of positivism marked a notable departure from strict empiricism, it left major objections to its applicability to IR largely unaddressed.

First, in most game-theoretic frameworks, actors’ preferences are seen as exogenous and defined a priori/ex ante, but such postulations regularly turn out to be quite arbitrary[14] —an internal contradiction that might be at odds with self-proclaimed standards of ‘neutrality’ (Wendt, 1998: 106). For instance, the assumption of bilateral diplomacy as a pre-existing norm in analyses of inter-state bargaining ignores the fact that the empirical centrality of this norm has only been made possible by a number of social practices (Keohane, 1988: 385). History therefore not only matters; it conditions the formation and availability of largely unquestioned preferences in the first place.[15]

Second, the mechanistic notion ‘that the social world can be understood by the same method that we used to understand a machine’ (Rosenberg, 1994: 97) is still to be found in many moderately positivist frameworks, resulting in the formulation of oversimplified stimulus-response models at the expense of ‘creative and emergent properties’ (Almond/Genco, 1977: 492). But, as argued above, the possibility of exact predictions may be questioned in many social and even natural scientific contexts.

Third, social scientists face major difficulties in creating ‘laboratory situations’ in which distinct stimuli and effects can be studied. The ceteris paribus rule and the four Humean conditions of causality (see Appendix) are frequently violated. The inability to rule out any influence of spurious correlations thus prompts the danger of analysing ‘accidental conjunctions’ (Almond/Genco, 1977: 497). Further problems arise because many social processes are not easily measurable[16] or even clearly definable[17].

3 Assessing the Conceptual Adequacy and Empirical Potential of a Synthetic Perspective on IR Research Methodology: Two Illustrations, One Suggestion

3.1 Game Theory and its IR Applications: Reasons for and Against a New Orthodoxy of Rationalist Social Analysis

One might question whether the principle of (bounded) [18] rationality can be attributed to every decision-maker in any given situation (see fn. 14). However, game theorists contend that the very reason for the creation of international institutions is their mediating effect on utter power bargaining, thus avoiding collective action dilemmas by reducing uncertainty and providing a framework for communication and monitoring.[19] For example, the establishment of disarmament regimes (Jervis, 1988: 328) may surely be seen as a ‘learning achievement’ that has reduced incentives to free-riding behaviour.

As regards the epistemological divide between description and prescription, many game theorists claim that they focus on the former rather than the latter. Consequently, the dynamics of the ‘security dilemma’ in the context of US-Soviet arms races during the Cold War could be analysed from a ‘distanced’ neorealist standpoint. However, such an attitude implies that the role of morality in world politics might be ignored.[20]

The analytical suitability of the Prisoner’s Dilemma for describing sub-optimal behaviour in coordination games may also conceal the empirical relevance of contingency in international relations. Most ‘equilibrium’ solutions to bargaining problems are not dichotomous, but should—as in the case of package deals or side payments—be conceived as lying on a continuum of choices between the poles of harmony and discord (Keohane, 1988: 380).[21] During the phase of anti-communist containment, for instance, US policy towards countries which had allegedly come under too much Soviet influence was characterised by ‘stick and carrot’ strategies.

Due to the dynamic nature of most games, preferences are more likely to be shaped by institutions than to shape institutions (Keohane, 1988: 382). Again, the building of security communities under the auspices of a hegemon (the role played by the US within NATO) and their subsequent change through processes of incremental norm diffusion and shifting perceptions[22] (NATO’s move from a staunchly defensive strategy towards ‘Partnership for Peace’) might serve as an example.[23]


[1] Many authors use the term ‘behaviourism’ to label the broad development of positivist thought in the social sciences, whereas ‘behaviour al ism’ denotes the application of positivist research methodology in the field of International Relations, the result of which was ‘an attempt to study the actual behaviour of actors rather than the meanings they assigned to this behaviour’ (Brown, 2005: 32).

[2] Positivism is conventionally described as a general philosophy of social scientific research, although it comprises various sub-categories (deductive/analytic as well as inductive/synthetic approaches).

[3] ‘International Relations/IR’ (upper case) is used here when referring to the academic discipline and ‘international relations’ (lower case) when referring to the empirical subject-matter of IR.

[4] I will use the terms ‘normativists’ and ‘traditionalists’ interchangeably throughout this essay.

[5] For a brief discussion of the ‘causes’ of war, see Brown (2005: 8-9). A comprehensive overview of the contemporary debate regarding civil wars (‘greed or grievance?’) is offered by Berdal/Malone (2000).

[6] Nicholson (1996: 135) employs medical analogies (‘curing diseases’) as a metaphor to characterise the positivists’ desire to improve social and political relations by means of a preceding ‘neutral’ analysis.

[7] Put differently, every social scientist should be aware of the fact that there are separate—albeit interrelated and equally important—theoretical accounts of ‘ why things happen […], what we should do [and] what something or other means. [W]e cannot explain an occurrence without simultaneously interpreting it and orienting ourselves towards it’ (Brown, 2005: 10; italics in original).

[8] To be sure, only critical and postmodernist theorists seem to deny the value of positivist methodology on a categorical basis.

[9] The conceptual power of the ceteris paribus rule in experimental designs is derived from the idea that the researcher can trace back all observed effects on the dependent variable to a variation in the independent variable(s) if—and only if—all outside influences except the stimulus are controlled for.

[10] In fact, there have always been competing theories and core beliefs in the natural sciences as well. This fact is perhaps best illustrated by enduring controversies in physics, where macro-accounts (general theory of relativity) and micro-theories (the indeterminacy of the quantum world) exist side by side.

[11] A researcher’s prior knowledge and personal interests regularly introduce some degree of selection bias and thereby reduce the possibility of ‘neutral’ random selection and observation (Lijphart, 1971: 688).

[12] Structural functionalists legitimised this approach by asserting that actors’ perceptions would be dependent on systemic pressures and necessities. Such actor-centric variables could therefore be reduced to intervening or residual ones—if they were to be defined as true variables at all.

[13] This type of philosophical ‘realism’ is not to be confused with ‘realism’ as a mainstream IR theory.

[14] It seems doubtful, for example, whether individuals are always subject to a utility-maximising kind of behaviour (Jervis, 1988: 341). On the contrary, a lot of actors in international politics might rather be seen as ‘satisficers,’ at least if one does not identify oneself with a purely neorealist perspective.

[15] To illustrate the significance of historical norm-building in another context, the existence of property rights is frequently deemed a necessary condition for the plausibility and accuracy of most models in neoclassical and monetarist economic theory (ibid.).

[16] It is often not obvious what sort of measurement scale (cardinal or ordinal) one should employ.

[17] For a categorisation of political power (attribute, relationship, structure) and a related discussion of the underlying definitional and conceptual problems, see Brown (2005: 81-91).

[18] There are two main types of classical games which have been extensively applied to IR contexts: the Prisoner’s Dilemma and the so-called Chicken Game (Jervis, 1988; Hollis/Smith, 1991: 135-42).

[19] A central unifying assumption of two-actor game-theoretic analyses is that, without any rules constraining their behaviour, players will almost always face an incentive to defection/non-cooperation because of their fear of being exploited by the other side.

[20] This might happen either subconsciously or intentionally. Even the most utilitarian analysis cannot escape a basic sense of normativity (Rawls, 1972). How else should it possible to devise concepts for mitigating violent conflict with the help of (rational) peace-building efforts (Nicholson, 1996: 140)?

[21] Allies may change from balancing to bandwagoning strategies and back, as was largely the case with France’s behaviour vis-à-vis the US during and after the Cold War.

[22] Brown (2005: 12-3) illustrates this finding by referring to the differing aims, calculations and perceptions of Hitler and Stalin during the Second World War. In a similar vein, Jervis (1976: 62ff.) describes the possibility of a ‘spiral’ of misperceptions between individual nation-states.

[23] The creation of the CSCE and its evolution into the OSCE further outline the ways in which an initial situation of crude bargaining might develop into a solid regime of ‘change through rapprochement ’. Of course, the impetus for change might also be induced from external shocks (Hall/Taylor, 1996). The breakdown of the Bretton Woods system of fixed exchange rates (1971/72) is a prominent example.

Excerpt out of 14 pages


Scientific Approaches to the Study of International Relations
London School of Economics  (Department of International Relations)
Catalog Number
ISBN (eBook)
ISBN (Book)
File size
415 KB
scientific, approaches, study, international, relations
Quote paper
Dipl.-Pol., MSc (IR) Jan-Henrik Petermann (Author), 2006, Scientific Approaches to the Study of International Relations, Munich, GRIN Verlag,


  • No comments yet.
Read the ebook
Title: Scientific Approaches to the Study of International Relations

Upload papers

Your term paper / thesis:

- Publication as eBook and book
- High royalties for the sales
- Completely free - with ISBN
- It only takes five minutes
- Every paper finds readers

Publish now - it's free