The ORE of Politics will be available for subscription in late September. Speak to your Oxford representative or contact us to find out more.

Show Summary Details

Page of

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, POLITICS ( (c) Oxford University Press USA, 2016. All Rights Reserved. Personal use only; commercial use is strictly prohibited. Please see applicable Privacy Policy and Legal Notice (for details see Privacy Policy).

date: 19 September 2017

Clarifying Causal Mechanisms in International Relations

Summary and Keywords

Theory and evidence about causal mechanisms, at some point (probably) long ago, reached the carrying capacity for integration into knowledge through expression in words alone. Causal mechanisms, through the implementation of systemism in the discipline of international relations, need clarifying. Systemism is used to convey and analyze the contents of a primary source, Causes of War, by Jack Levy and William Thompson. Explaining war is the most long-standing empirical problem, in the sense of Laudan, in the field of international relations. (Laudan suggested, quite helpfully, a shift from empirical content to problem-solving ability for assessing theories with regard to scientific progress.) The diagrammatic approach from systemism is used to translate a narrative from Levy and Thompson into a series of figures that include causal mechanisms from respective areas of theorizing about the causes of war. The overall purpose of this exercise is to show how the approach from systemism possesses the potential to convey causal mechanisms in a way that facilitates scientific progress. All of this augurs well for a visual turn—toward approaches, such as systemism, that can help to more effectively assemble the massive amount of information now available into knowledge about international relations.

Systemism’s essence has been conveyed by its most long-standing exponent, Bunge: a commitment to building comprehensive theories. Systemism transcends reductionism and holism as the other available “coherent views” with respect to operation of a social system. Instead of theorizing at the level of the system (holism) or its components (reductionism), systemism allows for linkages operating at macro- and microlevels, along with back and forth between them. Systemism also includes inputs from, and outputs to, the environment. This comprehensive procedure facilitates the comparison of alternative visions regarding cause and effect. Thus systemism is an approach rather than a substantive theory. One of its distinguishing merits is a capacity to facilitate criticism and comparison of theories through their representation in diagrams that are constructed under a set of rules to convey causal mechanisms.

Keywords: causal mechanisms, causation, conflict, systems, war, empirical international relations theory, visualization


Theory and evidence about causal mechanisms, at some point (probably) long ago, reached the carrying capacity for integration into knowledge through expression in words alone. A causal mechanism is defined as a declarative statement of the following type: “If x is true, then Y is more likely than otherwise.” Symbolically, this is expressed as “x → y.” The focus here is on clarifying causal mechanisms through the implementation of systemism in the discipline of international relations (IR). Systemism will be used to convey and analyze the contents of a primary source, Levy and Thompson (2010), which focuses on the causes of war. Explaining war is the most long-standing empirical problem, in the sense of Laudan (1977), in the field of IR. (Laudan suggested, quite helpfully, a shift from empirical content to problem-solving ability for assessing theories with regard to scientific progress.) The diagrammatic approach from systemism will be used to translate arguments from Levy and Thompson (2010) into a series of figures that include causal mechanisms from respective areas of theorizing about the causes of war. The overall purpose of this exercise is to show how the approach from systemism possesses the potential to convey causal mechanisms in a way that facilitates scientific progress. All of this augurs well for a visual turn—toward approaches, such as systemism, that can help to more effectively assemble massive amounts of available information into knowledge about IR.

Systemism’s most long-standing exponent, Bunge (1996), has conveyed it essence—a commitment to building comprehensive theories. Systemism transcends reductionism and holism, the other available “coherent views” with respect to operation of a social system (Bunge, 1996, p. 241). Instead of theorizing at the level of the system (holism) or of its components (reductionism), systemism allows for linkages that operate at macro- and microlevels, along with the back and forth between them. Systemism also includes inputs from, and outputs to, the environment. This comprehensive procedure facilitates comparison of alternative visions regarding cause and effect. Thus systemism is an approach rather than a substantive theory (Bunge, 1996, p. 265). One of its distinguishing merits is a capacity to facilitate criticism and a comparison of theories by representing them in diagrams that are constructed, under a set of rules, to convey causal mechanisms.

“International Relations: Yesterday, Today, and Tomorrow,” the next section, reviews the evolution of IR from a sociological point of view. The basic conclusion is that a vast amount of available information and the discipline’s internal complexity makes a turn toward visualization essential for scientific progress. In the section entitled “Systemism,” a diagrammatic approach is introduced that can clarify causal mechanisms through enhanced commensurability between and among expositions. A section on “War and Its Causes” applies systemism to a primary source: Levy and Thompson (2010) on the causes of war. The final section, “Summing Up,” reviews the contributions of the chapter and offers suggestions for future research.

International Relations: Yesterday, Today, and Tomorrow

The accumulation of further scholarship, in the absence of a visual scheme of organization, is inefficient and stands in the way of progress for IR. Identifying the operational nature of that problem—and what can and should be done about it—begins with a brief history of IR up to its state in the early 21st century. The bottom line is that a greater proportion of time and energy in IR needs to be allocated to visualizing the presumed causal mechanisms. Diagrammatic expositions of cause and effect are essential to commensurability between and among expositions, which, in turn, is required for constructive debate in the quest for scientific progress.

While various creation stories might be told about IR, a sociological one is most relevant for the present purposes. From a sociological standpoint, the beginning of IR as a discipline can be traced to the first academic unit in its name. Intense curiosity about the causes and prevention of major wars account for the origins of IR. The field came into existence immediately after the Great War, now known as World War I. Founded in 1919, the Department of International Relations at Aberystwyth University, in the United Kingdom, was the first entity to bring scholars together in an organized and sustained way to study the subject matter of IR. This means that, in sociological terms, IR is just under a century old. A centenarian IR could be regarded as relatively new or old, depending on what is emphasized in the designation. The development of IR over the last century coincided with revolutionary events in the world of technology, and that is most important in gaining perspective on its condition as a discipline.

Improvements in computing power, in particular, have produced an accelerated growth of theory and evidence in IR. These developments, in turn, lead to the properties that matter most in stimulating the increasing mismatch in the division of labor relative to prospects for scientific progress: (a) a discipline that is quite advanced in terms of sheer quantity of output; (b) in which research is increasing geometrically as a byproduct, of personal computers in particular; and (c) a drastically reduced ability to achieve synthesis through commensurable dialogue. These dynamics argue strongly in favor of a visual turn for IR, to clarify causal mechanisms in particular.

During the lifetime of IR as it was identified a moment ago, a series of communication technologies came into being.1 With the advent of the information age—high-speed, personal computing available throughout the developed world, and in expanding enclaves elsewhere—the sheer amount of scholarship produced and available for comparison escalated dramatically. The increasing size of the world’s population and of the number of people who are active as scholars reinforces that point. At present, the trajectory of the quantity of IR research is probably near the high end of the vertical take-off stage, in an s-shaped curve over time. Although population growth is leveling off, development and the attendant infrastructure for research in IR continues to spread around the globe. Further increase in the quantity of research and publication is therefore likely to continue, even if asymptotically, rather than at its recent pace. Thus the driving force behind the argument in favor of a visual turn—accumulation of research findings in primarily narrative form—continues unabated.

For that reason, the concept of division of labor and inefficiency as a byproduct of its current profile comes into play and supports a visual turn for IR. Scholarship in IR overwhelmingly consists of arguments expressed in words. This is true of both qualitative and quantitative studies, along with other standard dividing lines for work on IR. Only a relatively modest proportion of research in any sector of IR focuses on reassessing what is known, from the standpoint of explanation, regarding causal mechanisms. Although review essays appear regularly, generally speaking, these expositions also are expressed in words.2 Thus the overall quantity of written material increases even in the context of efforts to pause and take stock of what is known. It is worth mentioning that people read at the same pace as they did a century ago, which, given the vast amount of research that is available in print, tends to encourage selectivity based on specialization.

One response to the preceding argument might be to ask, “what about formal modeling?” Mathematical modeling, in general, and game-theoretic analysis, in particular, often include the visual presentation of analysis. Game trees and matrices, along with the Edgeworth box, are clear examples of visualization in formal modeling. For two reasons, however, these techniques do not address the problem at hand. First, formal modeling entails major barriers to entry, such as the need for training in mathematics, and thus is not an easily available option across the board. The technical apparatus of formal modeling also inhibits the transmission of its results to the policy world.3 Second, the types of visualization used in formal theorizing do not convey causal mechanisms. For example, diagrams in game theory convey strategic interaction between and among players in a tree- or root-like form. Cause and effect is communicated via equations that calculate equilibria in a game. Thus, while it certainly promotes rigor, formal modeling is not the answer to the challenge of integrating the vast amount of theory and evidence in IR through visual techniques.

Virtually unknown in the field of IR, and pointing toward a significant problem in terms of the division of labor, is the visualization of the discipline’s corpus of knowledge. This is an important observation because of the human inability to integrate large amounts of material—especially presentations with high levels of abstraction—into a coherent framework. While not presuming to convey the research in full, a theory from psychology tells a cautionary tale about the real, as opposed to imagined, human ability to absorb and manipulate concepts. The basic message from cognitive load theory is that people can grasp and manipulate a relatively small number of concepts at any given time, as such entities are introduced.4 With many thousands of books and articles in print at this stage of IR’s development, intuition suggests that the field is at a supra-optimal stage in terms of the sheer quantity of work in comparison to vision-based efforts to synthesize it into a coherent whole with respect to cause and effect.

Given cognitive limitations, priority should be placed on developing and implementing techniques that can summarize complex arguments in visual form. Most important of all is the identification of causal mechanisms. What is a book or article saying in terms of cause and effect? In addition, how might causal mechanisms, within and across such expositions, fit together? Questions like these have generated seemingly endless debates that, disappointingly, often focus on disagreements about the nature of the linkages posited in the first place. For example, multiple contributions to the Vasquez and Elman (2002) volume on realism and the balance of power include debates over the meaning of statements in that publication and others prior to it.

Another development that reinforces the need for a visual turn is the elaboration of IR’s intellectual infrastructure over time. Consider, again, the sociology of knowledge in IR, this time focusing on the International Studies Association (ISA).5 With approximately 7,000 members, the ISA is the principal outlet around the world for exchange of ideas and evidence about IR. The organization’s infrastructure is thus a key piece of information regarding theoretical points of view and substantive interests. “Sections are thematic sub-units,” according to ISA, “organized around specific topics and research interests in international studies” (International Studies Association, 2016). The ISA currently features 29 sections—a number that began to increase at more than a linear pace around the turn of the new millennium. It also is worth noting that the ISA has moved up from the one academic journal it published at its inception to the seven it publishes at present. These numbers point toward the increasing overall size and internal complexity for IR.

Are the properties just identified significant, however, in practice? Perhaps the sociology of the ISA, complex as it seems in terms of sections and other indicators that could be referenced, does not matter in the present context. A glance at some sections in relation to others suggests otherwise; the complex structure corresponds to maturity as a discipline but also represents a challenge to efforts toward integrating information into knowledge.

Some sections of the ISA, such as English School or Diplomatic Studies, can easily be imagined as maintaining close communications back and forth. Both sections are interested, in one way or another, in the idea of an international society. The same observation could be made about additional section dyads and even larger subsets, as well. So bridging elements can be identified in the infrastructure represented by sections of the ISA.

Other section pairings, however, look very much at odds with each other. Take, for instance, International Security Studies (ISS) and Scientific Study of International Processes (SSIP). Many members of each section do not regularly read each other’s scholarship, despite the common dependent variables in research designs that focus on international conflict. Furthermore, especially given the divide between quantitative and qualitative methods, communication generally takes the form of expressions of disapproval that go back and forth intermittently between those strongly affiliated with each section. This conflict comes out most directly about realism (e.g., Vasquez & Elman, 2002). Members of ISS and SSIP tend to be optimistic and pessimistic, respectively, about the place of realism in IR. The foundation for this difference, among many others between these sections, is in research methods. Research produced by SSIP favors quantitative analysis and formal modeling, whereas case studies represent the norm in ISS. This difference largely accounts for conflicting positions on cause and effect, with realism persisting as the center of disagreement.

Some sections of the ISA do not have much contact with each other—neither cooperation nor conflict is observed. A likely sample pairing would be Foreign Policy Analysis (FPA) with International Political Sociology (IPS). The IPS and FPA sections pursue different purposes and feature generally different epistemologies that go a long way toward accounting for their strikingly different research agendas. The basic difference rests in a primarily post-positivist identity for IPS, whereas FPA identifies more with empirical research.

While differences and even alienation exist between and among sections of the ISA, the leading organization in IR, does any of that matter with regard to prospects for more integrated knowledge? Is variation in degree and type of contact among academic specialists a cause for concern? From the standpoint of integration, the potential existence of silos in IR as a system of knowledge matters a great deal. Although a network analysis could do this rigorously, intuition is sufficient to infer that only a subset of units within the discipline of IR maintain regular, constructive communication with each other. Other sectors operate either independently or experience conflict that is very unlikely to produce any higher degree of shared agreement. As a result, the ability to work together toward a greater overall purpose is impaired. In fact, members of the IR community, when reflecting on the question, argue with great sincerity about whether a unifying sense of progress is feasible at all (Freyberg-Inan, Harrison, & James, 2016).

One reason for regular, cyclical “great debates” in IR, of expanding range and even intensity, is the suboptimal division of labor identified in this study.6 Lack of visualization as a technique to facilitate mutual understanding among members of IR as a discipline is the root of the problem. IR is massive in terms of sheer size and increasingly complex vis-à-vis infrastructure, and the ISA offers compelling evidence on both counts. Cycles of debate over fundamental issues such as ontology and epistemology, which are often equivalent to reinventing the academic “wheel” of IR, offer important evidence about the problem posed by the lack of a means toward management and integration of the vast and intricate base of information created by research in IR.

Discussion of the state of IR as a discipline has so far taken a relatively abstract form. Specific examples of the problem identified—vast amounts of research that call out for development of some means toward integration—are needed at this point. Consider as illustrations two classic studies, easily identified respectively with the ISS and SSIP sections from the ISA: Waltz (1979) and Bremer (1992). Each, for various good reasons related to original thinking, stimulated research and continues to be quite influential to this day. Both studies, however, also have had demonstrable reinforcing effects on segmentation in IR.

For security specialists, the structural realist exposition from Waltz (1979) plays an ongoing role in shaping visions of the international system. Its basic propositions are macroscopic; the structure of the international system, assessed in terms of the number of great powers, is expected to create variation in the likelihood of war. Specifically, a bipolar system is regarded as more stable than a multipolar system. This property is inferred from the simplicity of the bipolar system, which makes it more manageable for the two leading powers as opposed to multipolarity, within which the challenges posed by greater complexity produce higher risk of war among the great powers (Waltz, 1979). Thus multipolarity is anticipated to be a more warlike system than bipolarity.

Waltz (1979), moreover, claims to have a theory of international politics, not foreign policy. Thus his theory is self-limiting and can account, even if evidence is favorable, for patterns at only the level of the system itself. Thus the structure of the international system, based on the number of great powers, and its degree of stability, assessed via absence or presence of war, are the purview of structural realism as put forward by Waltz (1979). Power balancing and war are expected to recur in the international system, with no more specific predictions for the conduct of foreign policy (Waltz, 1979). The advent of structural realism caused a vast number of expositions in response, showing diverse reactions (James, 2002; Keohane, 1986).

Among those with an emphasis on the scientific investigation of conflict processes, Bremer (1992) shaped a generation of studies in terms of research design. Specifically, Bremer focused on “dangerous dyads”—that is, on identifying the characteristics that make it more likely that a given pair of states will experience a dispute or even a war in a given year—and found that among the characteristics that matter the most for the likelihood of war in a dyad are regime type and geographic proximity. This is not the only study of its kind from the same era (e.g., Maoz & Abdolali, 1989), but Bremer (1992) serves as an exemplar regarding the shift toward interstate dyad as the unit of analysis for studies of conflict processes. The title of Bremer’s (1992) study emphasizes the new focus on experiences of pairs of states on an annual basis. It is obvious, however, that the results apply to states as actors; traits of the system as a whole cannot be inferred from properties of respective dyads.

Waltz (1979) and Bremer (1992) have one obvious common trait: each focuses on the causes of war as an empirical problem. A less apparent characteristic, revealed when looking across levels of analysis, is that each study contains more ceteris paribus clauses than causal mechanisms. The statistical results from Bremer (1992) assume that effects from outside the dyads can be held constant. The exegesis from Waltz (1979) makes the same type of assumption, except in this context, everything outside of dynamics between social aggregates (i.e., polarity and stability) at the system level is subject to the ceteris paribus clause. Furthermore, a vast amount of research inspired by these studies displays the same property—holding constant all but one subset of potential causal mechanisms across levels of analysis.

What can be said of IR—yesterday, today, and tomorrow? IR is about a century old, in sociological terms, and it has a vast amount of research under its belt and a high level of internal complexity. All that calls to attention the need for a shift in the division of academic labor to paying greater attention to visualization in order to achieve more integrated knowledge. The trend toward greater size and complexity for IR as a discipline can be expected to continue, so the case for a visual turn becomes more pressing with time. Influential studies such as those of Waltz (1979) and Bremer (1992) confirm that sectors within the discipline gain insight but often are self-limiting as a result of multiple ceteris paribus clauses. These observations foreshadow content from the next section, “Systemism,” which introduces systemism as a visual approach with the potential to facilitate scientific progress for IR.


Systemism goes beyond holism and reductionism through inclusion of all causal mechanisms required to fully specify a theory.7 The approach entails a commitment to “understanding a system in terms of a comprehensive set of functional relationships” (James, 2002, p. 131). Figure 1 depicts functional relations in a social system from a systemist point of view. Variables operate at macro- (X, Y) and microlevels (x, y). Outside the system is the environment (E). The environment can be expected to provide inputs into, and experience outputs from, the system.

Clarifying Causal Mechanisms in International RelationsClick to view larger

Figure 1: Functional Relations in a Social System

Source: Adapted from Bunges (1996: 149; see also James 2012).

Within a system as depicted by Figure 1, four basic types of linkages are possible: macro–macro (X → Y), macro–micro (X → x), micro–macro (y → Y) and micro–micro (x → y). In addition, effects may go back and forth with the environment, such as E → X or y → E. In this figure, upper- and lower-case characters correspond to macro- and microlevel variables, respectively. It is difficult to specify in advance how cause and effect travels within and between levels. The nature of causation depends on the specification of mechanisms, which are identified by scholars addressing particular systems.

Systemism encourages specificity with regard to causal mechanisms. Of particular interest is functional form; what, for example, is the nature of Y = F(X)? Is the function incremental or something else? This question must be answered for all connections identified in the system. Assessment of Y as a function of X, by intuition, begins with an incremental or linear relationship, with complexity added as necessary. For example, some linkages may be incremental, such as water cooling down or heating up, and then step-level, with temperatures of 0°C and 100°C, resulting in freezing and boiling, respectively. Functional form also is important in strengthening the falsifiability of a theory by increasing the specificity of its causal mechanisms.

Systemism is of great practical value for articulation of theories to facilitate their comparison to each other. Causal mechanisms are put forward explicitly. This helps with a major problem identified by Cornut (2015, p. 51): the inability to construct explanations effectively in the absence of “contrastive questions.” It is inherently difficult to put together an answer if a query is put forward as, “Why is outcome X observed?” Instead, it makes more sense to ask, “Why is outcome X observed instead of outcome Y?” This approach provides context. So an answer may take the form “X is observed rather than Y because of factors F1, F2, F3, and so on” (Cornut, 2015, pp. 54, 57), Systemism is ideal for putting forward causal mechanisms in a way that facilitates responses to contrastive questions. Thus systemism is a method that facilitates progress through greater clarity of exposition among theories, which facilitates comparison and resolution of differences regarding causal mechanisms.

Look again at the classic expositions from Waltz (1979) and Bremer (1992) in the context of systemism. Following Singer (1961; see also Levy & Thompson 2010, p. 15), the international system is composed of states as its units; both Waltz (1979) and Bremer (1992) are consistent with that specification.8 The two studies, however, dwell in different worlds with regard to designation of causal mechanisms. Structural realism from Waltz (1979) is holistic; its principal causal connection is macro–macro—multipolarity leading to war proneness. Bremer (1992), by contrast, is reductionistic; cause and effect is micro–micro: Certain characteristics of a dyad increase the likelihood of war. In sum, studies such as these, which hold constant one or more types of essential causal mechanisms identified by systemism, can tell only a part of the story regarding war as an empirical problem.

Attention now turns to a study that offers a comprehensive survey of research on the causes of war. The goal in the section “War and Its Causes” is to go beyond examples, such as those just pursued, to show the value of systemism in wider application.

War and Its Causes

Entitled Causes of War, Levy and Thompson (2010) is an ideal choice for application of systemism to illustrate its potential value as a visual means toward converting information into knowledge about IR. The study focuses on causes of war, identified already as the most basic and sustained empirical problem in the subject matter of IR. “Our aim,” as Levy and Thompson (2010, p. 5) observe, “is not to present our own theory of war, but rather to survey some of the most influential theories advanced by scholars over the years and to point out some of the limitations of each of those theories.” The contents of the book are therefore well-suited to present purposes because theories deemed to be of particular interest are summarized and critiqued.

Levy and Thompson (2010) offer a thorough survey of research on the causes of war, and it is beyond the scope of the present study to address all of its contents. Theories included range from system-oriented realism to decision-making in individual and organizational contexts and beyond. Following on from the critique of Waltz (1979) and Bremer (1992), two areas of theorizing will be given further attention: (1) realism, in two self-designated structural variants; and (b) the steps-to-war explanation. Each area is linked to the noted Waltz and Bremer studies, respectively, in terms of intellectual mission. Realists of various stripes use power-oriented models to account for war, while the steps-to-war focuses on interstate dyads to explain why war occurs. Theorizing is presented through a series of figures derived from the narrative in Levy and Thompson (2010), followed by analysis of the properties of these visualizations. The overall objective is to identify the value added from implementation of systemism.

Structural realism from Waltz (1979) received a warm reception from realists across the board because it could offer an account, in their terms, of processes at the level of the international system. Figure 2 conveys structural realism based on the narrative from Levy and Thompson (2010, pp. 31–33). Causal mechanisms begin with MULTIPOLAR STRUCTURE → free riding. (Recall from Figure 1 that upper- and lower-case characters refer, respectively, to the macro- and microlevels.) Note the two broken arrows that follow in the sequence depicted by Figure 2. In this instance and elsewhere, broken arrows are used to indicate causal mechanisms added either by Levy and Thompson (2010) or here because of theoretical incompleteness. The broken arrows appear because Levy and Thompson (2010, p. 33) had to add these to the model from Waltz (1979), which cannot explain why multipolarity, in and of itself, should lack balancing and thus be a more war prone structure. Levy and Thompson add a macro–micro connection, MULTIPOLAR STRUCTURE → free riding. Free riding describes a phenomenon in which members of a group shirk their responsibilities because each prefers that others absorb the cost of taking action.9 This explains the impact of multipolarity in a way that is consistent with an implication from structural realism, namely, a system with more than two great powers entails greater complexity, and management problems are therefore anticipated to emerge. The precise form these difficulties take is free riding. Such behavior, in turn, produces a micro–macro linkage: free riding → LACK OF BALANCING. At this point, the connection made explicitly from Waltz (1979), LACK OF BALANCING → WAR PRONENESS, completes the figure.

Clarifying Causal Mechanisms in International RelationsClick to view larger

Figure 2: Structural Realism

Defensive realism appears in Figure 3.10 Adherents of this variant of structural realism “emphasize the importance of actual threats, of which intentions are an important component” (Levy & Thompson 2010, p. 34). The defensive realist story begins with a synergistic relationship, at the macrolevel: ANARCHY ← – – – → TECHNOLOGY FAVORS OFFENSE. Levy and Thompson, mention both factors (p. 35), but their relationship to each other within defensive realism is implicit. This is why the two-headed broken arrow appears in Figure 3. It is inferred from Levy and Thompson’s (2010) reading of defensive realist scholarship that each factor strengthens rather than weakens the other. The next causal mechanism is macro–micro: ANARCHY + TECHNOLOGY FAVORS OFFENSE → offensive doctrine and military posture. The latter, microlevel factor is listed, along with two others—proximity of a strong state and malevolent leaders, hostile regimes and decision-making pathologies—as making war more likely. This set of micro–micro linkages completes the set of causal mechanisms for defensive realism.

Clarifying Causal Mechanisms in International RelationsClick to view larger

Figure 3: Defensive Realism

Figure 4 displays the steps-to-war argument from Senese and Vasquez (2008), summarized by Levy and Thompson (2010, p. 61). This model of interstate dyads unfolds in a series of micro–micro linkages. The series begins with an interstate dispute and, when focusing on the horizontal arrows (with the exception of the one connecting interstate dispute with territorial dispute, which is vertical), looks like one domino toppling onto another. The final domino in the sequence is the outbreak of war. Note also the presence of vertical arrows that connect the outbreak of an interstate dispute with each of the other steps directly. Thus an interstate dispute, in combination with any of the other steps to war—even if some are not present—also increases the likelihood of war. Levy and Thompson (2010, p. 61) observe that Senese and Vasquez (2008) have not investigated whether the order in which the steps appear is significant.

Clarifying Causal Mechanisms in International RelationsClick to view larger

Figure 4: The Steps to War

With each figure specified, it is time to turn to the subject of value added. What are the individual and collective properties of these figures? Is it possible to identify valued added from the implementation of systemism? Assessing the figures individually and collectively will bring out six overarching points about the value provided from systemism that are not obtained via analysis expressed only in words.

First, Figures 2 to 4 represent major elements of theorizing about war, but each is identified as incomplete from the standpoint of systemism. Even with additions from Levy and Thompson (2010), Figure 2 still lacks any micro–micro component. Yet it took 17 years from the publication of Waltz (1979) for someone to put forward a full-fledged effort toward a structural realist model of foreign policy (Elman, 1996). With a visual representation of structural realism, as it could have been depicted via systemism at the time of its creation, micro–micro, micro–macro and macro–micro theorizing would have been more likely to emerge as priorities to address the self-limiting holism of structural realism. This is not to imply that various critics for years afterward did not complain about the disembodied and oversimplified character of structural realism, but only that systemism can communicate the foregoing problem in a graphic way that encourages complementary theorizing rather than rejection out of hand.

Figure 3, for defensive realism, also conveys incomplete theorizing. With addition of the implied connection at the macro–macro level (i.e., the two-headed broken arrow), the figure includes three of the four basic connections. Defensive realism excludes a micro–macro connection, but represents an improvement over structural realism in terms of the range of causal mechanisms included.

Finally, Figure 4 is exclusively micro–micro and thereby excludes three of four causal mechanisms. It descends from Bremer (1992) in terms of research design and is quite thorough in including aspects of the practice of realpolitik in the steps to war.

Second among the properties identified through the figures is the tendency for steps to be left out of statements of theory regarding the causes of war when these statements are restricted to words alone. Two of the three figures include broken arrows. These arrows had to be added in order to reach the outcome of war. The figures thereby reveal that neither structural realism nor its defensive variant, as stated by advocates, is logically complete. A picture may not be worth a thousand words, but the diagrams of the type advocated by systemism assist in identifying gaps in the analysis from a given theoretical exposition expressed in words alone.

None of the figures contains processes in which there is input or output involving the environment. This third characteristic of the visualizations undoubtedly reflects the tendency for theories about war to focus on the global system. Theorizing about regional processes should be encouraged to incorporate input and output regarding the environment.

Fourth among the traits of the figures is an assumption of incremental relationships. Functional form, encouraged by systemism to strengthen the specificity and attendant falsifiability of theorizing, is absent. The tendency toward implicit linearity in theorizing and an explicit devotion to it via the dominance of regression-style analysis is worth noting. Only a few studies of war, such as Moul (1993), experiment with research designs that include polynomial or other nonlinear expressions.

Seemingly paradoxical, perhaps, is a fifth observation about the series of figures: potential disagreement with representation of each theory in diagrammatic form as a positive aspect of systemism. James and James (2016) introduce this idea in the context of systemism as related to the evolving character of foreign policy analysis:

On the one hand, it is true that another scholar could read the same works and produce something different from the version conveyed by the current figure. On the other hand, that point reiterates a key strength of systemism—the fact that competing visions of a narrative can be compared to each other explicitly in terms of presumed causal mechanisms through a diagrammatic exposition.

(p. 308)

Imagine that authors of works converted to figures based on systemism objected to the contents. This would not be a bad thing. Instead, it would encourage the authors to communicate more clearly, and would-be exponents of systemism to accurately represent the contents of work to be reviewed. Given the many debates about what a particular study really meant, and how unproductive such debates have tended to be over the years, visualization would seem especially worth considering as a means toward greater commensurability in debate. A related point concerns potential value for pedagogy; encouraging students to create systemist figures and to exchange views on that basis could clarify causal mechanisms and enhance learning.

Sixth comes another positive trait: the ability to use figures from systemism to develop connections between and among respective theories. With the rise of analytic eclecticism (Sil & Katzenstein, 2010), systemism becomes of great interest to those who are attempting to put together ideas from across paradigmatic boundaries to obtain more comprehensive and policy-relevant explanations. Long ago identified as a problem by Kuhn (1962), incommensurability stands in the way of eclectic scholarship. Sil and Katzenstein (2010), in an otherwise compelling case for analytic eclecticism, do not provide a method for exponents of their approach to adopt. With its emphasis on the visualization of causal mechanisms and lack of concern for their points of origin, systemism emerges as a natural methodological option for analytic eclecticism. For example, efforts to combine the theories reviewed here—structural realism, defensive realism, and the steps to war—along with many others, would be facilitated by such a visual approach.

Summing Up

Systemism offers a rigorous option for the vast and intricate discipline of IR to consider in developing and comparing causal mechanisms. As a tool of visualization, systemism is especially suited to the digital era, in which competition for attention is extremely high. Cognitive load theory establishes the need for such a visual turn in IR because words alone are not enough to facilitate progress. More visualization of theorizing is essential and systemism stands ready as a means toward commensurable and effective communication in the quest for scientific progress. Diagrammatic representation of the contents from Levy and Thompson (2010), a comprehensive review of theorizing about the causes of war, confirms the value of systemism in particular and a visual turn for IR in general.

What about the future? Systemism can be applied to pry open black boxes of various kinds in IR and beyond. Consider, for example the program of research on civil war. One study that had great influence, Fearon and Laitin (2003), is sufficient to make a point regarding the methodological hegemony of regression-style models and how a misleading sense of cause and effect can be reinforced within such a program of research. Fearon and Laitin (2003, p. 84) offer a multivariate logit analysis of the determinants of civil war onset. Among the numerous, presumably independent variables in the model are prior experience in war, ethnic fractionalization, instability, per-capita income, and mountainous territory. Putting all these variables together in a linear combination to predict the likelihood of civil war seems off target from the outset from the standpoint of causal mechanisms. In visual terms, would it be appropriate to draw an arrow from each independent variable to the dependent variable, to the exclusion of other potential connections? This seems odd because the aforementioned variables also may impact upon each other and some are quite likely to do so. Mountainous territory, for instance, affects prospects for prior involvement in war, among various other variables included. Ethnic fractionalization could have implications for instability. In turn, instability could impact upon per capita income. It would be easy to keep going on in this way about the full set of variables from Fearon and Laitin (2003). Yet all of the preceding variables appear on the right-hand side of an equation with civil war on the left-hand side in a regression-style analysis. From the standpoint of systemism, which would emphasize exploring the full set of causal linkages that might arise, a technique such as structural equation modeling would be preferred. Yet this type of data analysis is exceedingly rare in comparison to regression models that treat all explanatory variables as exogenous. The resulting sense of cause and effect therefore becomes oversimplified in any number of contexts.


Bremer, S. A. (1992). Dangerous dyads: Conditions affecting the likelihood of interstate war, 1816–1965. Journal of Conflict Resolution, 36, 309–341.Find this resource:

Bueno de Mesquita, B. (2009). The predictioneer’s game: Using the logic of brazen self-interest to see and shape the future. New York: Random House.Find this resource:

Bunge, M. (1996). Finding philosophy in social science. New Haven, CT: Yale University Press.Find this resource:

Cornut, J. (2015). Analytic eclecticism in practice: A method for combining international relations theories. International Studies Perspectives, 16, 50–66.Find this resource:

Elman, C. (1996). Horses for courses: Why not neorealist theories of foreign policy? Security Studies, 6, 7–53.Find this resource:

Fearon, J. D., & Laitin, D. (2003). Ethnicity, insurgency, and civil war. American Political Science Review, 97, 75–90.Find this resource:

Freyberg-Inan, A., Harrison, E., & James, P. (2016). Evaluating progress in international relations: How do you know? New York: Routledge.Find this resource:

International Studies Association. 2016. “About Our Sections.” Retrieved from

James, C. C., & James, P. (2016). Systemism and foreign policy analysis: A new approach to the study of international conflict. In S. A. Yetiv & P. James (Eds.), Advancing interdisciplinary approaches to international relations (pp. 289–321). New York: Palgrave Macmillan.Find this resource:

James, P. (2002). International relations and scientific progress: Structural realism reconsidered. Columbus: Ohio State University Press.Find this resource:

James, P. (2012). Deterrence and systemism: A diagrammatic exposition of deterrence-related processes leading to the war in Iraq. St. Antony’s International Review, 7, 139–163.Find this resource:

James, P. (2016, September). A metatheory of progress for international relations: Identifying components. Paper presented at the Annual Meeting of the American Political Science Association, Philadelphia, PA.Find this resource:

Keohane, R. O. (1986). Neorealism and its critics. New York: Columbia University Press.Find this resource:

Kuhn, T. (1962). The structure of scientific revolutions. Chicago: University of Chicago Press.Find this resource:

Lapid, Y. (1989). The third debate: On the prospects of international theory in a post-positivist era. International Studies Quarterly, 33, 235–254.Find this resource:

Laudan, L. (1977). Progress and its problems: Toward a theory of scientific growth. Berkeley: University of California Press.Find this resource:

Levy, J. S., & Thompson, W. R. (2010). Causes of war. Malden, MA: Wiley-Blackwell.Find this resource:

Maoz, Z., & Abdolali, N. (1989). Regime types and international conflict, 1816–1976. Journal of Conflict Resolution, 33, 3–35.Find this resource:

Moul, W. B. (1993). Polarization, polynomials, and war. Journal of Conflict Resolution, 37, 735–748.Find this resource:

Olson, M. (1965). The logic of collective action. Cambridge, MA: Harvard University Press.Find this resource:

Ruane, A., & James, P. (2012). The international relations of Middle-Earth: Learning from The Lord of the Rings. Ann Arbor: University of Michigan Press.Find this resource:

Senese, P. D., & Vasquez, J. A. (2008). The steps to war: An empirical study. Princeton: Princeton University Press.Find this resource:

Sil, R., & Katzenstein, P. J. (2010). Beyond paradigms: Analytic eclecticism in the study of world politics. New York: Palgrave Macmillan.Find this resource:

Singer, J. D. (1961). The level-of-analysis problem in international relations. World Politics, 14, 77–92.Find this resource:

Vasquez, J. A., & Elman, C. (Eds). (2002). Realism and the balancing of power: A new debate. New York: Pearson.Find this resource:

Waltz, K. A. (1979). Theory of international politics. Reading, MA: Addison-Wesley.Find this resource:

Wright, Q. (1955). The study of international relations. New York: Appleton-Century-Crofts.Find this resource:


(1.) These technologies probably created conditions that stimulated rapid development of IR as a discipline, increasing the amount and pace of contact around the globe. As with its origin, the evolving nature of IR reflects events in the real world.

(2.) Technological developments also tend to discourage investing time in the large-scale, book-length expositions that would be required for a comprehensive reassessment of the field. A study such as Wright (1955), for instance, becomes a risky venture in the era of various citation-counting mechanisms that encourage a more rapid pace of publication in an article format.

(3.) A rare exception is the expected utility forecasting model; for accessible applications to IR and beyond, see Bueno de Mesquita (2009).

(4.) James (2016) summarizes findings from research in psychology, stimulated by cognitive load theory, which affirm the importance of limiting the amount of concept formation introduced—and especially manipulation of new abstractions—in order to maximize the likelihood of comprehension among the audience.

(5.) References in this article to the ISA in sociological terms are based on the author’s 37 years of involvement with the organization. This includes regular attendance at conferences and extensive service in governance.

(6.) Lapid (1989) identifies multiple great debates, with particular attention to epistemology and rising opposition to positivism in recent decades. See Ruane and James (2012) for a review of great debates, and the identification of a trend toward disagreement over more fundamental issues.

(7.) The diagrammatic exposition that follows is based primarily upon James (2012; see also Bunge, 1996).

(8.) This leaves aside the question of whether a state-centric designation of the international system is a desirable starting point for theorizing. However, increasing the complexity of the system, in terms of its units or otherwise, would strengthen rather than weaken the case to be made in favor of systemism as a visual representation of theory.

(9.) The classic exposition on the logic of collective action and the free-rider hypothesis is Olson (1965).

(10.) Levy and Thompson (2010, pp. 35–37) also cover offensive realism and other theories that purport to be system level, but it is beyond the scope of the present exposition to include all of that material.