The Steps to War: Theory and Evidence
- Andrew P. OwsiakAndrew P. OwsiakDepartment of International Affairs, University of Georgia
The steps-to-war theory maintains that war results from the issues under dispute and how states handle these issues. Its foundation rests on the territorial explanation of war, which argues that territorial issues are more conflict-prone than non-territorial ones because these issues constitute a salient security threat that realism recommends be addressed via power politics (i.e., the use of force, including alliance- and armament-building). When states employ power politics, however, the dispute festers, thereby causing recurring militarized conflict; creating feelings of threat, enmity, and competition (i.e., rivalry); producing counter-alliances and arms races; and generally building the more hostile, war-prone world that states originally sought to avoid. Each step taken—from a territorial dispute to rivalry (i.e., recurring militarized disputes) to alliance-building to armament building—therefore increases the probability that war will occur.
Existing empirical evidence supports the steps-to-war theory’s predictions in numerous ways. Tests of the entire theory, for example, demonstrate the dangerousness of territorial disputes, the tendency to manage territorial disputes via power politics, and that individual steps reinforce one another. Other bodies of research connect the individual steps directly to the likelihood that war will occur or highlight the connections between these individual steps—much as the theory predicts. Despite strong empirical support, however, much work remains to be done. Future research should consider the sequencing of the steps to war, investigate why the effects of certain steps vary across different epochs (e.g., alliances differ in their effects on war during the 18th and 19th centuries), identify the alternative paths to war, and study the paths to peace more explicitly—as obtaining peace may not be as simple as removing the known causes of war.
Some time ago, Clausewitz (1968, p. 211) noted that “the progress of events in War seldom proceeds from one simple cause, but from several in common.” War, in other words, is multicausal. This fact has not escaped modern scholars searching for answers to the causes of war. Blainey’s (1988) oft-cited work, for example, uncovers a series of factors that contribute to the outbreak of war (e.g., relative military power, expectations about the behavior of potential third-parties to the dispute, or domestic opinion). Similarly, in his canonical statistical analysis of “dangerous dyads,” Bremer (1992) finds that numerous commonly studied causes of war uniquely alter the likelihood of war onset (e.g., contiguity, alliances, or regime type), even after controlling for any potential interdependence between these causes. It therefore seems that Levy and Vasquez’s (2014, p. 4) conclusion regarding World War I has much broader applicability, for even outside this great war “[t]he processes leading to war [are] characterized by extraordinary causal complexity involving an intricate interplay of variables from all levels of analysis.”
Despite the widely espoused notion that war stems from numerous causes, much of the work on interstate war highlights and explores a single causal factor at a time while controlling for others (e.g., see Cashman, 2014; Levy & Thompson, 2010). This produces theoretical explanations of limited applicability, or what Most and Starr (1989) call “islands of theory.” Steps-to-war theory, however, breaks with this approach to researching the causes of war. It represents an explicit attempt to bridge these islands—that is, to incorporate various individual factors associated with war under a larger, coherent, theoretical explanation. As a result, the steps-to-war theory is partly an inductive exercise to integrate what is known. Yet it would be a mistake to suggest its contribution lies only in this integration, for it also advances a novel theoretical argument—namely, a territorial explanation of war. This explanation helped foster an extensive research program on territorial conflict (e.g., see Gibler, 2012; Huth, 1996; Owsiak, 2012; Wiegand, 2011), and its predictions receive strong, consistent empirical support. In the end, it offers a compelling mid-range theory that illuminates one prominent path to interstate war.
This article provides an overview of the steps-to-war theory. I begin by summarizing the theory’s argument, after which I review the empirical evidence that sheds light on the argument’s merits. Because the argument brings together empirical relationships between territorial disputes, rivalry, alliances, arms races, militiarized interstate disputes (MIDs), and war—and presents logic that operates at both the interstate and domestic levels—my article cannot be entirely exhaustive; nonetheless, the discussion demonstrates the strong empirical support the theory consistently garners.1 After reviewing existing work, I next supply a brief analysis that extends the empirical evaluation of the argument through 2010. Prior to 2015, data on militarized interstate disputes ended in 2001, causing many studies—including recent ones—to end in 2001 as well (see Ghosn, Palmer, & Bremer, 2004; Palmer, D’Orazio, Kenwick, & Lane, 2015). The release of the new militarized interstate dispute data (version 4.0), however, allows me to evaluate the argument through 2010, an important undertaking for those who believe the theory may not explain interstate conflict in the modern world. Finally, I end with a discussion about where future work can advance the research program started by the steps-to-war theory (for an overview, see Vasquez, 2009). As this latter discussion demonstrates, much remains to be done.
The steps-to-war theory adopts an issue-based paradigm, in which the raising and handling of issues constitutes the primary focus (Senese & Vasquez, 2003, 2005, 2008; Vasquez, 1993, 2009). It may now seem rather pedestrian to argue that what states disagree over affects how they behave. Nonetheless, this position—when advanced—diverged drastically from previous theoretical arguments (e.g., see Diehl, 1992). Earlier theories typically adopted the realist paradigm. For such scholars, the issues under dispute had little bearing on state behavior. Considerations of (relative) power, they believed, motivate states to fight, and this motivation arises regardless of the immediate issue that seems to trigger interstate conflict (for an overview and critique of the realist paradigm, see Vasquez, 1998a).
The steps-to-war theory takes a drastically different approach (Senese & Vasquez, 2005, 2008; Vasquez 1993, 2009). It begins from the premise that the issues over which states fight condition their conflict behavior. Of all the issues that states can dispute, territorial issues are most dangerous. According to Vasquez (1993, 2009), humans have a biological “proclivity” to divide the world into units and defend those units with force—a trait he refers to as territoriality. Moreover, territory possesses high levels of salience to leaders and their domestic audiences for a number of reasons, including the resources it contains (e.g., oil or water wells), its strategic or economic importance (e.g., control of straits or access to the sea), or the symbolic value that accompanies it (e.g., control of ethnic homeland). Finally, states learn that it is appropriate—and, in pursuit of their interests and preferences, perhaps best—to handle their territorial issues via violence.
This last point deserves greater explication. How do states learn to employ violence with respect to certain issues and not others? States can determine how to divide disputed goods through any number of allocation mechanisms, including through applying values (i.e., norms, such as equality), voting, bargaining, or the use of force (i.e., unilateral behavior; Vasquez, 1993, 2009). Which methods they employ in a particular context, as well as the order in which they employ them, derives from domestic and global institutions—the set of formal and informal rules and norms that govern international interactions. This suggests three conclusions. First, the use of force is not the breakdown of international order but serves a function within that order (see Bull, 1977). Second, force is only one mechanism for handling disputed issues. Alternatives exist and can be used. The goal is therefore to understand why states employ force more often when territory—but not other goods—is at stake. Finally, the use of force results when institutions constrain leaders and make war a permissible, appropriate means of handling their issues. In the end, the institutional context plays a key role in determining why and when states resort to violence against one another.
Three sets of institutions influence interstate decisions (Vasquez, 2009, pp. 170–171). At the international level, rules about the proper use of force exist. These rules change over time, with some periods being more accepting of violence than others. Currently, for example, the United Nations charter heavily restricts the use of violence, permitting states to use it only in self-defense (Article 51; see also the Concert of Europe period). At other times, states may have greater leeway to employ force in pursuit of their goals. At the dyadic (interstate) level, states build a history of interactions with one another that creates expectations about their future interactions.2 Repeated friendly interactions (e.g., United States–United Kingdom) build “a reservoir of psychological friendship” that can weather short-term negative encounters, while repeated hostile interactions yield “a reservoir of psychological hostility” that can overwhelm cooperation gestures (Vasquez, 2009, p. 171). These reservoirs can undermine the success of certain allocation mechanisms (e.g., successful bargaining will be challenging in the face of hostility). Finally, at the domestic level, leaders do not select policies in a vacuum, but rather within constraints set by the preferences of other government leaders, elites, and constituents (Putnam, 1988). As domestic preferences shift, so too will the foreign policy options available to leaders.
The issues under dispute can alter institutional constraints at all levels, thereby encouraging certain foreign policy actions at the expense of others. When a security threat arises, states employ prescriptions derived from realism—the dominant paradigm (see Vasquez, 1998a). Realist thought depicts an uncertain world in which states are selfish, self-reliant, and potentially aggressive. As a result, states must confront security threats through power politics—that is, unilateral efforts and the threat or use of force (e.g., demonstrations of resolve or military encounters)—for nobody will come to their aid and they will fall to the stronger if they are weak. A territorial dispute presents one such security threat, in which two (or more) states compete over a highly salient object (see above). States will try to manage these territorial disputes peacefully (e.g., see Hensel, McLaughlin Mitchell, Sowers, & Thyne, 2008; Owsiak et al., 2017). Yet their efforts will generally fail, as the high salience involved prevents them from making the concessions necessary to resolve the matter (see Wiegand, 2011) and discourages them from submitting the dispute to other allocation mechanisms over which they have less control (e.g., adjudication; see Owsiak & Mitchell, 2017). They therefore begin to threaten or use military force. Because territory is salient for both parties, however, threats produce counter-threats, and the use of force is met with a counter-force.
Unilateral (power politics) actions work well if one actor possesses an extreme power advantage (e.g., a major state versus a minor state). She can then achieve a decisive victory and impose her will on the opposing state. Yet this will be uncommon in territorial disagreements. Most of these disagreements occur where states come into contact with one another—along the borders between neighboring states, the vast majority of which are closer to relative power symmetry than extreme asymmetry. Given this configuration of relative power, neither state is likely to impose its will on the other, causing the use of force to produce stalemate outcomes. Force will therefore accomplish little in territorial disputes; instead, these disputes will fester.
When a territorial dispute festers, the threat persists. In fact, it intensifies, since the opponent’s willingness to use force confirms its aggressive intentions (see Maoz & Mor, 2002). Moreover, the salience of the disputed territory may also rise, as the opponent’s revealed hostile intentions can make it imperative that she not hold the territory, lest she destroy the state’s identity, resources, or populace, or turn any advantages the territory holds against the state. Once again, realism prescribes the use of force in the face of such a threat, and repeated militarized encounters follow, as each state finds it important to pursue the territory but incapable of imposing a decisive victory on its counterpart. These repeated, hostile interactions cause the involved states to see one another as a threat, enemy, and competitor—the core characteristics that define an interstate rivalry (Colaresi, Rasler, & Thompson, 2007; Diehl & Goertz, 2000).
Hostile, dyadic interactions between rivals yield two broad effects. First, the failure to resolve disputed issues, and the perceptions of threat and enmity that accompany it, encourage the involved states to institutionalize the rivalry. A history of repeated, negative interactions suggest that such interactions are likely to continue into the foreseeable future. As a result, states begin planning their foreign policy around the need to confront the rival. Second, persistence of the rivalry encourages a more hard-line (i.e., power politics) approach. Aggression convinces domestic audiences that the rival constitutes a threat to the state. Because it seems impossible to advocate or pursue the use of peaceful allocation mechanisms against an actor who behaves aggressively and threatens the state, domestic audiences will increasingly prefer a power politics approach. Moreover, failure to achieve the state’s goals is often attributed to not being aggressive enough (see Leng, 1983). Hardliners—as opposed to accommodationists—will therefore come to dominate domestic politics (Vasquez, 2009). This will constrain leaders’ actions, pushing them to rely increasingly on realist prescriptions (i.e., power politics).
A series of repeated militarized disputes that end in stalemate will convince a state that it is not powerful enough to achieve its goals. Realism then advocates two additional policies to enhance one’s power: building alliances and armaments. Alliance building represents an efficient, short-term means of gaining power. Through alliances, states augment their power by adding the alliance partner’s power to it. Realism maintains that this should either facilitate concessions (e.g., alliance partners making demands on a state without alliances) or deter aggression (e.g., a state without alliances choosing not to make demands on one with allies). Either way, they argue that the likelihood of interstate violence should decline. This, Senese and Vasquez (2008) propose, is misguided. As one rival builds an alliance, the other experiences heightened alarm—for it is now weaker than its counterpart, who it perceives to be aggressive toward it. One state’s search for security thereby promotes insecurity in the other (i.e., the security dilemma), and the non-allied state therefore begins searching for alliance partners as well. In short, alliances yield counter-alliances.
If alliances cannot offer a way to achieve their goals, states then turn to armaments. The process of building strength through armaments can be slow. Whereas allies can offer military forces immediately, creating ships or weapons, recruiting new troops, and finding money to expand the military can all take time. Nonetheless, rivals who expect a major threat to last for the foreseeable future might see armament-building as a prudent policy. As with alliances, however, armament-building is likely to threaten the adversary, causing them to do the same. An arms race then develops, with each state investing significantly in their military so as to prevent becoming weaker than—and, therefore, at the mercy of—their opponent.
The whole of this discussion constitutes the steps-to-war argument. It rests on a unique, territorial explanation of conflict. Territorial disputes are more likely to escalate to war than disputes over non-territorial issues—for the reasons identified earlier. Yet they are also more likely to set off a chain of events that activates the “realist road to war” (Senese & Vasquez, 2008; Vasquez, 1993, 2009). These disputes constitute a security threat for which realism prescribes power politics. A dyad then turns to militarized conflict, but neither dyad member is likely to win this conflict outright. Territorial disputes therefore fester, causing repeated militarized disputes and the formation of rivalry. The hostility and threat that accompany this process next promote alliance- and armament-building, which foster counter-alliances and arms races. As a dyad progresses through each step—territorial dispute, repeated militarized disputes/rivalry, alliance-building, and arms races—the likelihood of war rises.3
This broad argument requires three final caveats. First, the steps-to-war theory does not intend to explain all wars. Rather, it advances one causal pathway that purports to explain a significant number of interstate wars (i.e., those between states of relative power symmetry). Other explanations of war necessarily exist, and these can co-exist with it. Second, the steps to war seem to function as a sequence of events, but Senese and Vasquez (2008) maintain that they should not be viewed in this way. What matters most for increasing the likelihood of war, they argue, is how many steps the involved states take, not the order in which the steps are taken. I nevertheless return to the question of sequencing in the final section of this article. Finally, the steps-to-war theory potentially applies to non-territorial disagreements as well. One of its core arguments is that realist prescriptions (i.e., power politics) create the hostile world that realism seeks to escape. Power politics always heightens the probability of war, regardless of the issue under dispute. Thus, if states use power politics to manage non-territorial disputes, we expect that the likelihood of war rises as they advance further down the realist road to war in these disputes as well. Nonetheless, the steps-to-war theory advances strong theoretical reasons to believe that territorial disputes prove particularly dangerous—that is, more likely to escalate to war than non-territorial disputes—across each of the steps to war.
The steps-to-war theory finds broad empirical support, largely through three types of studies: those investigating all steps simultaneously, those demonstrating a relationship between individual steps and interstate conflict or war, and those examining the relationship between the individual steps themselves (i.e., how the steps mutually reinforce one another; e.g., territorial issues and rivalry).4 The first remains the least common, and it involves an examination of the entire theory’s logic. Senese and Vasquez (2005, 2008) exemplify this approach and investigate two broad points in their work. First, they study how each of the steps in their argument contributes to the likelihood of war.5 The empirical evidence here is strong. They conclude that the probability of war given a dyadic dispute over a policy matter is 0.09. In contrast, this probability sits higher—at 0.15—for territorial disputes. When they add an external alliance to the dyad, this latter probability then climbs to 0.45. Incorporating repeated militarized interstate disputes (MIDs) pushes the probability higher—to 0.55 after 6 MIDs and 0.65 after 15 MIDs. Finally, after they add an arms race to all these factors, the probability of war within the dyad reaches 0.90.6
Two points are worth noting about these general findings. First, Senese and Vasquez (2008) always find that territorial issues are more conflict-prone than non-territorial issues—in every model, within each distinct time period, and across all steps they examine (e.g., comparing policy disputes with arms races to territorial ones with arms races). They underscore this by first demonstrating that a territorial claim—that is, an official state position in which two or more states contest sovereignty over a given territorial space (see Frederick, Hensel, & Macaulay, 2017; Huth & Allee, 2002)—increases the likelihood that a MID will occur. Then, they show (via a two-stage model; see Reed, 2000) that a MID is more likely to escalate to war when it involves territorial issues. Finally, they illustrate how adding each additional step in their model generally raises the likelihood that war will occur; yet at each step, the likelihood of a territorial dispute escalating to war is higher than for non-territorial issues. This offers strong evidence in favor of the territorial explanation of war upon which the steps-to-war theory sits, as well as the larger theory itself. Second, the findings weaken in the 1946–1989 period, largely because alliances behave differently during the Cold War than either before or after it. During the Cold War, alliances appear to have a conflict-dampening, rather than conflict-promoting, effect. Nonetheless, territorial disputes still prove more conflict-prone in this anomalous period.
A second main point Senese and Vasquez (2008) explore concerns the interaction between the individual steps in their argument. The parsimonious version of the model is additive, proposing that each step taken increases the likelihood of war (on this point, see also Levy, 2012). It also, however, suggests potential interactive effects (Senese and Vasquez, 2008). A(n) (multiplicative) interaction can either amplify or dampen an additive effect. For example, territorial issues and rivalry may each raise the likelihood of war (i.e., an additive effect), but the presence of both factors together may raise it even higher than each would without the other (i.e., a multiplicative effect).
The steps-to-war theory predicts a few interactive effects. The strongest involve territorial issues, which should interact (multiplicatively) more often with the remaining steps than any other variable (Senese and Vasquez, 2008). This occurs because territorial issues arise prior to the remaining steps and set the entire theory in motion. Moreover, these issues possess high salience to domestic audiences and incentivize states to use power politics, thereby producing recurring disputes, the feelings of threat associated with rivalry, a search for allies, and the buildup of armaments. Additional interaction effects derive from rivalry. The feelings of threat and enmity, along with repeated hostile interactions, encourage states to search for allies and fund arms buildups. Empirical analysis confirms the presence of these various hypothesized interactive effects (Senese & Vasquez, 2008; see also Colaresi et al., 2007).
Besides Senese and Vasquez, few other authors test the entire steps-to-war theory. There are, however, three noteworthy exceptions. The first appears in Colaresi et al. (2007). These authors offer a thorough multivariate test of the theory using different data and model specifications than Senese and Vasquez (2008), and they reach similar conclusions. Territorial issues, repeated crises, and arms races significantly increase the likelihood of war. External alliances behave inconsistently across the models but generally exert no significant effect on the probability of war. A subsequent exploration of interaction effects, however, reveals a role for alliances: serial crises combined with either arms races or only one rival having an external ally raise the likelihood of war, whereas a serial crisis in tandem with both rivals having external allies decreases the likelihood of war. Finally, the key interaction once again concerns territorial issues, as Senese and Vasquez (2008) predict. Although territorial issues outside of rivalry remain dangerous, rivals that contest territorial issues are particularly conflict-prone. All of these findings remain consistent with what Senese and Vasquez (2008) argue and find.
The remaining studies involve more in-depth case work. Sample (2014), for example, examines the management of 42 territorial claims that experience war. Her goal is to determine whether the steps-to-war theory or the “anticipation of war” criticism of it—that is, that states anticipate war, use power politics as a result, and end up in the war they expected, thereby making the steps to war epiphenomenal—best explains these cases. In the end, she concludes that there are multiple paths to war, that the steps-to-war theory explains a significant number of her cases, and that the anticipation of war criticism performs poorly (see also Sample, 2017). Similarly, Vasquez (1998b) and Vasquez and Gibler (2001) review the origins of World War II in Europe and Asia, respectively. They select these cases because of their prominence and, with respect to Asia, because other common explanations of war cannot account for it (e.g., power transition theory). In the end, the authors determine that the steps-to-war model fits these cases well, although they acknowledge candidly that their analyses cannot be considered definitive tests of the theory. Nevertheless, such qualitative work complements the quantitative analysis, and both methodologies reveal consistent support for the theory.
Findings Connecting Individual Steps with War Onset
A second body of research sheds light on the relationship between the four individual “steps” in the steps-to-war argument—that is, territorial disputes, rivalry (or recurring conflict), alliance making, and arms races—and the likelihood of war. Because the steps-to-war theory proposes that each individual step adds to the probability that war will occur, we should observe an empirical connection between each step and an increase in the likelihood of war. Critics might dismiss this evidence, arguing that the theory should receive significant empirical support from this body of research because the theory emerged inductively through an accumulation of existing findings on war (Senese & Vasquez 2005, 2008; Vasquez, 1987, 1993, 2009). Senese and Vasquez (2008), however, offer a compelling counterargument: the steps-to-war theory rests on a territorial explanation of war, which was itself proposed before data on territorial disputes existed (for a history of the steps to war research program, see Senese and Vasquez, 2008). Evidence connecting territorial disputes with war therefore followed the theory’s construction and predictions. Moreover, the quantitative research on rivalries began independently of and simultaneously to the steps-to-war theory, making it unlikely that such findings could be used during the theory’s construction (see Goertz & Diehl, 1992; Thompson, 1995). In the end, evidence linking individual steps in the steps-to-war theory to the occurrence of war cannot be dismissed out of hand.
Territorial Disagreements and War
The strongest and most distinct findings in favor of the steps-to-war argument concern the connection between territorial issues and conflict. The steps-to-war theory grounds itself in a territorial explanation of conflict—a characteristic that distinguishes it markedly from other theoretical arguments. The theory proposes that territorial issues—when compared to non-territorial ones—are more likely to produce aggressive, militarized behavior; more likely to recur (i.e., defy easy settlement); and, once militarized, more likely to escalate to war. The empirical evidence strongly supports each contention. First, states behave significantly more aggressively toward one another when territory is at stake. For example, Hensel (1996) finds that when one state threatens, displays, or uses force against another state in the international system (i.e., a MID occurs; see Palmer et al., 2015), the targeted state is three times more likely to respond militarily when the disagreement concerns territory than when it does not (see also Tir & Diehl 2002). Moreover, MIDs over territorial issues lead to fatalities more often than those over non-territorial issues, even when the disputing dyad members are separated from one another by significant geographic distance (Hensel, 2000). Hensel (2000) argues that this occurs because territorial issues possess more salience than other issues over which states might disagree (see Hensel et al., 2008, for a full discussion). Indeed, when he investigates this proposition, he notes that militarized disputes increase as issue salience rises. Because Hensel only has salience measures for territorial disagreements, this latter finding is not direct evidence of the steps-to-war theory, which compares territorial and non-territorial issues. Nonetheless, if territorial issues possess more salience to states than non-territorial ones—as Hensel et al. (2008) argue—such findings support the logic of the steps-to-war theory.
Second, territorial issues persist and fester. Hensel (1994) provides the most thorough evidence on this point. He finds that a territorial MID is nearly two-and-a-half times more likely than a non-territorial MID to produce a subsequent MID concerning the same (i.e., territorial) issue. This stems from the fact that territorial MIDs end in stalemates more often than their non-territorial counterparts—the latter instead producing relatively more decisive military victories or compromise outcomes. In addition, territorial MIDs recur faster than disputes over other issues; a territorial MID recurs nearly twice as fast as a non-territorial MID. These recurrences (and the speed with which they happen) help explain why territorial issues generate hostility over a short period of time and therefore feature prominently in rivalries, precisely as the steps-to-war theory argues (Dreyer, 2010; Tir & Diehl, 2002).
Finally, territorial MIDs escalate to war more often than non-territorial ones.7 More precisely, Vasquez and Henehan (2001) find that 67% of all militarized interstate disputes that end in war are territorial. This finding holds across multiple time periods, as well as dyads containing different combinations of major and minor states. It is therefore quite robust. Further evidence derives from Holsti (1991), who shows that states have fought more wars over territory than any other issue during the entire 1648–1989 period, as well as the various subperiods within it. This suggests that territorial disagreements significantly increase the likelihood of war—a finding other scholars repeatedly confirm (see Toft, 2014, for an overview of this research).
More recent research explores the nuances of the territorial hypotheses that underlie the steps-to-war theory. Vasquez (1993, p. 141; 2009, p. 155) originally argued that a territorial explanation of conflict means that states will divide the world into territorial units (e.g., states) and defend these units with force. As a result, states pay particular attention to their borders, which they will contest until fully drawn and agreed upon. It is the contestation over the border’s placement that produces friction between neighboring states; after delimitation, neighboring states will therefore develop more peaceful relations, since they lack this highly salient source of disagreement.
Vasquez and others originally found data on territorial disagreements within the MID data (see Senese and Vasquez, 2008), shortly after publishing the steps-to-war theory. These data offered one means to test the argument but stopped short of considering border disagreements in particular. Gibler (2007, 2012) and Owsiak (2012), however, later supplied this data—Gibler by using a series of carefully selected proxy variables and Owsiak by collecting data on states’ signing of border treaties. Their analyses confirm the steps-to-war theory’s predictions. Neighboring states that lack clearly defined borders are at significantly greater risk of experiencing MIDs and wars. Once they settle the status of these borders, though, they are less likely to fight militarily over any issue, yielding notably long stretches of interstate peace (see also Owsiak, 2017).8 Such findings offer strong support to predictions advanced by the steps-to-war theory nearly two decades prior.
Rivalry and War
The steps-to-war argument next maintains that two states incapable of resolving their disagreement via diplomacy will eventually turn to militarized conflict. If these states are relatively equal in power, the militarized interstate disputes that result will also fail to resolve the disagreement—for neither state can unilaterally impose its preferred solution on the other. A highly salient issue (in this case, a territorial disagreement) plus recurring militarized conflict (see above) plus relative power parity means that states will develop a relationship in which they view one another as threats, enemies, and competitors (Colaresi et al., 2007). In short, a rivalry forms (see also Diehl & Goertz, 2000).
The empirical relationship between rivalry and war is well established. According to Diehl and Goertz (2000, pp. 62–63), 82% of all wars during the period 1816–1992 occur within a rivalry relationship (i.e., within only 108 dyads; see also Vasquez & Leskiw 2001). Colaresi et al. (2007, pp. 88–89) reach a nearly identical conclusion for the 1816–1999 period, noting in addition that wars fall increasingly within rivalries in more contemporary periods, as opposed to more historical ones. Why rivalry facilitates war is not particularly puzzling. Rivals contest highly salient issues that are difficult to settle (see below on the territory-rivalry connection; Rider & Owsiak, 2015; Tir & Diehl, 2002). As these salient (often territorial) issues persist, they generate feelings of threat, enmity, and competition, which then yield a generally more hostile relationship (Colaresi et al., 2007; Diehl & Goertz, 2000). This combination of salient issues, repeated negative interactions, and perceptions of a rivals’ malign intent significantly raise the probability of violence—including war—as rivals experience successive crises (Colaresi et al., 2007; Vasquez, 2009).
Beyond these larger findings about rivalry and war, the steps-to-war theory makes two additional predictions. First, it advances a specific mechanism that describes how domestic actors affect the process by which rivalry leads to war. As a dispute festers and militarized conflict recurs, the theory argues that the balance between accommodationists and hardliners tips in the latter’s favor. If true, two observations follow. First, tolerance within society should fall as opinions homogenize toward a hard-line perspective. Research generally confirms this. When a state faces a salient external threat (e.g., a rivalry, many of which contain a territorial dimension; see Tir & Diehl, 2002), political tolerance among individuals living within that state toward other groups in society declines (Hutchison & Gibler, 2007). This effect seems stronger when individuals view themselves as the target of a threat (see also Hutchison, 2011; Miller, 2013). Nonetheless, the trend toward less political tolerance when faced with a threat supports the steps-to-war theory’s general argument. Second, as hardliners increasingly dominate domestic politics over time, rivals should behave more aggressively in successive disputes. Leng (1983) offers evidence consistent with this point. He demonstrates that dyad members learn to use increasingly coercive bargaining strategies as their crises repeat—a finding confirmed by Colaresi et al. (2007). Evidence like this suggests that the steps-to-war theory has identified one causal mechanism by which domestic decisions influence interstate outcomes.
Beyond the domestic level, the steps-to-war theory predicts that, although rivals can dispute any number of individual issues and rivalries frequently begin specifically because of territorial issues (see below), rivals will frequently link their many disputed issues together into “one single grand issue—us versus them” (Vasquez, 2009, p. 80). This results from the perceptions of threat, enmity, and competition that drive rivalry, which cause rival states both to reorient their decision-making calculus toward inflicting harm upon and to institutionalize the expectation of repeated, future hostile interactions with their rival. As predicted, issue linkage seems extensive within rivalries. Dreyer (2010) and Mitchell and Thies (2011) estimate that 46–63% of all rivalries contain more than one issue. Moreover, the effects of this issue linkage can be considerable. Rivals with multiple issues under dispute experience significantly more MIDs (Dreyer, 2010; Mitchell & Thies, 2011). Once an issue is militarized (i.e., a single issue experiences more than one MID), the dyad is also at increased risk of additional MIDs and war occurring (Mitchell & Thies, 2011). The choice to use military force therefore substantially alters states’ relationship. That they initially turn to military force largely as a result of territorial issues (Mitchell & Thies, 2011) is consistent with the steps-to-war theory. It is important to note, however, that the steps-to-war theory also predicts issue linkage within rivalries and that this linkage will foster further militarized conflict—all of which is confirmed by empirical studies.
Alliances and War
The effect of alliances on the likelihood of war—a third step—exhibits significantly more complexity than the previous steps examined and represents one area in which future research remains to be done. Some research anticipates this complexity; Smith (1995), for example, shows that deterrence (i.e., peace) and escalation (i.e., war) can both result from effectively functioning alliances. The steps-to-war theory, however, maintains only that alliances foster conflict; it argues that realist folklore promotes alliance making as a means through which a threatened state can bolster its power. When one state forms an alliance with a third state, this then threatens the security of its counterpart, prompting the latter to form a counter-alliance with an additional, outside state. In short, alliance-building exacerbates the security dilemma, which raises the likelihood of war.
If the above argument is true, then alliances increase the probability of war; numerous studies uncover support for this conclusion. Senese and Vasquez (2008), for example, find that when two states in conflict have one or more external allies, the likelihood of war increases. Similarly, Kang (2017) concludes that alliances embolden their members; states with strong allies are more likely to initiate militarized disputes against non-allied states, especially when those inside the alliance share high levels of common security interests (i.e., they are highly motivated to assist one another during any potential dispute).
In contrast to the steps-to-war theory’s prediction, other scholars argue that alliances deter conflict instead. Leeds (2003), for example, proposes that alliance commitments vary tremendously, and these commitments condition the effect of alliances on conflict. After disaggregating alliances into commitment types—for example, defensive, offensive, and neutrality pacts—she finds that target states with defensive alliances experience less militarized conflict by deterring potential challengers, while offensive and neutrality pacts with potential challengers increase the likelihood of militarized conflict by emboldening alliance members to attack targets.9 Because a war cannot occur without a militarized dispute happening first, these findings suggest that alliances can therefore either increase or decrease the likelihood of war, depending on the exact commitment involved.
Benson (2011) reaches a similar conclusion. He argues that alliances designed to deter specific challengers from initiating a dispute generally reduce the likelihood of militarized conflict, while those designed to deter unspecified threats produce the opposite effect. Moreover, alliances in which members agree to assist one another regardless of who initiates a dispute significantly increase the likelihood that militarized conflict will follow. Specificity (or the lack thereof) therefore ultimately determines how alliances affect conflict behavior. Finally, Johnson and Leeds (2011) directly test whether defensive alliances deter (through an alliance with a potential target) or embolden (through an alliance with potential challengers). They find consistent support for a deterrence effect, but none for the emboldening effect; defensive alliances do not seem to encourage challengers to initiate disputes. Thus, it seems defensive pacts may promote peace, contrary to what the steps-to-war theory predicts (see also Reed, 2000).
As this discussion demonstrates, scholars do not yet agree on the role alliances play in interstate conflict processes. Three factors account for this lack of consensus. First, alliances are not monolithic. They vary tremendously in the commitments they impose on signatory states (e.g., the specificity, asymmetry, and type of commitments), and different commitments affect conflict behavior in distinct ways (e.g., securing the neutrality of a state may encourage a challenger to attack a target, while the target’s alliance structure may deter a potential challenger). Indeed, as scholars theorize more about alliance types (Benson, 2011; Leeds, 2003; Smith, 1995), they discover that some alliances promote peace, while others promote conflict. This nuance remains outside the steps-to-war theory, which instead proposes that all outside alliances increase the likelihood of war (Senese & Vasquez, 2008).
Despite its silence on alliance nuances, the steps-to-war theory can incorporate it. Vasquez (1993, 2009), for example, argues that a world war begins as a dyadic war that then expands via alliance ties. Leeds (2005) confirms this proposition more broadly. Defense pacts deter the onset of militarized interstate disputes, but if this deterrence fails, defense pacts pull other states into the dispute. Thus, even if certain alliances—for example, defense pacts—do not facilitate war onset, they may affect war by raising the likelihood that any war that occurs will involve more actors, possess greater severity, or persist for a longer duration. In these cases, defense pacts increase the probability of a certain type of war (for preliminary evidence on this point, see Valeriano and Vasquez, 2010). One might therefore theorize about the intersection of alliance types and war types to determine the conditions under which and the effects that alliances exert on war (on domain specific theories, see Most & Starr, 1989).
Second, the empirical effects of alliances vary over time. Levy’s (1981) analysis of alliances among great powers during the period 1495–1975 highlights this variation in detail, although it appears elsewhere too. For example, even though Senese and Vasquez (2008) find that two states that have one or more external alliances are more likely to experience a war with one another, this effect seems to be driven by the period 1816–1945. The relationship may disappear in the period 1990–2001 and run in the opposite direction during the period 1945–1989—that is, external alliances often reduce the likelihood of war during the Cold War (see Leeds, 2003; Reed, 2000). Kenwick, Vasquez, and Powers (2015) provide further evidence on this score. Using a matching technique, they examine the years immediately preceding and following the formation of a defensive alliance to ascertain whether these alliances deter as others expect (Johnson & Leeds, 2011; Leeds, 2003). The findings show that defensive pacts deter militarized conflict only in the period 1946–2000 and never deter war during the longer 1816–2000 period. In fact, defensive pacts increase the likelihood of both militarized conflict and war in the period 1816–1945. This leads them to conclude that their “analysis implies that the relationship between alliance formation and international conflict may hinge critically on the international context in which the alliance is formed” (Kenwick et al., 2015, p. 953). If true, then the international context may condition the steps-to-war theory’s logic, and a future refinement of the theory might consider more thoroughly why that occurs.
Finally, the alliance–war relationship may be obscured empirically. As Levy (1981, p. 611) notes, “The critical point is that nearly all alliances are formed in anticipation of some probability of war. This fear of war is the primary motivating factor underlying nearly all alliance formation . . .” In other words, alliance-making probably results from the realist logic that the steps-to-war theory identifies: actors determine that they are somewhere on the road to war and therefore engage in behaviors prescribed by realism—like alliance making—to ensure their security. Yet expecting some (non-zero) probability of war does not indicate whether that likelihood changes after or how quickly one should expect war to follow alliance formation. If, for example, a state wanted to start (i.e., expected) a war immediately, an alliance might help them along by securing third-party actors’ neutrality or assistance. War would then quickly follow alliance formation. In contrast, if a state feared (i.e., expected) it might be attacked in a war at some unknown time in the future (as members of the North Atlantic Treaty Organization feared at the time of the organization’s founding), securing a defensive pact may or may not deter that future war, depending on if and when a challenger attacked. Moreover, states may possess different thresholds at which they think the probability of war is sufficiently high enough to warrant the search for formal alliance partners. If many states possess a low threshold, we could see peace prevail for long periods after alliances form. War might not then quickly follow alliance formation. In short, the empirical effect of alliances on war remains a challenge to locate, as it is often unclear when and where it might appear. Despite this challenge, the steps-to-war theory’s logic seems to underlie alliance formation (Levy, 1981), and the empirical evidence often supports its predictions, even though more theoretical and empirical work remains to be done (see Kang, 2012, for an overview).
Arms Races and War
In addition to forming alliances, realism advises insecure states to build up their arms as well—the fourth step in the steps-to-war theory. Indeed, Sample (1997, p. 14) writes that it is “legitimate to assume that countries have generally relied, in the years before a conflict breaks out, on their own arms policies, rather than on their war allies to determine their security.” An insecure state that builds up its arms, however, is likely to appear threatening to the state from which it is trying to protect itself. Realism then prescribes an arms buildup to the latter state as well, and the two states find themselves in an arms race—i.e., a situation in which two or more states rely upon “military build-ups to compete over divergent preferences concerning the distribution of some stake(s) or the status quo” (Rider, 2013, p. 582). States often intend these arms races to be signals of resolve, expecting them to deter conflict (Rider, 2013). Unfortunately, arms races can theoretically produce the opposite effect—instead increasing the probability of war.
Empirical evidence generally finds a positive relationship between arms races and the onset of war (e.g., see Gibler, Rider, & Hutchison, 2005; Rider, Findley, & Diehl, 2011; Sample, 1997, 2002; Wallace, 1979). As with any research program, however, debates about this relationship exist, particularly with respect to the appropriate way to measure arms races (see Sample, 1997, for an overview of various measures and their performance) and the proper context in which the relationship holds. These latter contexts suggest that arms races foster war in asymmetrically powerful dyads (Sample, 2002), within rivalries specifically (Rider 2013; Rider et al., 2011), and (perhaps) only prior to 1946 (Sample, 2002; but see also Sample, 2012, which suggests no such limitation). Nonetheless, it seems that arms races generally promote war, rather than peace—a finding that aligns with the steps-to-war theory.
Findings Connecting the Steps to One Another
A third body of evidence supports the steps-to-war theory by demonstrating that the individual steps in the theory mutually reinforce one another, as Senese and Vasquez (2008) propose. Once again, the most compelling evidence stems from an investigation of territorial disputes, as the theory appeared before data on territorial issues existed and the theory itself rests heavily on a territorial explanation of conflict. A critical question therefore is: Do territorial disputes foster rivalries, alliances, and arms races? Research suggests an affirmative answer.
Territorial Issues and the Steps to War
Territorial disputes facilitate the formation and maintenance of interstate rivalry—primarily through two characteristics. First, territorial disputes fester. Hensel (1994), for example, shows that disputes over territory are more likely to end in a stalemate than either a compromise outcome or decisive victory. In other words, territorial disputes are difficult to resolve (see also Wiegand, 2011). The salience states attach to territorial issues—whether tangible (e.g., resources the territory contains) or intangible (e.g., the symbolic importance of the territory; see Goddard, 2006)—drives this characteristic. State leaders and their constituents find the possession of territory to be among the most vital interests the state has, and this, in turn, constrains the ability of leaders to offer the concessions necessary to settle territorial disagreements (Allee & Huth, 2006; Wiegand, 2011). Moreover, states may believe that the territory itself will enhance the power of the state that holds it permanently—for example, by giving the possessor access to a strategic military position (e.g., control of straits) or significant economic wealth (e.g., oil). In these cases, each state tries to prevent the other from possessing the territory because the holder may gain power via the territory, which then allows the more powerful state to exploit the weaker state in the future (Powell, 2006; Rider & Owsiak, 2015). States with such a mindset therefore eschew a negotiated territorial settlement, especially if—as with territorial issues—they expect they must live with that agreement for the indefinite future (Fearon, 1998).10
Second, territorial disputes encourage violence. I outlined this general relationship earlier; here, however, I sketch one additional aspect of this relationship—namely, that violent behavior increases in successive disputes and crises. Because territorial disputes recur both more often and more quickly than non-territorial disputes (Hensel, 1994) and violence increases in each successive dispute (Colaresi et al., 2007; Leng, 1983), territorial disputes should produce more violence than non-territorial disputes. These repeated, negative interactions then build a context defined by mutual threat, competition, and enmity—that is, interstate rivalry (Colaresi et al. 2007). It is this rivalry context that makes disputes particularly dangerous, as noted earlier (Colaresi et al., 2007; Tir & Diehl, 2002; Vasquez & Leskiw, 2001). Nonetheless, my point here is that territorial disputes cause this rivalry context, a point confirmed by numerous studies (Colaresi et al., 2007; Mitchell & Thies, 2011; Rider & Owsiak, 2015; Stinnett & Diehl, 2001; Tir & Diehl, 2002; Valeriano, 2012; Vasquez & Leskiw, 2001).
That territorial issues drive rivalry can also be deduced from studies of rivalry termination. Bennett (1996, 1997), for example, defines rivalry termination through the settlement of territorial issues, an indication that he believes such issues to underlie many rivalries. Diehl and Goertz (2012) and Owsiak and Rider (2013) provide more direct evidence on this point, however. The former show that territorial issues increase the likelihood that interstate rivalries persist, rather than end. The latter demonstrate that when a rival dyad settles its interstate border—that is, resolves one critical territorial issue (Vasquez, 1993, 2009)—the dyad’s rivalry is more likely to end. Both findings underscore that many interstate rivalries occur because of an inability to resolve territorial issues—just as the steps-to-war theory proposes.
In contrast to the strong link between territorial issues and rivalry, that between territorial issues and alliances proves more complex and nuanced—largely because of the myriad issues raised earlier (i.e., temporal variation, divergence in alliance terms, and how the alliance links to the disputing dyad). A first cut at investigating the territory-alliance relationship might consider whether these two factors reinforce one another. The evidence needed to evaluate this consideration is often indirect; nonetheless, it permits two broad conclusions. First, alliances form in response to threat—particularly the recurrence of militarized disputes and the existence of common enemies (Gibler & Wolford, 2006). Research shows that disagreements over territory recur more often than those over other issues (Hensel, 1994). Moreover, given that contiguity exerts one of the strongest effects on war onset (Bremer, 1992), enemies are likely to be contiguous and to therefore dispute border territory (Colaresi et al., 2007). One can therefore conclude that territorial threat incentivizes states to search for allies that are not a party to the territorial disagreement.
Second, the logic of the steps-to-war theory also implies that states should not become allies with those involved in territorial disagreements with them. Gibler (1996) confirms this; he finds that many alliance treaties resolve outstanding territorial disagreements, suggesting that alliances between states form when they remove territorial threats from their relationship. Moreover, these alliances exert a pacifying effect on the signatories’ relationship. Territorial threat therefore undermines alliance formation, while the resolution of territorial threats unlocks the pacific potential of alliances that others observe.
A second cut at the territorial-alliance relationship might entertain the idea that territorial issues mediate the effects of alliances on war. For example, because territorial issues are highly salient, it seems theoretically plausible that this salience might override the deterrent effect commonly associated with defensive pacts (see Leeds, 2003). Wright and Rider (2014) investigate this possibility, but they argue that no such mediating effect exists. Alliances form in response to specific threats (Senese & Vasquez, 2008). Consequently, Wright and Rider propose that target states with defensive alliances will still deter challengers from initiating militarized conflict—even when territorial issues are at stake—as the allied states will already have accounted for the issues under dispute when designing their alliance terms.11 Their evidence supports this position, although they admittedly explore the effects of alliances on the likelihood of militarized interstate disputes rather than war.
Despite such a finding, territorial issues might mediate the alliance–war relationship through one final avenue: norms. Kenwick et al. (2015) find that dyads with external allies experience a higher likelihood of war, but only during the period 1816–1945; after World War II, alliances do not affect war onset in a statistically significant way. They posit that this may result from the influence of nuclear weapons. An alternative possibility, however, is that the territorial integrity norm—the understanding that force cannot be used to alter interstate borders (Zacher, 2001)—facilitates peace as well. Zacher (2001) argues that this norm strengthens significantly after World War II. And if it has become less acceptable to use realist prescriptions (i.e., force) in pursuit of territorial ambitions in the post–World War II era, then alliances—even if they exist—will not increase the probability of war in this period. More research is needed, though, before one could attribute such a significant shift in alliance behavior to the territorial integrity norm.
Finally, territorial issues also promote arms races. The best evidence for this conclusion derives from the work of Rider (2009). He argues that high salience prompts states to use a military buildup to signal resolve over a disputed issue (see also Rider, 2013). If the issue also contains high salience for the state’s counterpart—as territorial issues often do (see Hensel et al., 2008)—then the counterpart builds military arms to signal its resolve over the issue as well. Thus, an arms race occurs. Consistent with this argument, Rider finds that territorial issues foster arms races, but non-territorial issues do not (see also Rider, 2013). Additional, indirect evidence in support of this conclusion derives from Rider et al. (2011), who find that after states experience roughly 10 militarized interstate disputes, the likelihood of an arms races increases. Because territorial disputes are most likely to recur (Hensel, 1994), this suggests that territorial issues will foster arms races—if not immediately, then after a history of repeated disputes develops as the steps-to-war theory argues.12
Relationships Between Rivalry, Alliances, and Arms Races
The connection between territorial disagreements and the other individual steps constitutes the most compelling evidence in favor of the steps-to-war theory. Nonetheless, the remaining steps reinforce one another as well—as the steps-to-war theory predicts. First, rivalry and alliances affect one another. The strongest evidence in support of the steps-to-war theory would demonstrate that rivalry produces alliance formation between the rival states and non-rival partners, presumably as a way for these rivals to augment their power relative to their opponent. Evidence on this score is difficult to obtain, largely because studies of alliance formation omit rivalry as an explanatory variable. Nonetheless, we see hints that such a relationship exists. Gibler and Wolford (2006), for example, find that states with a common enemy in a recent militarized dispute are more likely to form an alliance; were a rival state to be this common enemy, then rivalry might yield alliance formation. These authors also note that as states experience more militarized conflict, they are more likely to ally—evidence consistent with the steps-to-war theory’s position that states seek alliance partners when threat mounts. Because rivalry constitutes a threat, it too should produce alliance formation if such results are indicative of a larger behavioral pattern.
Two other relationships between rivalry and alliances are possible, however. The first derives almost exclusively from theory: rivals are less likely to be allied to one another. It would be illogical for those that viewed one another as threats, competitors, and enemies (i.e., rivals) to coordinate their security policies (i.e., to ally). Gibler and Wolford’s (2006) evidence is consistent with this expectation. When two states have recently fought militarily, they are less likely to form an alliance with one another. Models of rivalry formation provide similar findings. An alliance between states frequently reduces the likelihood of those states being rivals, although this relationship often is statistically insignificant (Colaresi et al., 2007; Rider & Owsiak, 2015). A second relationship runs opposite to the one discussed above: alliances might foster rivalry. Valeriano (2012) offers an argument for this relationship and uncovers evidence of its existence. He finds that when dyad members form alliances with states outside the dyad, this can encourage rivalry formation within the dyad. In such cases, states that view themselves as the target of an alliance quickly expect and prepare for negative, hostile interactions. Securing an external alliance thereby heightens perceptions of threat, enmity, and competition (i.e., rivalry forms).
A second connection between steps concerns how rivalries promote arms races. Numerous arms race studies use rivalry as a case-selection device (e.g., see Diehl, 1985; Gibler et al., 2005; Rider, 2009, 2013), arguing that “arms races make sense as a strategy only in the context of long-term military threats, evidenced by” a rivalry (Rider et al., 2011, p. 88). States, in other words, must perceive a long-term threat before significant, multi-year investment in militarization will occur. Such threats define rivalries (Colaresi, et al., 2007; Diehl & Goertz 2000), making it highly unlikely that arms races could occur outside of rivalries.
Research shows that this theoretical argument holds empirical merit too. Rider (2013), for example, explicitly studies whether rival states are more likely to find themselves in arms races. He concludes that they are and notes that this relationship is driven primarily by rivalries fighting over territorial issues (i.e., another step to war). Rider et al. (2011) offer additional confirmation of and nuance to this aggregate finding. They demonstrate not only that arms races are significantly more common in rivalries, but also that the cases of arms races within non-rival dyads generally lack face validity (i.e., seem to be false positives). Arms races, in other words, do not make sense outside of the rivalry context. Moreover, Rider et al. (2011) show that mature rivalries experience arms races, while young rivalries do not. This evidence all aligns with the steps-to-war theory, which proposes that rival states will use arms races as a security-enhancing device when threat persists over a longer time period (see also Diehl, 1985; Rider et al., 2011).
The final connection between steps considers how alliances may affect arms races. The evidence on this score is unclear. Theoretically, one would anticipate that allies (i.e., states with common security interests) do not arms-race (i.e., engage in behavior designed to counter threat from enemies) and vice versa. Studies of arms race onset, however, either find that states allied to one another are more likely to be in an arms race (Rider, 2013), that alliances do not affect arms races (Rider, 2009), or do not explain the connection between the two variables (Rider et al., 2011).13 These diverging results may stem from two sources: research design decisions or nuance. For example, allied states can share a common enemy, in which case we might see allies individually engaged in a military buildup to fight a third state. What looks like an arms race between allies then turns out to be two arms races with a third state external to the alliance. Alternatively, many states might belong to a larger regional alliance (e.g., the Arab League) but nevertheless can still threaten and arm against one another. Future research might investigate this possibility further—to reconcile these disparate findings and clarify the relationship that exists between alliances and arms races.
When one steps back to consider these various strands of evidence as part of a larger picture, it is important to remember that “[w]hat is crucial is not the sequence of the steps, but that they are mutually reinforcing” and that “each [step] increases the risk of war” (Senese & Vasquez 2008, p. 23). Territorial issues function as the trigger of the theoretical argument. It is these territorial issues that come first and then subsequently encourage actors to manage the issue via power politics tools (i.e., the “realist road to war”; Vasquez, 2009). The precise order in which actors use the power politics steps is unspecified, but each step heightens the security dilemma, incentivizes states to take additional steps along the realist road, and raises the likelihood that war will occur. As the above discussion demonstrates, evidence confirms these various claims. First, territorial issues promote the use of power politics. Second, each individual step increases the likelihood of both militarized interstate disputes and war. Finally, the individual steps reinforce one another, suggesting that the adoption of power politics tools moves states down a path that makes future aggression more likely. The cumulative weight of this evidence offers strong support to the steps-to-war theory and its predictions.
Thus far, scholars have tested predictions derived from the steps-to-war theory only during the period 1816–2001. This restriction resulted from data availability on MIDs. Nonetheless, it raises a commonly asked question: Is the steps-to-war theory still relevant in the contemporary world? Recently, MID data also became available for disputes during the period 2001–2010 (Palmer et al., 2015), which allows me to address this question directly.
In Table A1, I recreate Senese and Vasquez’s (2008) analysis of MID escalation to war—albeit for the extended 1816–2010 period. Before discussing these results, I wish to be candid about the decisions that underlie them. First, I rely upon the dyadic MID as the unit of analysis (Palmer et al., 2015)—analyzing only dyads in which both members were involved in each MID on its first day (i.e., originator dyads, which excludes dyads that join a dispute after it begins). Second, the dependent variable denotes whether each dyad involved in each MID went to war within five years of that MID. Because this dependent variable is dichotomous, I use a logistic regression. The full results of this regression appear in the appendix and produce the estimates in Table A1. Third, various independent variables capture the steps to war. A series of three dichotomous variables track the issues under dispute in each MID, including disagreements over territory, a specific policy, the ruling regime of a state, or some other issue (Palmer et al., 2015). To measure rivalry, I use Thompson and Dreyer’s (2012) strategic rivalry data.14 A rivalry exists when two states perceive one another as threats, enemies, and competitors. Alliance relationships derive from the Correlates of War Project’s alliance dataset (Gibler, 2009). Using these data, I determine for each dyadic MID whether one state had an outside alliance, both states had an outside alliance, or the states are allied to one another and there exists (one or more) outside alliances. Finally, I follow Rider (2013) to measure arms races, which occur when both states increase their military expenditures by 8% for three or more consecutive years. The rivalry, alliance, and arms race variables are all dichotomous.
Table 1. Probability of a MID Going to War Within Five Years, 1816–2010a
+ One with outside alliance
+ Both with outside alliance
+ Allied and outside alliance(s)
+ Both with outside alliance and arms race
a 95% confidence intervals appear below the mean probability estimate in each cell.
Three conclusions derive from the regression results (Table A1) and the probabilities generated from them (Table 1). Each echoes the various predictions, findings, and conclusions discussed earlier.15 First, territorial issues are more war-prone than other issues. Territorial MIDs always display a higher likelihood of going to war than MIDs over policy disagreements—the modal MID in the dataset. For example, absent the other steps to war, the probability that a territorial MID escalates to war is 0.1116—nearly twice as high as for a policy MID (0.0588). When we compare the values in the first column of Table 1 (territorial MIDs) with those in the second column (policy MIDs), we see that territorial MIDs are always about twice as likely to escalate to war than policy MIDs, regardless of which exact steps to war a dyad takes. The steps-to-war theory predicts this outcome. Second, as we move down Table 1—thereby adding steps to war—the probability of war climbs. As noted above, the probability that a territorial MID escalates to war is 0.1116. If we add a rivalry to the disputing dyad, the probability rises to 0.2667. An outside alliances increases it still further to 0.2891. And an arms race pushes the probability even higher—to 0.3479. This pattern appears also for policy disputes, although the probabilities are lower because territorial issues are not at stake. All of this is consistent with the steps-to-war theory.
Finally, in line with earlier work, alliances behave irregularly. One dyad member possessing an outside alliance slightly increases the probability of war, but outside alliance ties to both states may reduce it slightly (see Table 1). These relationships are admittedly difficult to discern, however, as the changes in probability are minuscule (roughly 0.02–0.03). In addition, alliance ties between the dyad members significantly reduce the likelihood of war (see Table 1). We might expect this, as allies should be less likely to war with one another. Nonetheless, their involvement in a MID in the first place suggests that they are willing to threaten, display, or use force against one another despite their alliance status. It therefore seems that alliance relationships are complex. More research is needed to understand which alliance relationships promote peace, which promote war, and whether their effects remain constant over time.
The results presented in Tables A1 & 1 cumulatively demonstrate that the steps-to-war theory remains valid when we analyze data more recent than 2001. Territorial issues prove especially war-prone, and the addition of each of the individual steps to war raises the likelihood that a MID will escalate to war. Given the number of outstanding territorial claims in the world (Hensel et al., 2008) and the borders that remain unsettled (Owsiak et al., 2017), I expect the steps-to-war theory will remain a valid explanation of interstate conflict for the foreseeable future.
Future Directions for the Research Program
Steps-to-war theory represents an important stage in the progression of scholarly research on why war occurs. It advances a novel, territorial explanation of conflict that also synthesizes large bodies of work on many major factors known to affect the outbreak of war. Future work might further contribute to the steps to war research program in four ways. First, scholars might theorize more about and test the sequencing of the steps to war.16 Senese and Vasquez (2008) maintain that sequencing matters little in their argument; yet hints of sequencing do appear in it. For example, territorial issues start the process, which means territorial issues must arise before the other steps occur. Similarly, arms races develop “fairly late in a rivalry” when “most all of the other risk factors are present” (Senese & Vasquez, 2008, p. 245). This implies that rivalry precedes arms races (Rider, 2013) and that arms races appear late in rivalries’ life spans (a point seemingly inconsistent with current research; see Rider et al., 2011).
If territorial issues appear first and arms races last, then the only sequencing possibilities concern rivalry and alliance-building. One might, for example, ask whether the rivalry context prompts a search for allies, or whether securing allies enhances the probability of rivalry (on the latter, see Valeriano, 2012). Similarly, does one state securing an ally encourage its counterpart to do the same, as the steps to war argument proposes (i.e., do counter-alliances form)? When does alliance-seeking behavior begin? What factors—either within or outside the rivalry—promote this behavior? How long does it take before rivalry yields alliance-building or vice versa?
Beyond these questions, one might test the theoretical sequencing predictions of the theory. Do territorial issues begin the process? Under what conditions do they not (e.g., Sample, 2014, finds numerous instances in which territorial issues seem to be a justification for a pre-determined war)? What are the paths by which non-territorial issues lead to war? If power politics provides the vehicle for these latter wars, what determines when states merge on to the realist road to war and when they travel a different path? Preliminary research on sequencing exists (e.g., Sample, 2014, 2017), but as the discussion here demonstrates, many questions remain.
Second, empirical evaluations of the theory show varying effects across different epochs. This appears most clearly with respect to alliance behavior (e.g., conflict promoting in the 19th century, but conflict suppressing during the 20th century). Scholars might therefore theorize more about the functions of alliances, their effects on interstate conflict, and how each of these may change over time. More broadly, we might ask whether and why patterns of territorial contestation, conflict recurrence, and rivalry onset/termination change over time. At first glance, this may seem an unimportant endeavor; yet I would argue the opposite. As Most and Starr (1989) note, many (mid-range) theories will likely be domain-specific. The value of any theory therefore depends on specifying its domain properly.
Third, the steps-to-war theory offers one compelling causal path to interstate war. In doing so, it is clear that there are other paths to war as well. Levy (2012, p. 284) sees “a causal path as a combination and/or sequence of factors that leads, with high probability, to a particular outcome.” Thus, to find the other causal paths, we need to theorize about and search for the combinations and sequences of factors that explain the wars that the steps-to-war theory cannot. Sample’s (2014) work begins shedding light on what these factors might be. She notes, for example, that a significant number of dyadic cases she studies experience a war within one year of the initiation of a territorial claim. The steps-to-war theory cannot fully account for these wars, she argues, because no time existed in which the steps to war could fully occur.17 Two alternatives seem plausible to her: that states anticipate war and pursue policies to enhance the expected war’s success, or that rivalry dynamics account for the war. I find the former unsatisfying, for it leaves us wondering why states expected war. What did they disagree about? Why was there hostility? Even if we assume a realist, Hobbesian world, most states are at peace with most other states most of the time. What prompted violence? The latter suggestion, too, is unsettling. How did the rivalries whose dynamics are driving war begin? What characteristics constitute their “dynamics”? Beyond these, a number of wars are non-territorial (see Colaresi et al., 2007; Dreyer, 2010). What factors explain the path from disagreement over non-territorial issues to war?
Finally, we need to study the “paths to peace” more directly. Many scholars—including those in the steps-to-war tradition—study war to reduce its occurrence or effects. They hope, in other words, to find the keys to a more peaceful world by studying war. This, however, may be an illusion. The world may not be as manipulable as we think. If territorial issues, for example, activate a biological proclivity toward violence, encouraging actors to behave non-violently may be a challenging, if not impossible, task. This would be especially true if realist folklore tells leaders to manage all security threats that arise with power politics. In addition, some scholars suggest that removing the causes of war will not necessarily produce peace (i.e., causal asymmetry exists; see Goertz & Mahoney, 2012). Instead, we might be left with a temporary state of “not war,” which is far from the more permanent, peaceful world most people envision (see also Goertz, Diehl, & Balas, 2016; Owsiak, Diehl, & Goertz, 2017). It may therefore behoove us to spend more time theorizing explicitly about the causes of interstate peace (see also Regan, 2014).
I thank Toby Rider, Doug Atkinson, Josh Jackson, two anonymous reviewers, and the editorial team for helpful comments on earlier drafts of this work.
Table A1. Logistic Regression of Militarized Interstate Disputes Escalating to War Within 5 Years, 1816–2010a
One with outside alliance
Both with outside alliance
Allied and outside alliance
aRobust standard errors presented in parentheses
* p < 0.10
** p < 0.05
*** p < 0.01
- Allee, T. L., & Huth, P. K. (2006). Legitimizing dispute settlement: International legal rulings as domestic political cover. American Political Science Review, 100(2), 219–234.
- Bennett, D. S. (1996). Security, bargaining, and the end of interstate rivalry. International Studies Quarterly, 40(2), 157–183.
- Bennett, D. S. (1997). Measuring rivalry termination, 1816–1992. Journal of Conflict Resolution, 41(2), 227–254.
- Benson, B. V. (2011). Unpacking alliances: Deterrent and compellent alliances and their relationship with conflict, 1816–2000. Journal of Politics, 73(4), 1111–1127.
- Blainey, G. (1988). The causes of war (3d ed.). New York: Free Press.
- Bremer, S. A. (1992). Dangerous dyads: Conditions affecting the likelihood of interstate war, 1816–1965. Journal of Conflict Resolution, 36(2), 309–341.
- Bull, H. (1977). The anarchical society. New York: Columbia University Press.
- Cashman, G. (2014). What causes war? (2d ed.). Lanham, MD: Rowman and Littlefield.
- Clausewitz, C. von. (1968). On war. New York: Penguin.
- Colaresi, M. P., Rasler, K., & Thompson, W. R. (2007). Strategic rivalries in world politics. Cambridge, U.K.: Cambridge University Press.
- Diehl, P. F. (1985). Arms races to war: Testing some empirical linkages. Sociological Quarterly, 26(3), 331–349.
- Diehl, P. F. (1992). What are they fighting for? The importance of issues in international conflict research. Journal of Peace Research, 29(3), 333–344.
- Diehl, P. F., & Goertz, G. (2000). War and peace in international rivalry. Ann Arbor: University of Michigan Press.
- Diehl, P. F., & Goertz, G. (2012). The rivalry process: How rivalries are sustained and terminated. In J. A. Vazquez, (Ed.), What do we know about war? (2d ed., pp. 83–109). Lanham, MD: Rowman and Littlefield.
- Dreyer, D. R. (2010). Issue conflict accumulation and dynamics of strategic rivalry. International Studies Quarterly, 54(3), 779–795.
- Fearon, J. D. (1998). Bargaining, enforcement, and international cooperation. International Organization, 52(2), 269–305.
- Frederick, B. A., Hensel, P. R., & Macaulay, C. (2017). The issue correlates of war territorial claims data, 1816–2001. Journal of Peace Research, 54(1), 99–108.
- Ghosn, F., Palmer, G., & Bremer, S. A. (2004). The MID3 data set, 1993–2001: Procedures, coding rules, and description. Conflict Management and Peace Science, 21(2), 133–154.
- Gibler, D. M. (1996). Alliances that never balance: The territorial settlement treaty. Conflict Management and Peace Science, 15(1), 75–97.
- Gibler, D. M. (2007). Bordering on peace: Democracy, territorial issues, and conflict. International Studies Quarterly, 51(3), 509–532.
- Gibler, D. M. (2009). International military alliances, 1648–2008. Washington, DC: CQ Press.
- Gibler, D. M. (2012). The territorial peace. Cambridge, U.K.: Cambridge University Press.
- Gibler, D. M., Rider, T. J., & Hutchison, M. L. (2005). Taking arms against a sea of troubles: Conventional arms races during periods of rivalry. Journal of Peace Research, 42(2), 131–147.
- Gibler, D. M., & Wolford, S. (2006). Alliances, then democracy: An examination of the relationship between regime type and alliance formation. Journal of Conflict Resolution, 50(1), 129–153.
- Goddard, S. E. (2006). Uncommon ground: Indivisible territory and the politics of legitimacy. International Organization, 60(1), 35–68.
- Goertz, G., & Diehl, P. F. (1992). The empirical importance of enduring rivalries. International Interactions, 18(2), 151–163.
- Goertz, G., Diehl, P. F., & Balas, A. (2016). The puzzle of peace. Oxford: Oxford University Press.
- Goertz, G., & Mahoney, J. (2012). A tale of two cultures. Princeton, NJ: Princeton University Press.
- Hensel, P. R. (1994). One thing leads to another: Recurrent militarized disputes in Latin America, 1816–1986. Journal of Peace Research, 31(3), 281–297.
- Hensel, P. R. (1996). Charting a course to conflict: Territorial issues and interstate conflict, 1816–1992. Conflict Management and Peace Science, 15(1), 43–73.
- Hensel, P. R. (2000). Territory: Theory and evidence on geography and conflict. In J. A. Vasquez (Ed.), What do we know about war? (pp. 57–84). Lanham, MD: Rowman and Littlefield.
- Hensel, P. R., McLaughlin Mitchell, S., Sowers, T. E., & Thyne, C. L. (2008). Bones of Contention: Comparing territorial, maritime, and river issues. Journal of Conflict Resolution, 52(1), 117–143.
- Holsti, K. J. (1991). Peace and war: Arms conflicts and international order, 1648–1989. Cambridge, U.K.: Cambridge University Press.
- Hutchison, M. L. (2011). Territorial threat and the decline of political trust in Africa: A multilevel analysis. Polity, 43(4), 432–461.
- Hutchison, M. L., & Gibler, D. M. (2007). Political tolerance and territorial threat: A cross-national study. Journal of Politics, 69(1), 128–142.
- Huth, P. K. (1996). Standing Your Ground. Ann Arbor: University of Michigan.
- Huth, P. K., & Allee, T. L. (2002). The democratic peace and territorial conflict in the twentieth century. Cambridge, U.K.: Cambridge University Press.
- Johnson, J. C., & Leeds, B. A. (2011). Defense pacts: A prescription for peace? Foreign Policy Analysis, 7(1), 45–65.
- Kang, C.-N. (2012). Alliances: Path to peace or path to war? In J. A. Vasquez (Ed.),What do we know about war? (2d ed., pp. 45–62). Lanham, MD: Rowman and Littlefield.
- Kang, C.-N. (2017). Capability revisited: Ally’s capability and dispute initiation. Conflict Management and Peace Science, forthcoming.
- Kenwick, M. R., Vasquez, J. A., & Powers, M. A. (2015). Do alliances really deter? Journal of Politics, 77(4), 943–954.
- Leeds, B. A. (2003). Do alliances deter aggression? The influence of military alliances on the initiation of militarized interstate disputes. American Journal of Political Science, 47(3), 427–439.
- Leeds, B. A. (2005). Alliances and the expansion and escalation of militarized interstate disputes. In A. Mintz & B. Russett (Eds.), New directions for international relations (pp. 117–134). Lanham, MD: Lexington Books.
- Leng, R. J. (1983). When will they ever learn? Coercive bargaining in recurrent crises. Journal of Conflict Resolution, 27(3), 379–419.
- Levy, J. S. (1981). Alliance formation and war behavior: An analysis of the great powers, 1495–1975. Journal of Conflict Resolution, 25(4), 581–613.
- Levy, J. S. (1988). Domestic politics and war. Journal of Interdisciplinary History, 18(4), 653–673.
- Levy, J. S. (2012). The “paths-to-war” concept. In J. A. Vasquez (Ed.), What do we know about war? (2d ed., pp. 281–290). Lanham, MD: Rowman and Littlefield.
- Levy, J. S., & Thompson, W. R. (2010). Causes of war. West Sussex, U.K.: Wiley-Blackwell.
- Levy, J. S., & John A. Vasquez, J. A. (2014). Introduction: Historians, political scientists, and the causes of the first world war. In J. S. Levy & J. A. Vasquez (Eds.), The Outbreak of the First World War (pp. 3–29). Cambridge, U.K.: Cambridge University Press.
- Maoz, Z., & Mor, B. D. (2002). Bound by struggle. Ann Arbor: University of Michigan Press.
- Miller, S. V. (2013). Territorial disputes and the politics of individual well-being. Journal of Peace Research, 50(6), 677–690.
- Mitchell, S. M., & Thies, C. G. (2011). Issue rivalries. Conflict Management and Peace Science, 28(3), 230–260.
- Most, B. A., & Starr, H. (1989). Inquiry, logic and international politics. Columbia: University of South Carolina Press.
- Owsiak, A. P. (2012). Signing up for peace: International boundary agreements, democracy, and militarized interstate conflict. International Studies Quarterly, 56(1), 51–66.
- Owsiak, A. P. (2017). Foundations for integrating the democratic and territorial peace arguments. Conflict Management and Peace Science, forthcoming.
- Owsiak, A. P., Cuttner, A. K., & Buck, B. (2017). The international border agreements dataset. Conflict Management and Peace Science, forthcoming.
- Owsiak, A. P., Diehl, P. F., & Goertz, G. (2017). Border settlement and the movement toward and from negative peace. Conflict Management and Peace Science, forthcoming.
- Owsiak, A. P., & McLaughlin Mitchell, S. (2017). Conflict management in land, river, and maritime claims. Political Science Research and Methods, forthcoming.
- Owsiak, A. P., & Rider, T. J. (2013). Clearing the hurdle: Border settlement and rivalry termination. Journal of Politics, 75(3), 757–772.
- Palmer, G., D’Orazio, V., Kenwick, M., & Lane, M. (2015). The MID4 dataset, 2002–2010: Procedures, coding rules, and description. Conflict Management and Peace Science, 32(2), 222–242.
- Powell, R. (2006). “War as a Commitment Problem.” International Organization, 60(1), 169–203.
- Putnam, R. D. (1988). Diplomacy and domestic politics: The logic of two-level games. International Organization, 42(3), 427–460.
- Rasler, K. A., & Thompson, W. R. (2006). Contested territory, strategic rivalries, and conflict escalation. International Studies Quarterly, 50(1), 145–167.
- Reed, W. (2000). A unified statistical model of conflict onset and escalation. American Journal of Political Science, 44(1), 84–93.
- Regan, P. M. (2014). Bringing peace back in: Presidential address to the Peace Science Society, 2013. Conflict Management and Peace Science, 31(4), 345–356.
- Rider, T. J. (2009). Understanding arms race onset: Rivalry, threat, and territorial competition. Journal of Politics, 71(2), 693–703.
- Rider, T. J. (2013). Uncertainty, salient stakes, and the causes of conventional arms races. International Studies Quarterly, 57(3), 580–591.
- Rider, T. J., Findley, M. G., & Diehl, P. F. (2011). Just part of the game? arms races, rivalry, and war. Journal of Peace Research, 48(1), 85–100.
- Rider, T. J., & Owsiak, A. P. (2015). Border settlement, commitment problems, and the causes of contiguous rivalry. Journal of Peace Research, 52(4), 508–521.
- Sample, S. G. (1997). Arms races and dispute escalation: Resolving the debate. Journal of Peace Research, 34(1), 7–22.
- Sample, S. G. (2002). The outcomes of military buildups: Minor states vs. major powers. Journal of Peace Research, 39(6), 669–691.
- Sample, S. G. (2012). Arms races: A cause or a symptom? In J. A. Vasquez (Ed.), What do we know about war? (2d ed., pp. 111–138). Lanham, MD: Rowman and Littlefield.
- Sample, S. G. (2014). From territorial claim to war: Timing, causation, and the steps-to-war. International Interactions, 40(2), 270–285.
- Sample, S. G. (2017). Anticipating war? War preparations and the steps-to-war thesis. British Journal of Political Science, forthcoming.
- Sarkees, M. R., & Wayman, F. (2010). Resort to war: 1816–2007. Washington, DC: CQ Press.
- Senese, P. D., & Vasquez, J. A. (2003). A unified explanation of territorial conflict: Testing the impact of sampling bias, 1919–1992. International Studies Quarterly, 47(2), 275–298.
- Senese, P. D., & Vasquez, J. A. (2005). Assessing the steps to war. British Journal of Political Science, 35(4), 607–633.
- Senese, P. D., & Vasquez, J. A. (2008). The steps to war: An empirical study. Princeton, NJ: Princeton University Press.
- Smith, A. (1995). Alliance formation and war. International Studies Quarterly, 39(4), 405–425.
- Stinnett, D. M., & Diehl, P. F. (2001). The path(s) to rivalry: Behavioral and structural explanations of rivalry development. Journal of Politics, 63(3), 717–740.
- Thompson, W. R. (1995). Principal rivalries. Journal of Conflict Resolution, 39(2), 195–223.
- Thompson, W. R., & Dreyer, D. R. (2012). Handbook of international rivalries, 1494–2010. Washington, DC: CQ Press.
- Tir, J., & Diehl, P. F. (2002). Geographic dimensions of enduring rivalries. Political Geography, 21(2), 263–286.
- Toft, M. D. (2014). Territory and war. Journal of Peace Research, 51(2), 185–198.
- Valeriano, B. (2012). Becoming rivals: The process of rivalry development. In J. A. Vasquez (Ed.), What do we know about war?(2d ed., pp. 63–82). Lanham, MD: Rowman and Littlefield.
- Valeriano, B., & Vasquez, J. A. (2010). Identifying and classifying complex interstate wars. International Studies Quarterly, 54(2), 561–582.
- Vasquez, J. A. (1987). The steps to war: Toward a scientific explanation of correlates of war findings. World Politics, 40(1), 108–145.
- Vasquez, J. A. (1993). The war puzzle. Cambridge, U.K.: Cambridge University Press.
- Vasquez, J. A. (1998a). The power of power politics. Cambridge, U.K.: Cambridge University Press.
- Vasquez, J. A. (1998b). The steps to war in Europe, 1933–1941. In F. P. Harvey & B. D. Mor (Eds.), Conflict in world politics: Advances in the study of crisis, war, and peace (pp. 207–240). London: McMillan.
- Vasquez, J. A. (2009). The war puzzle revisited. Cambridge, U.K.: Cambridge University Press.
- Vasquez, J. A., & Gibler, D. M. (2001). The steps to war in Asia, 1931–1945. Security Studies, 10(3), 1–45.
- Vasquez, J. A., & Henehan, M. (2001). Territorial disputes and the probability of war, 1816–1992. Journal of Peace Research, 38(2), 123–138.
- Vasquez, J. A., & Leskiw, C. S. (2001). The origins and war proneness of interstate rivalries. Annual Review of Political Science, 4, 295–316.
- Wallace, M. D. (1979). Arms races and escalation. Journal of Conflict Resolution, 23(1), 3–16.
- Wiegand, K. E. (2011). Enduring territorial disputes. Athens: University of Georgia.
- Wright, T. M., & Rider, T. J. (2014). Disputed territory, defensive alliances, and conflict initiation. Conflict Management and Peace Science, 31(2), 119–144.
- Zacher, M. W. (2001). The territorial integrity norm: International boundaries and the use of force. International Organization, 55(2), 215–250.
1. A war involves sustained combat between the organized armed forces of two or more states, which results in at least 1,000 battle-related fatalities (Sarkees & Wayman, 2010). A militarized interstate dispute (MID) exists when one state threatens, displays, or uses force against another state in the international system (Palmer et al., 2015). The steps-to-war theory—although about war—also incorporates MIDs. I therefore use the terms “war” and “MID” to refer to these specific, respective events, and the term “conflict” to refer to them together.
2. A dyad refers to a pairing of two states; for example, the United States–Canada is one dyad, and the United States–Mexico another.
3. Territorial issues are therefore an underlying cause of interstate conflict, while the realist prescriptions—including repeated militarized disputes (i.e., rivalry), alliance-building, and arms races—are the proximate causes (see Vasquez, 1993, 2009; Senese & Vasquez, 2008).
4. The latter two types of studies are necessarily part of the first, but the converse is not true.
5. Senese and Vasquez (2008) operationalize war in various ways, including as the likelihood that: (a) a dyad ever experiences a war during its entire history, (b) the current MID escalates to war, and (c) the current or any other MID escalates to war within five years of the current MID. Results using these various operationalizations yield conclusions consistent to those discussed throughout this article.
6. These values come from Table 8.2 of Senese and Vasquez (2008, p. 273).
7. Because a threat, display, or use of force precedes war, a MID is a (near) necessary condition for war—the reason why scholars study the escalation of MIDs.
8. Gibler (2007, 2012) proposes that border settlement also affects democratization. After controlling for a border’s status, he argues (and concludes) that the democratic peace is spurious. I do not engage this ongoing debate here. If additional empirical testing confirms this finding (e.g., see Owsiak, 2017), however, the steps-to-war theory may have supplied the foundation for explaining the democratic peace—the closest thing we have in international relations to an empirical law (Levy, 1988).
9. A challenger initiates a dispute against a target. Either party might have an alliance, and this can alter how we expect alliances to affect conflict initiation.
10. Once settled, the territorial integrity norm prevents states from altering borders by force (Zacher, 2001). States therefore expect territorial settlements to persist indefinitely, which causes them to bargain harder during the negotiation process.
11. In other words, territorial issues do not have an exogenous effect on the alliance–war relationship.
12. Note, however, that Rider, Findley, and Diehl (2011) do not directly test this supposition.
13. The alliance variable drops from Rider et al.’s (2011) empirical model of arms race onset. This usually happens when a dichotomous independent variable perfectly predicts one value of the dependent variable. In this case, though, it is unclear whether alliances would perfectly predict arms race onset (Rider, 2013) or the opposite.
14. Counting the number of disputes—as Senese and Vasquez (2008) do—does not change the results presented.
15. Each conclusion is consistent with earlier work by Senese and Vasquez (2008). I do not directly compare these results to their study, however, because: (a) my research design differs slightly from theirs, (b) they find stronger/weaker effects across different time periods (e.g., 1816–1945 shows stronger results than 1946–1989), and (c) they never present probabilities for the entire 1816–2001 period they examine. In this article, I therefore simply wish to demonstrate that evidence in favor of the theory holds when I add more recent data to the analysis.
17. I disagree with Sample’s (2014) conclusion. The steps-to-war theory does not require every step to occur before a war; it simply says that each step taken increases the probability of war further.