Multiparty Disputes and the Probability of War: 1816-1992 Revisions Submitted to Conflict Management and Peace Science 19 March 2004 Karen K. Petersen, Vanderbilt University; John A. Vasquez, Colgate University; and Yijia Wang, Vanderbilt University 1 MULTIPARTY DISPUTES AND THE PROBABILITY OF WAR: 1816-1992 Abstract Previous theory and research have suggested that multiparty disputes might be significantly more likely to escalate to war than bilateral disputes, because of the difficulty of reaching a mutually acceptable agreement as the number of parties increases. This study presents a systematic test of this hypothesis. Efforts to provide such a test have been hampered by the absence of data that distinguish the number of participants in a militarized interstate dispute prior to the outbreak of war from the number of participants after the war breaks out. We find that multiparty disputes do have an increased probability to escalate to war. In addition, we find that the issue over which the disputants contend has an important effect on the probability that the dispute will escalate to war, regardless of whether the dispute involves multiple participants or is simply a bilateral dispute. In order to contribute to future scholarship on this topic, the data for the new classification scheme of multiparty disputes are published in the Appendix. 2 INTRODUCTION Among the most important legacies of Stuart Bremer were his completion of the Militarized Interstate Dispute (MID2.1) data set and the theoretical and methodological changes he helped bring about in light of the possibilities created by these data. His work on MIDs resulted in two key shifts in the intellectual history of peace research. The first was the shift from systemic analyses of why war occurs to an emphasis on what distinguishes militarized disputes that escalate to war from those that do not. The second was the shift from the correlates of war to what factors increase the probability of war. Each of these shifts has made research within the field much more successful and cumulative than it was before Bremer’s work. What distinguishes MIDs that escalate to war from those that do not has been a new theoretical approach within international relations research, and it has given rise to a new research program centered on a particular way of conceptualizing how war arises. Much of the theoretical edifice for this approach was first laid out in Bremer and Cusack (1995), which set out several stages or sequences that depicted the various factors that might affect the process by which war might emerge. The mere fact of looking at war as a process that might emerge from the interaction of states was in itself somewhat revolutionary, given the systemic emphasis of the previous two decades. If we think about the factors that might distinguish MIDs that escalate to war from those that do not, we can see that there are four obvious sets of characteristics that should be examined: first, the attributes of the MID itself—e.g. the kind of issue under contention, such as territory; second, the characteristics of the actors in the disputes, e.g. whether they are neighbors or democracies; third, what these states actually do to each 3 other, e.g. the nature of their bargaining and the history of their prior interactions; and lastly, the systemic context in which the MID occurs, e.g. whether the norms of the system are restrictive or permissive (see Raymond 2000). Much of the research associated with the second set of variables has been framed in terms of dyadic analyses, and here again Bremer (1992) led the way by looking at the attributes that would distinguish the dangerous dyads from the ones that would be less likely to go to war or in some cases “never” or rarely go to war. Of course, some of the variables used in his analysis of dangerous dyads actually included factors that reflected behavior that states had previously taken and not just static attributes—these would include variables like whether states were allied to each other or militarized, and in Bremer (2000) whether they had a history of militarized confrontations. The dyadic analyses brought about by Bremer’s (1992) work have resulted in a major research design breakthrough for peace science in that they have produced a number of important findings, best illustrated by the work on the democratic peace, but also by the work on rivalry (Diehl and Goertz 2000) and on territorial disputes (Hensel 1996; Vasquez and Henehan 2001; Senese and Vasquez 2003). The extensive popularity of dyadic analyses, however, has diverted attention away from characteristics of MIDs that may increase the probability of war that are associated with the dispute itself and not the dyad. In particular, researchers have tended not to think about the various specific characteristics of disputes that might account for its escalation to war. Perhaps the variable that has been most forgotten by dyadic analyses is the number of actors involved in a dispute; i.e. whether an MID involves multiple parties. This is the subject of this study. 4 Previous research, particularly that of Cusack and Eberwein (1982) on a forerunner of the MID data, has found a significant relationship between multiparty disputes and the onset of war. Analyzing the relationship between multiparty disputes and the probability of war, however, is not easily done with the current MID data, primarily because no distinction has been made in the data as to whether the MID becomes multiparty before or after the onset of war. In what is regarded by most as a seminal article in the scientific study of international conflict, Stuart Bremer commented on a data limitation that we correct herein. When laying out the research design for Dangerous Dyads he claimed: If all wars began as one-on-one confrontations, then for each of the 56 interstate wars that began during the period under study there would be one dyad of original participants, but the historical record as not quite so simple. In 13 of the 56 qualifying wars, two or more states became involved in war with one or more other state on the very first day of the war. These may be instances of genuine collusion or very fast joining behavior (I favor the former interpretation), but unfortunately the available historical record does not allow us to distinguish reliably between the two. (Bremer 1992, 321) We do, in fact, believe that distinguishing between collusion and joining is both empirically important and feasible. In our discussion herein dealing with multiparty disputes, collusion refers to instances when three or more states are party to a dispute that leads to war; whereas joining refers to instances when two states go to war and are subsequently joined by one or more additional states. 5 Distinguishing between collusion and joining is important because previous studies have attempted to distinguish between bilateral and multiparty dispute escalation without explicitly discussing the distinction. Cusack and Eberwien (1982: 25) found evidence that the expansion of disputes makes war more likely and conclude that “the entrance of third parties into ongoing serious international disputes has been associated with the escalation of such disputes to war.” Gochman and Maoz analyze Militarized Interstate Disputes (MIDs) from 1816 to 1976 and find that 22 percent of the multiparty disputes escalated to war compared to 5 percent of the dyadic disputes (1984: 602). They posit several reasons for the war-proneness of multiparty disputes. First, they claim that the larger aggregate of capabilities makes the sustenance of full-scale conflict possible. Second, major states tend to intervene more, and major states have greater capacity to sustain hostilities. Finally, they claim that “the decision- making dynamics of multiparty disputes may make it difficult to control the escalatory potential of conflict spirals” (602). Likewise, Vasquez (1993: 190-193) argues that the size of a dispute affects its probability of war because it is more difficult to negotiate among several parties than one and that outcomes from multiparty disputes are more apt to leave one or more parties less satisfied than a bilateral negotiation would. Nonetheless, the empirical evidence discussed above is based on incomplete information. This is so because of the restrictions of the current MID data; it is not possible to make valid claims about the escalation of multiparty disputes because joining and collusion are indistinguishable. The crisis literature also addresses the issue of multiple partic ipants. In his work on international crises, Brecher (1993) finds that “the larger the number of parties in a bargaining sequence, the more difficult it is to attain a solution that will satisfy all 6 concerned. Under such circumstances, there is reason to expect that one or more actors will resort to violence in the escalation phase” (151). Brecher also finds that crises of short duration are more likely to involve only two participants (245) and that crises with multiple actors are more intense and more likely to have a high impact (331). Additionally, Brecher and Wilkenfeld (1997) find that crises with multiple actors are significantly more likely to experience intervention by global organizations than crises with only two actors (858). Likewise, Jame s and Wilkenfeld (1984) find that crises that involve three or more parties tend to be longer and have outcomes that are less satisfactory because the bargaining dynamics are more complicated. While the ICB crisis data is not directly comparable to the MID data, it is encouraging that the findings from the crisis data indicate that there is in fact a difference between bilateral and multiparty disputes/crises. However, within the MID data there is not an explicit distinction between original participants and joiners to the ensuing war (or collusion and joining). Given the existing coding rules, it is not possible to tell whether all the actors enter the MID before or after the MID escalates to war. Since it is clear that a distinction should be made between bilateral and multiparty disputes, we posit that it is important that multiparty disputes be properly classified so as to be able to clearly and confidently state the effect that multiple participants have on escalation. THEORY The conflict literature contains some discussion of the difficulty of conflict resolution when multiple parties are involved in a dispute. Both Gochman and Maoz (1984) and Brecher (1993) briefly address the issue of conflict resolution and how multiple participants might make resolution more difficult. Additionally, in her study of 7 the impact of alliances on conflict behavior, Leeds notes that “… it is reasonable to assume that all else equal, increasing the number of parties that must be satisfied by a bargain to avoid war makes the negotiation of a successful settlement less likely, and war correspondingly more likely” (2003: 8). i Negotiations in situations of militarized conflict, whether mediated through a third party or resolved among the participants to the dispute, require consideration of the issues underlying the dispute in order to understand the nature of the escalation to or termination short of war. ii It then follows that the more participants the less similarity of interests and, perhaps more important, settlement potential will also vary depending on the issue(s) at stake. The inability to reach negotiated settlements inevitably leads to prolonged and protracted conflicts that produce more battle casualties and become “wars” under the Correlates of War operational definition of war. According to Leeds, “escalation to war and longer more costly wars are a result of particularly intractable bargaining problems (Wagner 2000, Filson and Werner 2002, Smith and Stam n.d.)” (2003: 5). With the increase in the number of parties in a negotiation, the likelihood of reaching a unanimous agreement will also be correspondingly diminished, leading to conflict escalation. This “intransigence factor” may help explain why the "size factor" facilitates the escalation of militarized disputes to all-out wars. Based on the “intransigence factor,” the international conflict literature, and the business dispute resolution literature, we develop an issuebased explanation for the war-proneness of multiparty disputes. Issue space is an important component of our theory of the war-proneness of multiparty disputes. Previous research has focused on the role of issues in militarized conflict, but not jointly on the role of the size of the dispute (or number of participants) 8 and the issue at stake. For example, Vasquez (1993) lays out the theoretical foundation for an analysis of conflict through the lens of an issue-based paradigm (see also Mansbach and Vasquez 1981); and Holsti (1991) examines armed conflicts from 1648 to 1989 using an issue-based approach. Issues are important because the common ground available to participants may vary by issue type. Some issues may be easier to resolve simply because states may be more willing to compromise over an issue that does not threaten their existence, for example. If issues matter, then the resolution of a dispute prior to the outbreak of war requires that the participants find some common ground that will satisfy (or at least “satisfice”) the participants. iii The reduction in common ground, or more technically the reduction in overlapping issue positions, means that what would have been acceptable to the two parties might not be acceptable to additional parties, which thereby reduces the number of acceptable proposals. In this manner, proposals that would resolve the issue are reduced to the least common denominator, especially since any party to the dispute can often exercise an informal veto over settlement. In multiparty disputes, all are held hostage to the actor that is most intransigent or least able to sign an agreement, so long as all seek to remain within this coalition. Therefore, disputes involving more than two states will be less likely to be resolved prior to the escalation to war (if in fact the addition of third parties reduces the common ground) and thus more likely to lead to war than disputes between two parties. Two historical examples show that such a dynamic can be at play in the real world. Both of these examples are cases involving actors that are willing to settle, but do not because of the issue positions of their coalition partners. So they end up going to 9 war. In the events leading up to the Crimean War, for instance, Napoleon III of France, initiated the crisis by placing demands on the Sultan of the Ottoman Empire that he be granted certain rights to protect Christians of the Latin rite. This in turn raised the ire of the Russian Tsar. Britain was brought into the crisis because it is aligned with France. Later, when Napoleon III was willing to settle because his basic demands were met, he does not because Britain is unwilling to settle, and he ends up going to war (see Richardson 1994: Ch. 5). Similarly, during the negotiations between Japan and the U.S. in the fall of 1941, Franklin D. Roosevelt developed a modus vivendi proposal that would have loosened the oil embargo placed on Japan in return for partial withdrawal of Japanese troops from Indochina. Churchill and Chiang Kai-shek objected and the proposal was never offered (Iriye 1987: 180-81) with the attack on Pearl Harbor being the result. Here, we have an example of a proposal that FDR thought would have delayed war (and it probably would have (Richardson 1994: 190)) being put aside, in part because of objections of other members of the coalition. Another factor that makes it more difficult to resolve multiparty disputes is that, with intractable issues, it is often necessary to frame the issue in a radically different way to reach a mutually satisfactory agreement (see Burton 1990; Kriesberg 2003). Such reframing, especially of intractable issues, is required to move actors from zero-sum proposals to win- win proposals, or as Figure 1 shows, from the quadrant with the distribution of proposed costs and benefits that is unequal to the quadrant where the distribution is equal. Getting two actors to do this is difficult enough; getting three or five to do it is obviously a more arduous task, especially since each actor may have 10 slightly different "interests" at stake and, more importantly, different domestic constituencies (or audiences) that may prohibit it from signing on to certain classes of proposals (see Roy 1997). (Figure 1 about here) All of these factors lead us to believe that multiparty disputes, ceteris paribus, will be more difficult to resolve than dyadic disputes. Disputes that are difficult to resolve tend to repeat if they are salient (both in terms of their importance to decision makers and public awareness of the issue). Empirical research has sho wn that disputes that recur have a higher probability of going to war (Hensel 1994). Therefore, if multiparty disputes are intrinsically more difficult to resolve and if they are salient, and therefore recur, it follows that they should have a higher probability of escalating to war than dyadic disputes. It follows from the above that while multiparty disputes may be generally more difficult to resolve, the key to whether they will escalate to war depends on their salience. Salience also impacts their tendency to recur, thus making resolution even less likely. The territorial explanation of war (Vasquez 1993: Ch. 4) maintains that territorial issues are potentially the most salient and most likely to go to war; and indeed, a growing body of empirical work has found militarized disputes over territory to be the most war prone disputes (Ben-Yehuda 1997, 2000; Hensel 1996; Huth 1996; Senese 1996; Vasquez and Henehan 2001; Senese and Vasquez 2003). Vasquez (1993: Ch. 4) sees territorial issues as an important underlying cause of war rather than a proximate cause of war because they do not directly “cause” war but rather set off a train of events that can culminate in war. Whether they do or not depends on how they are handled. If they are handled 11 through the use of power politics, then the probability of war increases greatly. Power politics is seen as the more proximate "cause" of war (see also Vasquez and Hene han 2001: 123). In addition, Hensel (1994) shows that territorial disputes are apt to recur, and the rivalry literature shows that recurring disputes are associated with having wars (Diehl and Goertz 2000). Given these findings, while we would expect multiparty disputes in general to escalate to war, we would expect multiparty territorial disputes to be particularly prone to war. In summary, we have presented a theoretical rationale for the war-proneness of multiparty disputes, and based on that, we examine the effect of a contentious issue (territory) when multiple parties are present with the propensity for dispute escalation. Any one of those factors alone should increase the probability of war, but disputes that combine territoria l issues and multiple parties should be especially prone to escalation. In order to test our positions, we created a rigorous system for classifying disputes as bilateral or multiparty thereby contributing to the existing MID data set. It is our contention that other analyses of the war-proneness of multiparty disputes are less than optimal in that they do not distinguish between joining and genuine collusion. Our analysis improves upon previous work by incorporating more rigorous criteria for the classification of multiparty wars and by incorporating the role of issues into an analysis of the probability that a militarized dispute will escalate to war. 12 RESEARCH DESIGN We are basing our analyses on theoretical premises that claim that multiple parties involved in a dispute make dispute resolution less likely and that territorial disputes are more prone to escalation. Therefore, proposition 1 states: 1) Disputes with multiple participants will escalate to war more frequently than expected by chance. If multiparty disputes are as difficult to resolve as we suspect, then it would be expected that multiparty disputes escalate to war more frequently than bilateral disputes as well as more frequently than would be expected by chance. Therefore, proposition 2 states: 2) Disputes with multiple participant are more likely to escalate to war than are disputes between only two parties. Additionally, if disputes over territory are more war prone in general, then multiparty disputes that involve territory should be more war prone than multiparty disputes over other issues. Therefore, proposition 3 states: 3) Multiparty disputes that involve issues of territory are more likely to escalate than multiparty disputes over other issues. Our propositions necessitate the use of the MID as the unit of analysis. Therefore, there is one observation per militarized dispute iv ; and the temporal domain is 1816 through 1992 for which we employ the Militarized Interstate Dispute Data and the Correlates of War Interstate War Data (see Jones et al. 1996 for MID coding information, and Singer and Small 1982, for more information on the COW Interstate War data). Dispute size refers to the number of participants (nation-states) in a MID. A war is a MID that involves a total of 1,000 or more battle fatalities across all of the participants 13 (Small and Singer 1982: 55). To be considered a war participant, a state must either suffer a minimum of 100 fatalities or commit at least 1,000 troops to the effort. The main problem in testing our propositions is that the MID data do not contain a variable representing the dispute size as discussed above; therefore, we created the variable by comparing the start dates for MIDs with multiple participants that escalate to war with the start dates for these wars in the COW Interstate War data set. v MIDs with more than two participants that escalate to war are problematic because the data do not distinguish whether the participants joined before or after the MID escalated to war. In order to classify MIDs as multiparty or bilateral, the date of entry in the MID and the date that the war began (according to the COW Interstate War Data) for all of the MIDs that escalated to a hostility level of 5 (war) and contained more than two participants were examined. If there were more than two participants in the MID on or before the day that war broke out, then the MID is coded as multipartyvi. If only two of the participants were actually party to the MID prior to the outbreak of war, then the MID is coded as bilateralvii. All MIDs that did not escalate to war, but contained more than two participants are coded as multiparty. Given these coding rules, about 16% of all MIDs are multiparty MIDs. Of these 16%, about 12% escalate to war as opposed to roughly 4% of the bilateral MIDs. viii The primary problem with classifying disputes that have multiple parties without considering when participants join is that it may conflate war joining and war initiation. If start dates are not taken into consideration and all multiparty wars are included regardless of whether the participants were party to the MID prior to the outbreak of war, then about 15% of multiparty disputes escalate to war—more than our findings indicate. ix 14 In addition to the size of the dispute, the issue at stake is also analyzed. The issue is determined based on the MID code for revision type. The revision type represents the goal of the state(s) attempting to alter the status quo and can take on four possible values: territory, regime, policy, and other. All disputes require an explicit claim of some type in order to classify the revision type. For example, in order to be coded as a territorial dispute, the revisionist actor must make explicit a claim to some or all of the territory of the target state prior to the onset of an MID (see Jones et al. 1996: 178 for complete coding rules). Because the MID is the unit is analysis, many of the standard control variables commonly used in the conflict literature cannot be feasibly employed herein. For example, contiguity is a dyadic measure and many of the disputes have three or more participants. The presence or absence of democracy is not something that can be measured for particular disputes, but is instead a measure applied to states or pairs of states. Using capabilities as control variable follows the same logic. Therefore, only variables relevant to disputes as a unit of analysis are used in the analyses below. In the following section, we begin with a bivariate analysis of the relationship between dispute size and war, followed by multivariate logistic regression analysis of the same relationship. After establishing the importance of dispute size, we then evaluate the impact of the issue at stake in the dispute as well. FINDINGS We test Proposition 1 by comparing the probability that bilateral and multiparty disputes will escalate to war with the base probability that any given dispute will escalate to war. Table 1 reports the findings of our analysis of Proposition 1. 15 (Table 1 about here) Table 1 clearly demonstrates that multiparty disputes have a higher probability of going to war than expected by chance. Multiparty disputes have more than twice as many observed instances of war than would be expected. In the case of multiparty disputes, the conditional probability is 0.12, compared to an overall base probability of 0.0547; and the Z-score x of 4.850 is statistically significant at the .001 level, indicating that there is a statistically significant difference in the conditional and base probability of war for multiparty disputes. Conversely, bilateral disputes have a conditional probability of war that is less than the base probability (.042) and thus less observed wars than expected. The Z-score for bilateral disputes is negative and statistically significant at the .05 level, indicating that bilateral disputes are less likely to escalate to war than expected by chance. Proposition 2 is evaluated by comparing the conditional probability of war for the two types of disputes. The conditional probability of war for multiparty disputes is about three times greater than the conditional probability of war for bilateral disputes (0.12 versus 0.04, respectively). Only 59 of 1403 bilateral disputes go to war, whereas 32 of 260 multiparty disputes go to war. Therefore, multiparty disputes are significantly more likely to escalate to war than bilateral disputes. We find considerable support for both Proposition 1 and Proposition 2 on the basis of the results presented in Table 1. Proposition 3 compares the probability that disputes with multiple participants will escalate to war by the type of issue at stake. Proposition 3 is tested by comparing the conditional probabilities of war in disputes with multiple participants across issue type. The results are presented in Table 2 below. 16 (Table 2 about here) Proposition 3 claims that multiparty disputes should similarly be affected by issue type, and the results in Table 2 support Proposition 3. Only disputes over territorial issues have a higher conditional probability relative to the base probability (.243 versus .123, respectively). The difference in the conditional and base probability is statistically significant at the .001 level. Policy disputes have a lower conditional probability (.082), and that difference barely misses statistical significance at the .05 level. Comparing the conditional probabilities across categories reveals that multiparty territorial disputes are more than three times more likely to escalate to war than multiparty policy disputes or multiparty regime disputes and more than twice as likely to escalate to war than multiparty “other” disputes xi. It should also be kept in mind that neither the regime nor the “other” disputes reach statistical significance. In addition to a comparison of the conditional and base probabilities, the results from a logistic regression analysis of the effect of multiple participants and issue types on the probability of war are reported in Table 3 below. Estimating a logit model is necessary due to the dichotomous nature of the dependent variable. The reported coefficients are not exponentiated; however, the percentage change in odds are reported. (Table 3 about here) When compared to policy disputes xii, territorial disputes are significantly more likely to lead to war. In fact, the log odds ratio increases by 207%, indicating that the odds of a territorial dispute escalating are more than 200% greater than the odds of a policy dispute escalating. Multiparty disputes are more likely to escalate to war than bilateral disputes, with a percentage change in odds of 232%, even when controlling for 17 issue type. Again, this change in the odds ratio indicates that multiparty disputes are more than 200% more likely to escalate to war than are bilateral disputes. Regime disputes are not statistically significantly different from policy disputes in their escalation propensity. “Other” disputes, while statistically significantly different from policy disputes only account for 1 of the 31 multiparty wars (one should also keep in mind the coding issues related to “other”--see note ix). One interesting way to analyze the results of regression analysis is to look at simulated probabilities. Table 4 lists the simulated probability of war under different conditions generated using CLARIFY (King et al. 2000). xiii (Table 4 about here) The base probability of war from Tables 1 and 2 above is about 5% overall and 12% for multiparty disputes. The simulated probability of escalation to war for a territorial dispute with multiple participants is 20%, four times higher than the overall probability and almost twice as high as the probability of escalation for multiparty disputes as a whole. Bilateral disputes over territory are more than three times as likely to escalate to war then bilateral disputes over policy (.07 versus .02, respectively), and multilateral disputes over territory are almost three times more likely to escalate to war than multiparty disputes over policy issues (.20 and .08, respectively), which supports Proposition 3. xiv CONCLUSIONS We set out to correct the limitations of the MID data when it comes to analyzing multiparty dispute escalation and then empirically confirm what has come to be known as “conventional wisdom” in international conflict studies. We were able to create a 18 measure of multiparty dispute escalation that distinguishes between war joining and collusion and we did confirm that multiparty disputes do escalate to war more frequently than bilateral disputes. However, the rate of escalation is less than previous studies indicated—a disparity that could have an impact on analyses of interstate conflict. In the process of correcting the above problem, we set out to explain why multiparty disputes might be more likely escalate and what role the issue at stake in a dispute might play in the propensity for these disputes to escalate. A set of propositions about the relationship between escalation and the number of partic ipants and the issue at stake followed naturally from our discussion. The findings presented in Table 1 above support our contention that disputes that involve more than two parties are more difficult to resolve short of war. The common ground available to disputants probably shrinks drastically with the addition of each participant, making war more likely as opposed to conflict resolution. Additionally, as Tables 2 through 4 show, disputes over territory are more likely to escalate to war than disputes over other issues, even in the context of a multiparty dispute, which supports our propositions that the issue type at stake should impact the probability of escalation to war for multiparty disputes. It is likely that dispute escalation is not the only phase that is affected by the presence of multiple parties. The resolution of disputes is probably very different when there are multiple parties involved. Thus, our findings suggest that more attention should be given to the study of conflict resolution with particular attention to methods for successfully dealing with multiple participants and the issues over which they contend. Increasing the common ground may not be the only option available. 19 As we mentioned at the outset, the territorial explanation of war centers on territory as a predisposing cause of war. The more proximate cause may be the way in which territorial disputes are handled. If such disputes are being handled in a power politics fashion, then this may be the driving force behind escalation. The hostile spiral that often develops out of such behavior should be expected to be exacerbated when there are more than two participants involved; thus, exploring conflict management techniques that reduce the propensity to employ power politics techniques could go a long way towards ameliorating the most dangerous types of disputes. When combined with methods for softening issue positions and increasing the common ground available to disputants, a reduction in power politics behavior could ha ve an impact on the occurrence of dispute escalation. 20 Bibliography Ben-Yehuda, Hemda. 1997. “Territoriality, Crisis and War: An Examination of the Theory and 20th Century Evidence. ” Paper presented to the International Studies Association, Toronto, 19 March 1997. Ben-Yehuda, Hemda. 2001. “Territoriality, Crisis and War in the Arab-Israel Conflict, 1947-94. ” The Journal of Conflict Studies 21 (Winter): 78-108. Bremer, Stuart A. 1992. “Dangerous Dyads: Conditions Affecting the Likelihood of Interstate War, 1816-1965.” Journal of Conflict Resolution 36 (2): 309-341. Bremer, Stuart, and Thomas Cusack. eds. 1995. The Process of War. Amsterdam: Gordon and Breach. Burton, John. 1990. Conflict: Resolution and Prevention. London: Macmillan Carr, W. 1963. Schleswig-Holstein 1815-48: A Study in National Conflict. Manchester: Manchester University Press. Carr, William. 1991. The Origins of the Wars of German Unification. London: Longman. Cusack, Thomas R. and Wolf Dieter Eberwein. 1982. “Prelude to War: Incidence, Escalation and Intervention in International Disputes, 1900-1976.” International Interactions 9 (1): 9-28. Diehl, Paul F. and Gary Goertz. 2000. War and Peace in International Rivalry. Ann Arbor: University of Michigan Press. 21 Filson, Darren and Suzanne Werner. 2002. "A Bargaining Model of War and Peace: Anticipating the Onset, Duration, and Outcome of War." American Journal of Political Science 46: 819-837. Gochman, Charles S. and Zeev Maoz. 1984. “Militarized Interstate Disputes, 18161976.” Journal of Conflict Resolution 28 (4): 585-615. Hensel, Paul R. 1994. “One Thing Leads to Another: Recurrent Militarized Disputes in Latin America, 1816-1986, ” Journal of Peace Research 31: 281-298. Hensel, Paul R. 1996. “Charting a Course to Conflict: Territorial Issues and Interstate Conflict, 1816-1992.” Conflict Management and Peace Science 15 (1): 43-73. Holsti, K. J. 1991. Peace and War: Armed Conflicts and International Order, 16481989. Cambridge: Cambridge University Press. Huth, Paul K. 1996. Standing Your Ground: Territorial Disputes and International Conflict. Ann Arbor, MI: University of Michigan Press. Institute of International Business Law and Practice. 1980. Multiparty Business Disputes. Paris: ICC Services. Iriye, Akira. 1987. The Origins of the Second World War in Asia and the Pacific. London and New York: Longman. Jones, Daniel M., Stuart. A. Bremer and J. D. Singer. 1996. “Militarized Interstate Disputes, 1816-1992: Rationale, Coding Rules, and Empirical Applications.” Conflict Management and Peace Science 15 (2): 163-213. King, Gary, Jason Wittenberg, and Michael Tomz. 2000. “Making the Most of Statistical Analyses: Improving Interpretation and Presentation.” American Journal of Political Science 44 (2): 341-355. 22 Kriesberg, Louis. 2003. Constructive Conflicts: From Escalation to Resolution, 2nd ed. Lanham, Md.: Rowman and Littlefield. Leeds, Brett Ashley. 2003. "Alliances and the Expansion and Escalation of Militarized Interstate Disputes." Unpublished Manuscript. Rice University. Long, J.Scott, & J. Freese. 2001. Regression Models for Categorical Dependent Variables. College Station, TX: Stata Press Mansbach, Richard W. and John A. Vasquez. 1981. In Search of Theory: A New Paradigm for Global Politics. Columbia, NY: Columbia University Press. Menkel-Meadow, Carrie, Ed. 2001. Mediation. The International Library of Essays in Law and Legal Theory. Dartmouth: Ashgate. Raymond, Gregory. 2000. “International Norms: Normative Orders and Peace.” In What Do We Know about War?, ed. John Vasquez, pp. 281-297. Lanham, MD: Rowman & Littlefield. Richardson, James L. 1994. Crisis Diplomacy: The Great Powers since the MidNineteenth Century. Cambridge: Cambridge University Press. Roy, A. Bikash. 1997. “Intervention across Bisecting Borders. ” Journal of Peace Research 34: 3-14. Sandiford, Keith A. P. 1975. Great Britian and the Schleswig-Holstein Question 184864: a study in diplomacy, politics, and public opinion. Toronto: University of Toronto Press. Senese, Paul D. 1996. “Geographical Proximity and Issue Salience: Their Effects on the Escalation of Militarized Interstate Conflict.” Conflict Management and Peace Science 15 (2): 133-161. 23 Senese, Paul, and John Vasquez. 2003. “A Unified Explanation of Territorial Conflict: Testing the Impact of Sampling Bias, 1919-1992.” International Studies Quarterly 47: 275-98. Simon, Herbert Alexander. 1957. Models of Man: Social and Rational; Mathematical Essays on Rational Human Behavior in a Social Setting. New York: Wiley. Small, Melvin, and J. David Singer. 1982. Resort to Arms: International and Civil Wars, 1816-1980. Beverly Hills, CA: Sage. Smith Alastair and Allan C. Stam. n.d. "Bargaining and the Nature of War." Unpublished Manuscript. New York University and Dartmouth College. Steefel, Lawrence D. 1932. The Schleswig-Holstein Question. Cambridge: Harvard University Press. Toyne, S. M. 1945. The Scandinavians in History. Freeport, N.Y.: Books for Libraries Press. Vasquez, John A. 1993. The War Puzzle. Cambridge: Cambridge University Press. Vasquez, John and Marie T. Henehan. 2001. “Territorial Disputes and the Probability of War, 1816-1992.” Journal of Peace Research 38 (2): 123-138. Vasquez, John and Richard W. Mansbach. 1984. “The Role of Issues in Global Conflict and Cooperation,” British Journal of Political Science 14 (Sept.): 411-33. Wagner, R. Harrison. 2000. “Bargaining and War.” American Journal of Political Science 44: 469-484. 24 25 Appendix A: Multiparty MIDs that Escalate to Warxv Appendix A contains a list of the wars with multiple participants in the MID prior to the outbreak of war. The following information is presented: the MID ID number, the name of the war from the COW War Data, the names of all participants in the MID (those who reach level 5 hostility are in bold), the revision type of the MID, and the multiparty status. Multiparty status indicates whether there were multiple participants in the MID one, three, and five (1, 3, 5) days prior to the outbreak of war. MID ID Number COW War Name 19 Austro-Sardinian 31 Boxer Rebellion 57 Crimean War 88 Franco-Prussian War 89 Franco-Spanish War 111 112 135 Italo-Ethiopian War Italo-Roman War Wars of Italian Unification Franco-Mexican War 175 Roman Republic 180 Russo-Japanese War 194 Second SchleswigHolstein War 196 Franco- Thai War 200 Sinai War 115 MID Participants Italy, Tuscany, Modena, AustriaHungary Japan, France, U.K., Russia, China, U.S., Austria-Hungary, Italy, Germany France, Russia, Italy, U.K., Turkey/Ottoman Empire, AustriaHungary Germany/Prussia, France, Bavaria, Wuerttemburg, Baden Spain, France, Austria-Hungary, Germany, Russia Italy, Ethiopia, U.K. Papal States, Italy, France Italy, Austria-Hungary, France, Germany Mexico, France, U.K., Spain Austria-Hungary, Papal States, France, Two Sicilies Japan, Russia, Korea Germany, Denmark, U.K., AustriaHungary, Saxony, Wuerttemburg, Hesse Grand Ducal, Baden, Hanover France, Thailand, U.K. Egypt, France, Israel, U.K., U.S., U.S.S.R MID Revision Type Multiparty Status Territory 1,3,5 Policy 1,3,5 Policy 1,3,5 Other 1,3 Policy 1,3,5 Territory Territory 1,3,5 1 Territory 1,3,5 Regime 1,3,5 Policy 1,3,5 Territory 1,3,5 Territory 1,3,5 Territory 1,3,5 Policy 1,3,5 26 MID ID Number COW War Name 257 World War I 258 World War II 261 Seven Weeks’ War 375 First SchleswigHolstein War 611 Vietnamese War 1035 Six Day War 1046 Yom Kippur War 1250 First Balkan War 1251 Second Balkan War 1265 Hungarian Allies War 1293 1490 1535 1793 Turko-Cypriot War Sino-Japanese War Second Central American War Palestine War MID Participants Austria-Hungary, Belgium, Bulgaria, France, Germany, Greece, Italy, Japan, Portugal, Rumania, Turkey, U.K., U.S., Russia, Yugoslavia Australia, Belgium, Brazil, Bulgaria, Canada, China, Ethiopia, Finland, France, Germany, Greece, Hungary, Italy, Japan, Mongolia, New Zealand, Norway, Netherlands, Poland, Rumania, South Africa, U.K., U.S., U.S.S.R., Yugoslavia, Spain Austria-Hungary, Baden, Bavaria, Germany, Hanover, HesseElectoral, Hesse Grand Ducal, Italy, Meckelnburg Schwerin, Saxony, Wuerttemburg Denmark, Germany, U.K., Russia, Bavaria, Hanover, Saxony, Sweden Australia, Cambodia, Democratic Republic of Vietnam, Philippines, Republic of Korea, Republic of Vietnam, Thailand, U.S., China, U.S.S.R. Egypt, Israel, Jordan, Syria, Kuwait, Iraq, Saudi Arabia Egypt, Iraq, Israel, Jordan, Saudi Arabia, Syria Bulgaria, Greece, Turkey, Yugoslavia Bulgaria, Greece, Rumania, Turkey, Yugoslavia Czechoslovakia, Hungary, Rumania, France, Italy, U.K., Yugoslavia Cyprus, Turkey, Greece China, Japan, Korea El Salvador, Guatemala, Mexico, U.S. Egypt, Iraq, Israel, Jordan, Lebanon, Syria MID Revision Type Policy Multiparty Status 1,3 Territory 1,3,5 Territory 1,3,5 Policy 1 Regime 1,3,5 Territory 1,3,5 Territory 1,3,5 Territory 1,3,5 Territory 1,3,5 Territory 1,3,5 Policy Regime 1 1,3,5 Policy 1,3,5 Territory 1 27 MID ID Number COW War Name 3725 Ethiopian-Somalian War Sino-Vietnamese War U.K.-Egypt War 3957 Gulf War 2069 3007 MID Participants Cuba, Ethiopia, Somalia China, Democratic Republic of Vietnam, U.S.S.R Egypt, U.K., France Canada, Egypt, France, Iraq, Italy, Saudi Arabia, Syria, U.K., U.S., Jordan, Argentina, Belgium, Bangladesh, Germany, Greece, Morocco, Niger, Netherlands, Pakistan, Senegal, Sierra Leone, Spain, Turkey, Australia, Bahrain, Kuwait, Oman, Qatar, United Arab Emirates MID Revision Type Territory Multiparty Status 1,3 Policy 1,3,5 Policy 1,3,5 Territory 1,3,5 28 Appendix B: Bilateral MIDs that Escalate to War with Subsequent War-Joining Appendix B lists all of the wars with multiple participants where there were only two participants in the MID prior to the outbreak of war (the remaining participants joined after the war began). The following information is presented: the MID ID number, the name of the war from the COW War Data, the names of all participants in the MID (those who reach level 5 hostility are in bold, the original two participants are in italics), and the revision type of the MID. MID ID Number COW War Name 51 Korean War 113 183 Italo-Sicilian War Nomonhan War Third Central American War Lithuanian-Polish War Israeli- Egyptian War Spanish-Chilean War Pacific War Lopez War Ugandan-Tanzanian War 1205 1272 1480 1482 1518 1590 2141 MID Participants Australia, Belgium, Canada, China, Columbia, Ethiopia, France, Greece, New Zealand, Netherlands, Philippines, People’s Republic of Korea, Republic of Korea, Thailand, Turkey, U.K., U.S. Italy, Two Sicilies, France Japan, Mongolia, U.S.S.R Guatemala, Honduras, El Salvador, Nicaragua Lithuania, Poland, Latvia Egypt, Israel, U.S.S.R. Chile, Peru, Spain, Bolivia, Ecuador Bolivia, Chile, Peru Argentina, Brazil, Paraguay Libya, Tanzania, Uganda MID Revision Type Territory Territory Territory Regime Territory Territory Other Territory Other Territoty 29 Figure 1 Benefits Equal win-win Costs Equal Unequal zero sum Unequal 30 TABLE 1: DISPUTE S IZE AND WAR Multiparty Observed # Expected # MID Cond. Dispute? of Wars of Wars Total Prob. (War) No 59 77 1403 Yes 32 14 260 Total 91 91 1663xvi Base probability of war = 91 / 1663 = 0.0547 Z Sig. .042 -2.092 .018 .123 4.850 <.001 31 TABLE 2: ISSUE TYPE AND WAR IN M ULTIPARTY D ISPUTES Issue Observed # Expected # Multiparty Cond. Z p Type of Wars of Wars MIDs Prob. (War) Territory 17 8.62 70 .243 3.057 <.001 Policy 11 16.61 135 .082 -1.450 .074 Regime 3 5.66 46 .065 -1.198 .117 Other 1 1.11 9 .111 -0.110 .456 Total 32 32 260 Base probability of war = 32 / 260 = 0.123 (for multiparty disputes) 32 TABLE 3: LOGISTIC R EGRESSION OF D ISPUTE S IZE AND ISSUE TYPE ON WAR xvii War Unexponentiated Coefficients Robust S. E. Z p > ¦ z¦ [95% conf. Interval] % Change in Odds xviii Multiparty 1.201 .255 4.70 0.000 0.700 to 1.702 232.3 Territory 1.123 .243 4.63 0.000 0.647 to 1.598 207.3 Regime 0.627 .432 1.45 0.147 -0.220 to 1.474 87.1 Other 1.323 .607 2.18 0.029 0.134 to 2.512 275.4 Constant -3.722 .205 -18.17 0.000 -4.124 to –3.321 N=1663, Wald Chi2= 47.08, Prob. > Chi2 = 0.000, Pseudo R2 = 0.066, 94.53% of cases correctly classified 33 TABLE 4: S IMULATED PROBABILITY OF WAR Variable Combination Probability of War Multiparty - Territorial .20 Bilateral - Territorial .07 Multiparty - Policy .08 Bilateral - Policy .02 34 i Leeds does find empirical support for this claim. Using different data than employed herein, Leeds finds that 5.2% of bilateral disputes escalate to war as opposed to 12.5% of multilateral disputes. ii The business dispute literature also addresses the particularly difficult nature of multiparty dispute resolution and the importance of underlying issues. See Menkel-Meadow (2001). iii Satisficing behavior is an alternative to maximizing (or optimizing) behavior. Under conditions of bounded rationality, decision makers will attempt to achieve at least some minimally acceptable goal as opposed to the rational actor striving for utility maximization (Simon 1957). In situations with multiple participants negotiating an agreement, satisficing (compromise) may be the best possible outcome given that it is unlikely that all participants could be satisfied by an agreement. iv The following disputes are excluded from all of the analyses below: MID number 1129 is a war according to War Data, but not according to MID data so it is excluded; MID numbers 320, 324, 397, 518, 3813, 3826 all have only one participant that reached level 5 hostility and for that reason do not appear in the War Data, so they are excluded. Additionally, two disputes have more than one revisionist actor with more than one revision type coded (MID numbers 21 and 3564). These two disputes as well as any dispute where all revisionist actors are coded as non-applicable are excluded from all of the analyses. v In all cases (both with MIDs and Wars) the start year and start month were available. In a few cases (with MIDs), the start day was unknown. For those cases, we simply substituted “1”. We could have used any other day for these cases without impacting our classification. See Appendix A for a list of all multiparty disputes that escalate to a multiparty war and Appendix B for a list of Bilateral MIDs that escalate to a multiparty war. vi Alternatively, we used two other coding schemes. The first required more than two participants in the MID at least three days prior to the outbreak of war, reducing the number of multiparty MIDs from 260 to 256. The second required that more than two participants be in the MID at least five days prior to the outbreak of war, reducing the number of multiparty MIDs from 260 to 253. Neither of these more restrictive coding schemes changed the outcomes of the analyses presented herein appreciably, so results using the least restrictive rule are presented. 35 vii A war can still have multiple participants and not arise from a multiparty MID if states join the war after the first day (see footnote 4 above). There are 10 cases of war-joining behavior. Given our coding scheme, these cases are considered bilateral MIDs that escalate to war (listed in Appendix B). viii All exclusions listed in footnote 2 apply to this description of the data. 260 of the 1663 disputes are classified as multiparty under our guidelines. 32 of these multiparty disputes escalate to war versus 59 of the 1403 bilateral disputes. ix We found that if entry dates are not taken into consideration, then about 15.6% of all multiparty disputes go to war and about 3.5% of all bilateral disputes go to war for the time period 1816-1992. x The Z-score compares the base and conditional probability of going to war. The Z-statistic is calculated as follows: Z = (Pc-Pb) / vPb (1-Pb) / Nc where Pc = conditional probability; Pb = base or unconditional probability; Nc = total number of cases having the condition (multiparty or bilateral), and v indicates square root. Our use follows Bremer (1992). Significance levels based on a two-tailed test. xi There are four wars in the MID data with a revision type coding of “other,” the Franco- Prussian War, the Football War, the Spanish-Peruvian/Chilean War of 1866, and the Lopez War. The Franco-Prussian War is the one case of a multiparty war with a revision type (issue) coding of other. In the case of the Franco Prussian war, Prussia is coded as seeking a revision of type “other”, involving a dispute over succession rights. The revision type is coded based only on explicit claims made publicly by official representatives of the state before the outbreak of war. However, Vasquez and Henehan discuss this issue in more detail with particular emphasis on the territorial elements of the four “other” wars and claim that “None of these cases indicates that there is some ‘fourth’ factor that makes more war probable than territory; in fact three of them suggest that territory played a role in the war, even if it was not explicit in the claims made before the war. Therefore, they should not be treated as a distinct type of issue with their own level of probability” (2001, 129: fn 9). xii xiii Policy is used as the reference category because it is the modal category. CLARIFY is an ado file written for use with STATA in order to make data interpretation and presentation more accessible and reader-friendly. CLARIFY employs simulation to approximate features of the probability distribution (King, et al 2000: 349). For more information, see Michael Tomz, Jason Wittenberg, and Gary King. (1999). CLARIFY: Software for Interpreting and Presenting Statistical 36 Results. Version 1.2.1. Cambridge, MA: Harvard University. June 1. http://gking.harvard.edu. See also, King, et al (2000). xiv In order to subject proposition 3 to further scrutiny, we collapsed policy, regime, and other into a category “non-territorial”, thereby creating a dichotomous measure of issue type (territory/non-territory) and tested the proposition using the multiparty variable and this dichotomous measure. The results (not shown) were supportive of proposition 3. Multiparty territorial disputes have a simulated probability of escalation equal to 0.21, multiparty non-territorial disputes 0.09, bilateral territorial disputes 0.07, bilateral non-territorial disputes 0.03. This model is a better predictor of escalation to war according to the Pearson’s Chi-2 statistic; however, the model presented in Table 3 is better suited to demonstrating the varying effects of the different types of issues as opposed to just classifying the issue at stake as territory or non-territory. xv All exclusions from footnote 2 apply in the appendices. xvi All exclusions from footnote 2 apply to this and the following analyses. xvii We also ran this model with interaction terms for revision type and multiparty (not shown). None of the interaction terms were statistically significant and the coefficients for the multiparty and revision variables remained significant in all cases. The model with the interaction terms provided a slight (.01) increase in the Psuedo R2 . xviii See Long and Freese (2001) or <http://www.indiana.edu/~jsl650/stata/> for more information.