Ethnicity, Civil War and International Disputes


This study explores the role played by ethnicity in civil war in terms of sparking international disputes. Much of the seminal literature has found that states experiencing civil wars are more apt to experience interstate disputes than others; however, comparatively less is understood as to whether different types of civil wars impact interstate relations differently. Expanding on this notion, it is theorized here that ethnic civil wars will be more likely to spur international disputes than non-ethnic civil wars given the presence of a transnational ethnic group. This theory is then empirically tested over the time-period 1945 to 2001, and, while transactional ethnicity is found to share a positive relationship with international dispute initiation, ethnic civil wars do not.

Table of Contents: 


    Beginning in the early 1990s, the state of Yugoslavia began to disintegrate as each of its major provinces sought complete political autonomy. This began with the states of Slovenia and Croatia, but slowly expanded to include the states of Macedonia and Bosnia. During this period, civil conflict often erupted within these seceding territories, most violently within Bosnia and Croatia, as each prominent ethnic group in these regions sought to secure maximum assurances of political rights and representation, while simultaneously fearing repression under the rule of the others. These ethnic struggles, however, did not remain contained within Yugoslavia, as transnational ethnic groups with political power in neighboring state—most notably the Croats and the Serbs—provided aid and assistance to their ethnic brethren abroad (Central Intelligence Agency, 2002). Consequently, intrastate conflict within Bosnia became increasingly intertwined with interstate conflict between Croatia and Serbia in particular. Given the salience of the ethnic component in the Bosnian conflict, this case demonstrates how ethnic civil wars often spark international disputes. Obviously, the Bosnian case may be unique from similar international cases as it occurred during the fall of a larger state, Yugoslavia, and territorial boundaries and national sovereignty were not yet fully established. Nevertheless, it suggests the question, are ethnic conflicts more likely to spark international disputes than non-ethnic conflicts?

    The question presented here is hardly independent from existing research on civil conflict, ethnicity, and international disputes. Several seminal studies on domestic instability have speculated that this phenomenon will be positively related to international dispute initiation. Some from the diversionary body of literature speculate that states with civil unrest will be more prone to initiate an international dispute in order to draw public attention away from civil factionalism and toward a common, international opponent (Gelpi, 1997; Davis, 2002). Others have speculated that external opponents will be more likely to take advantage of a state riddled with unrest by launching an opportunistic attack to attain previously sought benefits (Walt, 1996). Others still have suggested that a state engaged in civil war is more apt to experience an international dispute due to reasoning endogenous to the civil war itself (Gleditsch, Salehyan, & Schultz, 2008).

    In this study, I hope to expand the understanding of this relationship by empirically testing whether an ethnic dimension in a civil war makes a state more likely to initiate an international dispute. In order to perform this test, I first review the seminal literature specific to civil war cause and typology, the possible rationale for dispute initiation given the presence of a civil conflict, and finally the salience of ethnicity in international conflict. In the second section of the study, I develop a theory, fitting the component of ethnic conflict into existing understandings of conflict diffusion and transnational ethnicity. I predict a positive relationship between ethnic civil war and conflict initiation, given transnational ethnic ties. I also predict that scenarios involving ethnic civil wars and transnational ethnic groupings will be more dispute prone than non-ethnic civil wars. Third, I lay out a plan to test this postulation using a logistic regression model that incorporates core components of ethnic conflict combined with factors traditionally used in explaining international conflict onset, such as capabilities ratios and trade dependence. In the fourth section of the study, I discuss the results of this test, and analyze its implications for the presented theory. Ultimately, I find little evidence suggesting that ethnic civil wars impact dispute initiation differently than non-ethnic conflicts, but discover continued support for the positive relationship between transnational ethnicity and international dispute initiation.

    Literature Review

    Explanations of Civil War Onset

    To date, much of the seminal literature on civil wars has focused on the causal mechanisms that bring them about. As these factors may also account for why civil wars may spark international disputes, an understanding of this work is necessary. In order to explain the outbreak of civil wars, many scholars have utilized paradigms focused on grievance, greed, and opportunity. Studies focused on grievances have often sought to explain the onset of civil war through the presence of economic, social, or political grievances directed by a people against a given regime (Humphreys & Weinstein, 2008). More to this end, civil war onset has also been explained through relative deprivation, or the apparent disparity between the conditions of life to which one thinks they are entitled and the actual conditions they believe they are presently capable of attaining (Gurr, 1970). Others have suggested that the presence of a democratic or a strongly authoritarian regime decreases the probability of civil war outbreak, as grievances of the citizenry are either curbed by political participation or suppressed under despotic rule (Gates, Hegre, Jones, & Strand, 2006; Hegre, Elligsen, Gates, & Gleditsch, 2001). There is also evidence to suggest that the presence of highly organized ethnic minority groups who are deprived of political power increases the probability of a civil war, as both grievance and capability are present in such scenarios (Buhaug & Cederman, 2008). These last explanations of civil war onset merge the presence of particular grievances with potential opportunity as ingredients for likely rebellion. Still, other scholars have focused more exclusively along the lines of opportunity, such as Fearon and Latin (2003), who find the presence of a weak state, accompanied by a large population and favorable geography, as the primary factors causing civil war outbreak, as opposed to ethnic or religious characteristics.

    Alternatively, many others have sought to focus on the role played by greed and opportunity in sparking civil wars. These scholars typically contend that the presence of grievance is difficult to quantify, and that the grievances of a general population often fail to coincide with the forces motivating a rebel faction (Collier & Hoeffler, 2004). Works on greed more often view civil war onset as being caused by the drive of rebel groups to attain economic benefits and lootable goods (Collier & Hoeffler, 1998; Collier & Hoeffler, 2004; Keen, 1998; Malone & Nitzschke, 2005). Like studies on grievances, those focusing on the impact of greed have often incorporated opportunity as a potential causal mechanism in spurring civil wars. Collier and Hoeffler (2004) find that the probability of conflict is raised by the availability of finances for rebel groups, measured through primary commodity exports, and is lowered as viable alternatives to rebellion are raised, measured through per capita GDP, GDP growth, and primary education enrollment. Other studies have found this drive for profit to be directed toward natural resource exploitation (Le Billon, 2001; Lujala, Gleditsch, & Gilmore, 2005; Soysa, 2001; Weinstein, 2005). Still, approaches focused on greed have raised a chicken-egg question–that is, whether it is the prospect of economic gains that caused rebel groups to initiate wars, or if it was the initiation of wars that required these groups to seek economic resources (Levy & Thompson, 2010).

    One common characteristic of the studies mentioned above is that all civil wars are evaluated similarly, i.e., no typology is built to evaluate whether particular types of civil wars are inherently different from others. One notable exception to this is Sambanis (2001), who finds that identity wars, or wars fought over religious or ethnic lines, may have different causes than non-identity, or ideological wars. More specifically, the author concludes that identity wars tend to be associated with political grievances rather than economic factors, are negatively correlated with democracy, and tend to occur most often when located in a violent regional neighborhood. These findings are particularly interesting, as they may account for some of the disparities in the existing literature, and suggest that different types of civil wars may be driven by different causal mechanisms. If Sambanis is correct, and civil wars are born of different causes, it may also be the case that they behave differently once begun. That is, ethnic civil wars may behave differently than non-ethnic civil wars once underway.

    Intra-State to Inter-State Warfare: The Internationalization of Civil Wars

    While many scholars have noted the positive relationship between civil unrest and international dispute initiation (Gleditsch, Salehyan, & Schultz, 2008; Woodwell, 2004; Davis, 2002), there is little consensus as to why exactly this is. Some of the most common explanations of how civil conflict leads to international conflict revolve around concepts of opportunism, diversion, and spillover. Opportunism occurs when a state experiencing a civil war is perceived as weak by international actors, who seek to exploit this vulnerability in order to advance their national interests (Walt, 1996; Lake & Rothchild, 1998). For example, if a given state has an outstanding territorial claim against another, and the latter is suffering from internal conflict, the first may see this as an ideal time to threaten to seize that piece of territory, in the hopes of achieving concessions. In other words, states engaging in opportunistic attacks or threats attempt to exploit weakness, presented in the form of internal conflict, in order to secure some form of national interest, be it tangible or intangible.

    Diversion, on the other hand, focuses on international conflict launched by the state experiencing internal conflict toward an international opponent. Diversionary theories are steeped heavily in the “rally round the flag” effect (Mueller, 1973), which claims that domestic support for a standing government increases when an external conflict is launched. That is, by initiating an international dispute, a government sparks an “us versus them” mentality amongst the people, which then increases nationalism and consolidates support under the existing leadership. Therefore, leaders facing domestic unrest are said to be more likely to initiate disputes in an attempt to divert attention away from home and consolidate support (Davies, 2002; Gelpi, 1997; Lake & Rothchild, 1998). Many scholars have claimed that the diversionary model does not apply equally well to all regime types. Gelpi (1997), for example, finds that democratic regimes are more apt to engage in diversionary conflicts than authoritarian regimes. Similarly, Davis (2002) finds that democracies are more apt to initiate diversionary conflicts in the presence of both violent and nonviolent domestic protest, whereas authoritarian regimes typically only respond to violent protests in this manner, as suppression is the preferred means of dealing with non-violent protests. Finally, several studies have found that diversionary conflicts, or at least conflicts initiated by a civil war state, are more common than disputes initiated by an outside state against a state engaged in civil war (Gleditsch, Salehyan, & Schultz, 2008; Trumbore, 2003).

    Intrastate violence is also thought to spur interstate violence through diffusion or spillover. Here, internal violence in one state spreads across borders, heightening international tensions as a consequence. More specifically, rebel groups may station themselves abroad, conduct operations from these bases, and increase the probability of trans-border and interstate conflict (Lake & Rothchild, 1998). In his work on transnational rebels, Salehyan (2007, 2009), finds that the majority of all rebels groups since 1945 have used external bases, and hypothesizes that bases are most likely to be found in weak states, rival states, and states with large refugee communities. While Salehyan attempts to explain the onset and continuation of civil wars, his propositions on transnational rebel base locations are relevant to Lake and Rothchild’s explanation of spillover. Taken together, these works suggest that the greater the probability of transnational rebels are present, the greater should be the probability of spillover and international disputes.

    Regardless of which occurs most often, opportunism, diversion, or spillover, each of these bodies of thought provides theoretical models detailing why the presence of a civil war may increase the probability of a dispute within a given dyad. Like work on the causes of civil war, however, these works often fail to explore whether different types of civil wars behave in an inherently different manner. While some of the studies listed above, such as Davis (2002), do categorize civil wars based on their level of violence, little thought is given to what is actually being fought over. Moreover, in a recent study, Gleditsch, Salehyan, and Schultz (2008), found that the increased risk of interstate disputes stemming from internal war is associated more often with attempts to affect the outcome of the civil war itself, than to the pure pursuit of ulterior motives through opportunism or diversion. This finding lends further credence to the notion that the primary characteristics of the civil war itself may be an influential factor in the decision to initiate conflict. While little work has been done comparing different types of civil wars, there has been an abundance of work characterizing the nature of ethnic conflicts and their implications for international relations.

    Ethnic Conflict in International Relations

    Much of the literature on ethnicity in international relations seeks first to explain why ethnic ties may remain politically salient across borders. To this end, Lake and Rothchild (1998) seek to explain transnational group cohesion by claiming the members of a given ethnic group in one state will be compelled by feelings of solidarity to support their ethnic kin in another. Modifying this perspective slightly, Moore and Davis (1998) contend that transnational ethnic groups show group cohesion under the fundamental assumption they share similar policy preferences. In this way, transnational ethnic groupings are said to be best conceptualized as international alliances. Others have presented evidence that may offer an even more practical source perpetuating this cohesion. Collier and Hoeffler (2004) and Salehyan (2009) note that ethnic rebel groups often receive funding from their ethnic kin abroad. If this is the case, ethnic rebel groups have ample motivation to perpetuate notions of solidarity with their ethnic brethren abroad. Whether motivated by feelings of solidarity, perceived policy goals, or finances, these works suggest that transnational ethnic groups are likely to demonstrate significant degrees of cohesion and mutual support.

    The presence of a transnational ethnic identity has also been utilized in explanations of international dispute initiation. For example, ethnic communities in one state may effectively pressure their political elite to initiate an international dispute in order to support their oppressed ethnic brethren abroad (Saideman, 2002). From the perspective of the political elite, dispute initiation may also become more appealing, particularly in terms of launching an opportunistic attack, given the prospect of support from this domestic ethnic group (Lake & Rothchild, 1998; Saideman, 2002; Trumbore, 2003). Alternatively, from the perception of the state with that oppressed minority, should civil unrest be present, a diversionary act may be more appealing in order to spark an “us versus them” mentality against the rebelling group’s ethnic kin abroad (Lake & Rothchild, 1998; Trumbore, 2003). Each of these studies delineates how the presence of ethnicity may raise the prospect of domestic political support and, in turn, lower the cost of initiating a conflict.

    In addition to identifying why ethnic groups may support their brethren abroad, many studies have sought to explain when this relationship may have the most significant impact on dispute initiation. Most of these explanations seem to revolve around what percentage of the total population an ethnic group comprises in any pair of states, along with the political rights this group enjoys in each state. For example, Moore and Davis (1998) find that scenarios where an ethnic group comprises a persecuted minority in one state, and controls or shares political power in another are particularly prone to international disputes. Expanding on this idea, Saideman (1997, 2002) claims that ethnic groups can most effectively pressure political leaders to support common groups abroad if they comprise a portion of the given leader’s constituency. In each of these works, ethnic cohesion is found to be most internationally influential when a group is a minority, possibly oppressed, in one state, and holds political power in another.

    Based on these propositions, several recent studies have sought to quantitatively test the relationship between transnational ethnicity and dispute initiation. Moore and Davis (1998) find a positive correlation between transnational ethnic groups and interstate disputes, although this study is severely limited by its time frame, which only ranges from 1977 to 1978. On the other hand, Woodwell (2004) examines transnational ethnicity and dispute initiation from 1951 to 1991 in terms of population dynamics. Woodwell categorizes transnational ethnicity into three dyadic categories: situations where an ethnic group comprises a majority in both states; a majority in one state and minority in the other; or a minority in both states. The author finds only transnational ethnic groups with a majority population in at least one state impact dispute initiation positively, explaining that the majority-majority combination spurs ideological competition amongst ethnic elites. While each of these studies provides evidence for the significance of ethnicity in international relations, neither takes into account which direction hostility typically comes from, the state with the ethnic minority, or the ethnic majority. In addressing this, Trumbore (2003) finds that ethnic conflict makes a state more likely to initiate a dispute, and less likely to have a dispute initiated against it.

    While these works provide valuable insights as to the salience of ethnicity in international relations, much is still left to explain in terms of the internationalization of civil wars. In particular, many of these studies focus exclusively on the occurrence of ethnic civil conflict, but fail to determine why. That is, these studies seek to explain whether and how the presence of ethnic rebellion impacts interstate relations, but little is said as to whether this particular type of rebellion differs meaningfully from non-ethnic rebellions given the presence of transnational ethnicity. Given the fact that identity wars may be fundamentally different from non- identity wars (Sambanis, 2000), and given that factors endogenous to a civil war often impact dispute initiation (Gleditsch, Salehyan, & Schultz, 2008), this may be an important distinction to draw, and one missing from the existing literature. Last, data limitations have prevented previous scholars from comparing the behavior of these ethnic civil wars across multiple international systems. As some scholars speculate, the Cold War period saw less international conflict due to its bipolar structure (Senese & Vasquez, 2003), so testing ethnic conflict and international dispute initiation across multiple systems seems pertinent and yet to be extensively tested.


    Multiple studies have found that a civil war or civil conflict within a given pairing of states increases the probability those states will experience a militarized dispute (Gleditsch, Salehyan, & Schultz, 2008; Woodwell, 2004). Nevertheless, as it has been found that ethnic and non-ethnic wars may not have the same causes (Sambanis, 2001), it cannot be assumed that all types of civil wars impact international relations in a similar manner. The presence of an ethnic civil war within a given state might plausibly impact international relations in a fundamentally different way than a non-ethnic civil war would. A defining characteristic of an ethnic conflict is the organization of combatants along ethnic lines. When combined with the presence of a transnational ethnic group, an ethnic conflict may be more likely to lead to interstate conflict than a non-ethnic conflict. In order to understand why this is, one must consider how and when these transnational groups may impact the relations between a given pair of states. Next, it is also important to consider how these ethnic factors modify the decision to initiate an armed conflict, both from the perspective of a state experiencing an ethnic conflict, let’s say state B, and from the perspective of a given state with relations to this state, let’s say state A. With these factors in mind, the differences, if any, between ethnic and non-ethnic civil wars can be better understood in terms of how they impact international relations between states.

    In any ethnic conflict, ethnic divisions become inflamed within the state experiencing the conflict, and these divisions are apt to spread to the ethnic communities of other states. More specifically, communities with the same ethnic identity have been shown to have a tendency toward mutually supporting one another, because of either a perceived sense of ethnic solidarity (Lake & Rothchild, 1998; Saideman, 2002) or a perception of similar policy preferences (Moore & Davis, 1998). For this reason, if a community in state A shares a common ethnic identity with a minority group engaged in civil conflict within state B, it will likely show support for the latter.  Moreover, if this ethnic community is viewed as politically significant to the ruling elite, the group’s affinity for its ethnic kin can be expected to affect state A’s relationship with state B. Determining whether such a community is relevant is not always easy and it cannot be assumed that all ethnic groups are viewed as politically significant by the ruling elite. Nevertheless, it can be suggested that, at least within more highly democratic states, gathering public support is crucial to maintaining an elected leadership position. By extension, if a given ethnic group comprises the majority of the total population within a democratic state, their support can be considered crucial for individuals holding elected offices. Once the ethnic population has reached this size, its policy preferences are likely too important to be ignored, as they hold the ability to remove the political elite from power.

    In authoritarian regimes, political relevance is more difficult to determine. Borrowing from work on authoritarian regimes, while an authoritarian ruler may not be accountable to the public at large, he or she is still subject to the whims of the ruling elite (Bueno de Mesquita, Morrow, Siverson, & Smith, 2004; McGillivray & Smith, 2008). Therefore, if this ethnic group is present within an authoritarian leader’s ruling coalition, its preferences can be deemed politically important. Even if this is not the case, authoritarian rulers may seek the support of a large ethnic community, if they fear that domestic instability may be forthcoming. Returning to Davis (2002), the cost of repressing citizens may plausibly be higher than the cost of consolidating support through external conflict initiation. Taken together, the presence of a majority ethnic community with ties to an ethnic group engaged in civil war abroad will likely be viewed as politically relevant in democratic states, and, under certain conditions, politically relevant within authoritarian states.

    The set of conditions outlined above would likely affect state A’s (or the non-civil war state’s) decision to initiate an armed conflict in two general ways. In the first potential scenario, pressure from the ethnic community with ties to state B may compel the political elite of state A to take some sort of action against state B in order to increase its public support. In other words, in order to appease its ethnic population, the political elite of a given state may seek to defend its embattled ethnic brethren abroad (Saideman, 2002). As an example of this, Saideman (2002) draws upon Turkish support of Bosnian Muslims during the Yugoslavian disintegration of the 1990s. It is claimed that that the Turkish regime would likely have preferred not to support the Bosnian Muslims, given their predisposition to maintaining a secular regime, yet ultimately supported this group due to a general fear of repercussions from the Turkish Muslim community (Saideman, 2002; Brown, 1993). Without the ethnic dimension here, in a scenario without ethnic ties or an ethnic conflict, such mobilized support from the citizenry in state A would be less likely to exist, as no ethnic ties are present, and initiation of an armed conflict would be less likely.

    In an alternative scenario, the political elite of state A may also use the conflict in State B to pursue pre-existing national security or policy goals. In other words, state A would be engaging in opportunism, attacking another state while it is weak in order to obtain some perceived benefits (e.g., capture disputed territory, obtain diplomatic concessions, and/or intangible national security gains), separable from the civil war itself (Gleditsch, Salehyan, & Schultz, 2008). Here again, the presence of a common ethnic community in state A may change the way this decision is calculated. More specifically, the political elite in state A will likely assume that an opportunistic attack against state B will be met with approval from the majority ethnic community. This lowers the perceived cost of initiating an armed conflict as the ordinary benefits sought through opportunism will be compounded by the expected support from the population sharing ethnic ties. For the reasons outlined above, this would be expected to have the greatest effect on primarily democratic states, though authoritarian elites may also use the presence of trans-border ethnic ties as justification at the international level. In either of these two scenarios, the prospect of support from the ethnic communities with transnational ethnic ties to those in conflict in state B lowers the cost of initiating a militarized dispute and increases the probability of one occurring. It can also be assumed that this effect will be most pronounced when the conflict in state B is ethnic in nature, as this should raise the salience of the transnational ethnic connection. In other words, one may not expect the transnational ethnic connection to be quite so relevant if the conflict in state B is primarily drawn along economic lines as opposed to ethnic lines. From this, the following hypotheses are generated.

    H1: A state with an ethnic majority that shares ties with an ethnic minority community in a state engaged in ethnic civil war will be more likely to initiate a dispute with that state than it would given the presence of a non-ethnic civil war.

    H1a: Democratic regimes with ethnic ties to a group engaged in an ethnic civil war will be more likely to initiate a dispute than will autocratic regimes.

    From state B’s perspective, the presence of a common ethnic group will also modify its decision to initiate a conflict with state A.  As it can be suggested that ethnic populations will support their brethren abroad, state B will likely suspect state A of supporting state B’s rebelling group, regardless of whether or not this is truly the case. The presence of a common ethnic group will likely exacerbate the notion that state A is, in some capacity, meddling in state B’s affairs (Moore & Davis, 1998). Under this assumption, state B is likely to perceive state A with greater hostility, and may therefore be more likely to engage in hostile behavior against that state. In non-ethnic conflicts, state B may be less likely to come to this conclusion, as the presence of a transnational ethnic group may seem less closely tied to the civil conflict at hand. Unlike the previous scenarios, the polity type should have little influence here, as public sentiment is not being taken into account in this case.

    Alternatively, even if the political elite of state B do not perceive state A as hostile or a threat, they may attempt to perpetuate this illusion in order to launch a diversionary conflict. Once more, a diversionary conflict is one in which the political elite of a state with internal strife or unrest initiate a conflict with an outside force in order to divert attention to the international level and consolidate support at home through the “rally round the flag” effect (Davis, 2002; Gelpi, 1997; Mueller, 1973). It is expected that diversion will be more common within dyads that have ethnic civil wars and the previously described transnational ethnic dynamic where a majority community in state A shares ethnic ties to a minority community in state B. This is because the ethnic connection many in state A share with the rebels in state B will more effectively bind state A into the “us vs. them” mentality sought after in diversionary conflicts, and, in turn, make the prospect of public support more credible for state B’s leadership (Lake & Rothchild, 1998). In non-ethnic conflicts, this identity-based connection between the rebels and state A would likely be less salient, so the prospect of public support may consequentially be lower and diversion should then be less appealing. Given the existence of a civil war, this effect would likely affect autocratic and democratic regimes equally, as each has been found to engage in diversionary activity given violent political unrest (Davis, 2002). In sum, it can be expected that the presence of ethnic conflict and the majority-minority transnational ethnic dynamic will make all regime types more likely to engage in diversionary disputes, when compared to non-ethnic conflicts. From this the following hypothesis is generated.

    H2: A state engaged in an ethnic civil war will be more apt to initiate an armed dispute with a state whose majority population shares ethnic ties to an ethnic minority in that state than it would given the presence of a non-ethnic civil war.

    A closely related factor that would also affect the decision to enter a dispute would be the presence of transnational rebels and violent spillover. Salehyan (2007, 2009) suggests the presence of transnational rebels complicates negotiations between rebels and the host state by introducing a third actor, besides the civil war state’s government. Based on these findings, it is not unreasonable to suspect that the presence of transnational rebels will also increase the probability of an armed conflict between state A and state B. Salehyan (2007, 2009) suggests transnational rebels are most apt to establish bases in weak states, rival states, and states with large refugee communities; it also seems plausible, however, that such rebels may be drawn to communities with common ethnic ties. This is based on the assumption that trans-border ethnic groups will display some sort of group cohesion or support, and that rebels will likely exploit these connections in order to establish international bases. This does not necessarily require ethnic ties with the majority of state A’s population, nor the consent of the state A government. For this reason, this may be equally apt to occur in the presence of any type of transnational ethnic community, not simply the majority-minority ethnic connection described earlier. The presence of a transnational ethnic community would be more likely to generate an international conflict, as state B may be more likely to pursue rebels across borders, and state A is likely to retaliate. Similarly, state A may send troops to the border to ensure security, in order to prevent great degrees of spillover, in turn generating an armed conflict (Lake & Rothchild, 1998). Rebels drawn by ethnic ties across borders hold the potential to bring armed conflict to these borders, and in turn, spark international disputes. As I suggested, this effect is expected to be greatest during ethnic civil wars, as transnational ethnic ties are seen as being most relevant under these circumstances. In short, the existence of ethnic ties increases the probability that transnational rebels will exist, which further increases the probability of conflict between state A and state B. Upon this logic, the following hypothesis is generated.

    H3:  A dyad with an ethnic civil war and a transnational ethnic community will be more likely to experience an armed dispute than a dyad with a non-ethnic civil war.

    In sum, under particular conditions, the presence of an ethnic civil war within a dyad should make this pair of states more prone to an international dispute than it would otherwise be in the presence of a non-ethnic civil war. The first set of such conditions occurs when a minority group in an ethnic civil war state shares ethnic ties with a majority community in another state. In this scenario, it is suggested that the state with the majority ethnic community will be more likely to initiate a dispute due to pressure from the ethnic communities with ties to their international brethren, a desire of political leaders to support their ethnic brethren, or the lowered cost of opportunism based on the expected support from the majority community. From the perspective of the state experiencing a civil war, dispute initiation is more likely to occur due to perceptions that the outside state is supporting their ethnic brethren who may be engaged in rebellion and the lowered cost of diversion based on these perceptions among the citizenry. As the motivation to initiate a conflict is often difficult to decipher and may be caused by multiple factors, this study does not take into account which of these factors is the primary motivation for dispute initiation (i.e., cases of diversion or opportunism are not differentiated from cases where conflict is initiated due to public pressure or heightened threat perceptions). The second set of conditions does not require transnational ethnic groups to be characterized by a majority community in one state and a minority in the other. Rather, the sheer presence of a transnational ethnic community is expected to increase the probability of transnational rebel activities, and in turn, international conflict. In all likelihood, this will then increase the probability of military operations crossing international borders and in turn sparking international disputes. Therefore, ethnic civil wars are expected to behave in a fundamentally different manner than non-ethnic civil wars on the international level, being more likely to lead to international dispute initiation.

    Research Design

    In testing our hypotheses, it is first important to bear in mind that this is a study seeking to explain the connection between what are two relatively rare phenomena in civil and international relations, civil wars and interstate disputes.  The rarity of these events is further compounded by the fact that this study seeks to differentiate between different types of civil wars, along ethnic and non-ethnic lines. As a consequence, it is necessary to include as many cases of civil war as possible, in an attempt to attain the highest possible degree of explanatory and comparative power. Therefore, to test these hypotheses, the temporal domain of this study will begin at the end of World War II. This is an appropriate starting point, as it constituted the rough beginning of the decolonization movement, changing the international system and bringing with this change new questions over self rule and nationalism and increasing the potential for civil conflict. The data employed will then stretch to as recent a year as possible given data restraints, which is here found to be 2001. Spatially, the study will include all contiguous states present in the international system at the given time. Contiguity is appropriate here, as states within close proximity are most apt to have interest in the outcome of a neighbor’s civil war (Hensel, 2000).

    Finally, as civil wars are evaluated in terms of their impact on interstate relations, the appropriate unit of analysis would be the directed interstate dyad. The incorporation of direction is necessary, as it allows one to determine which state in the dyad initiated the international dispute, and in turn whether a given characteristic of either state is driving this decision (Most & Starr, 1989; Siverson & Starr, 1991; Trumbore, 2003;). This is particularly important here, as different sizes of ethnic communities within each state in a given dyad are expected to influence dyadic relations differently. In other words, direction is necessary to determine whether an ethnic group comprises a minority or a majority in a given state, as this is said to impact the decision over dispute initiation differently. Moreover, as the escalation process from a civil war to a militarized dispute is being analyzed, the presence of a civil war should be incorporated into the base unit of analysis. Therefore, the unit of analysis employed here is theDirected Dyad Civil War Year.

    Dependent Variable

    Militarized Interstate Dispute (MID). The dependant variable MID is used to determine whether or not two states share a dispute in a given year.  According to the Correlates of War project, a MID occurs when high-level officials from a given state threaten, show, or use force against another (Jones, Bremmer, & Singer, 1996). This measure is appropriate as it incorporates only disputes serious enough to reach a level where the use of force is plausible, thus demonstrating a concrete worsening of interstate relations and ruling out purely rhetorical disputes of a less serious nature. Moreover, the alternative of setting a military clash as the threshold for measuring interstate disputes would likely be too high. As stated earlier, political leaders may often escalate interstate tensions as a means of attaining political support from a significant ethnic group in their constituency, but without the intention of directly initiating a military attack. Capturing this phenomenon then requires a measure of dispute short of the use of force. Finally, as this study is focusing on relations between pairs of states, the dyadic Militarized Interstate Dispute database will be utilized (Maoz, 2005). This dataset runs up to 2001, allowing for analysis of both the Cold War and post-Cold War era. The variable “MID” is dichotomous and will be coded “0” indicating there was no dispute present in a given year, and “1” indicating there was.

    Explanatory Variables

    Civil War (CIVWAR). The presence of a civil war is coded using the UCDP/PRIO Armed Conflict Dataset. Here, a civil war is defined as a contested incompatibility that concerns government and/or territory, includes two parties, one of which is a state, and results in at least 1,000 battle-related deaths in a given year (Gleditsch, Wallensteen, Eriksson, Sollenberg, & Strand, 2009). This data spans the entire temporal domain of this study, 1945 to 2001. The presence of a civil war will be coded using the dichotomous variable CIVWAR, which will be coded “0” in the absence of a civil war and “1” in the presence.

    Ethnic Civil War/Non-Ethnic Civil War (ETHCIV/NONETHCIV). In order to differentiate between ethnic and non-ethnic civil wars, the Political Instability Task Force (Marshall, Gurr, & Harff, 2009), State Failure Problem Set data is used to code for the presence of an ethnic civil war. Ethnic civil wars are defined as “episodes of violent conflict between governments and national, ethnic, religious, or other communal minorities (ethnic challengers) in which the challengers seek major changes in their status” (Marshall, Gurr, & Harff, 2009, p. 6). This definition suits this study as it necessitates not only the presence of an ethnic community engaged in civil conflict, but also one in which their status is directly tied to the conflict itself. This ensures that the conflict is fought over ethnicity, and not simply between different ethnic groups over issues apart from their ethnic identities (Sambanis, 2000). In order to be classified as a civil war, the PITF dataset necessitates at least 1,000 battlefield deaths over the course of a conflict, with at least one year containing 100 or more casualties (Marshall, Gurr, & Harff, 2009). As this threshold is lower than the PRIO definition, the PITF dataset will be used to classify PRIO civil wars as either ethnic or non-ethnic. The variable ETHCIV will be dichotomous with “1” indicating the presence of a PRIO civil war coded as ethnic by the State Failure Project, and “0” if it is not. The variable NONETHCIV will function identically for non-ethnic civil wars. If a civil war is coded by PRIO, and not coded as ethnic by PITF, it is coded as non-ethnic.

    Transnational Ethnic Ties (MINMAJ/ETHTIES). In order to characterize transnational ethnic ties, data from Woodwell (2004) will be utilized. This dataset codes the presence of ethnic, religious, or linguistic communities based on the Minorities at Risk Dataset, The C.I.A. World Factbook (2000), Vanhanen’s ethnic/linguistic/religious/racial division data, and the Ethnologue (2000) data. Woodwell then codes this data into interstate dyads based on three descriptive variables, from which two will be created for the purposes of this study. The first is titled minmin, and codes for when there is an ethnic community present on both sides of the dyad, comprising a minority community in each state. In order to constitute a politically relevant minority, this group must comprise at least three percent of the total population, or be coded as a Minority at Risk (Woodwell, 2004). Secondly, the variable minmaj indicates that there is a transnational ethnic group that comprises a majority in one state and a minority in the other. For example, the France-Belgium dyad would be coded minmaj, as the French compose a majority in France, and an ethnic minority in Belgium. Last, the variablemajmaj codes for the presence of an ethnic group that comprises the majority in two states. From these three variables, a new variable, ETHTIES is created, which codes for the presence of any transnational community in any capacity.

    Secondly, the minmaj variable is modified in order to incorporate direction. To do this, the Woodwell (2003) data was referenced with the C.I.A. World Factbook (2010), Minorities at Risk (2008), and Ellingsen (2000) in order to determine which side of the dyad the majority and minority communities were on. From this, the variable MINMAJAcreated and coded “1” if the ethnic majority was located on the state A side of the dyad (the state not experiencing a civil war), as this is the circumstance under which theminmaj variable is hypothesized to impact dispute initiation. Finally, this data is currently only available to the year 1991. As a consequence, transnational ethnic community characteristics are assumed to remain consistent from 1991 until 2001. While not ideal, this is based on the assumption that transnational demographic data will not fluctuate at such a rapid rate.

    Polity (POLITY2). In order to determine regime type, the Polity2 measure from the POLITY IV dataset will be used. POLITY2 ranges from -10, being the most highly autocratic score possible, to 10, being the most democratic (Marshall & Jaggers, 2009). Within the directed dyad, the polity score of the primary state considered will be coded. In terms of this study, the Polity II score will be used to determine whether or not democratic or authoritarian states are more likely to initiate a dispute given the presence of a domestic or foreign civil war.

    Control Variables

    Capabilities Ratio (CAPRATA/CAPRATB/CAPRATIO). In order to gather information on the war-making capabilities of each side of a given dyad, a capabilities ratio is utilized. This measures the power ratio within a dyad through population, iron and steel production, military manpower, energy consumption, and military expenditures (Singer, Bremer, & Stuckey, 1972). This ratio is created in three different ways, each using the composite index of national capabilities score (CINC score) as recorded by the National Material Capabilities dataset hosted by the Correlates of War project (Singer, 1987). The first measure is presented in the variable CAPRATA, which is calculated by dividing the CINC score of state A over the summed CINC score of the dyad. The second measure is calculated similarly and is titled CAPRATB, which measures the CINC score of state B divided by summed dyadic CINC scores. Last, CAPRAT codes for the dyadic capabilities ratio are calculated using the larger of the dyadic CINC scores divided by their sum. These measures account for the probability that some states will be less likely to initiate a militarized dispute against another based on overwhelming power preponderance.

    Cold War (COLDWAR). The variable COLDWAR codes for whether or not an observation occurred during the years of the Cold War. Past studies have indicated the system structure of the Cold War may have dramatically affected violent behavior in international relations (Senese & Vasquez, 2003). All data points before 1990 are said to have occurred during the Cold War. Observations are coded “1” if they occurred during this time and “0” if they did not.

    Trade Dependence. Liberal theory indicates an integrated trade network between two states decreases the possibility those two will initiate a conflict against one another (Russett & Oneal, 2000; Woodwell, 2004). Trade dependence data is gathered from the international trade dataset hosted by the Correlates of War project (Barbieri, Keshk, & Pollins, 2008).  Trade dependence is measured using three different variables:TRADEDEPA, TRADEDEPB, and TRADEDEP. TRADEDEPA measures state A’s trade dependence on state B by calculating state A’s imports from state B over its total state imports. TRADEDEPB is then calculated in the same way to determine state B’s dependence on state A. TRADEDEPA and TRADEDEPB are used to test hypothesis one and two respectively. TRADEDEP determines the mutual trade dependence in the dyad by averaging each state’s dependence on the other, and is used in non-directional models to test hypothesis 3.

    Alliance (ALLY).  The ALLY variable determines whether or not the members of the dyad are allied to each other. Allied states may be less likely to initiate a dispute against one another (Russett & Oneal, 2000). This data can be taken from Gibler (2004). This variable is dichotomous, with “1” indicating the given states are allied, and “0” indicating they are not.


    As the dependent variable presented here, a militarized interstate dispute, is a dichotomous variable, a logistic regression model is appropriate (Long, 1997). Through this model, the relationship and significance level between each independent variable shares with the dependent variable, MID initiation, will be determined. At the core of the theory presented here is the expectation that ethnic civil wars will interact positively with different compositions of transnational ethnicity, and will be more dispute prone than non-ethnic civil wars. Therefore, if the preceding hypotheses are correct, the presence of an ethnic civil war, coded ETHCIV, should have a greater, positive impact on dispute initiation than do non-ethnic civil wars, coded NONETHCIV. This should hold across each direction of the dyad. That is, states engaged in an ethnic civil war should be found to be more likely to initiate a MID than states engaged in an non-ethnic civil war, and outside states should be more likely to target states engaged in ethnic civil wars, compared to non-ethnic civil wars. If the ETHCIV variable is found to be statistically insignificant, or in the wrong direction, these hypotheses will be disproved. Moreover, in accordance with hypotheses 1 and 2, the interaction between the MINMAJvariable and ETHCIV variable should be positive and significant. If this is not found to be the case, these hypotheses will be disproved. Finally, in accordance with hypothesis 3 the MINMIN should share a positive and significant relationship with dispute initiation in non-directional models, and should interact positively with the ETHCIV variable. Insignificance or an unexpected direction among these variables and their interactions will either indicate ethnic civil wars are not, in fact, more likely to lead to MIDs, or that they are for different reasons than those outlined here.

    Analysis and Discussion

    Table one presents the results from the logit regression tests for Hypothesis One in models 1 and 2. As previously stated, hypothesis one claims that the non-civil war state in a civil war dyad should be more likely to initiate an interstate dispute given the presence of an ethnic majority community with ethnic ties to a minority community in a state experiencing an ethnic civil war, than it otherwise would be given the presence of a non-ethnic civil war. In order to test this specifically, an interaction term for the presence of an ethnic civil war and a majority-minority dynamic was included within model 1. This interaction term, the presence of a majority-minority dynamic, and the presence of an ethnic civil war, however, are found to be statistically insignificant within this model. I then constructed Model 2 to test the effect a non-ethnic civil war has on dispute initiation, but here too, the presence of a non-ethnic civil war, along with the presence of an ethnic majority with ties to the civil war state, is statistically insignificant. Model 2 contained no interaction term, as there was no proposed relationship between non-ethnic civil wars and transnational ethnicity. The results from both models suggest that the combination of an ethnic civil war and the presence of an ethnic majority-minority dynamic have no discernible impact on the initiation of a militarized dispute by the non-civil war state. Moreover, both the majority-minority dynamic and the civil war type are found to be individually insignificant in both models. As these factors both fail to interact positively, and fail to attain significance individually, Hypothesis One must be rejected. Whether a civil war in a neighboring state is ethnic, and whether a majority community shares ties with a minority community within this state seem to have no systematic impact on the decision of the bordering state to initiate a militarized dispute.

    This notwithstanding, models 1 and 2 do seem to provide some insight into the salient factors driving the non-civil war state toward dispute initiation. First, within both the ethnic civil war and the non-ethnic civil war model, the presence of ethnic ties remains significant, and positively related to dispute initiation. As previously stated, this variable differs from the majority-minority variable, as it is coded for all dyads with any shared and politically relevant ethnic community. This indicates that, while the civil war type may not be significant from this direction, transnational ethnicity remains a significant force driving dispute initiation against civil war states, reaffirming the findings of previous studies (Trumbore, 2004; Woodwell, 2004). In order to develop the predicted probabilities such effects have on dispute initiation, the program, CLAIRIFY, is utilized to hold all other variables at their mean while fluctuating the presence of a given variable (King, Wittenberg, & Tomz, 2001). The results of such tests are presented in table 3, and here, the presence of transnational ethnicity is shown to raise the probability of dispute initiation by the non-civil war state by roughly 16 percent. As this relationship is significant given the presence of ethnic ties in general, but not for the previously discussed majority-minority dynamic, an alternative explanation of ethnicity and international conflict must be developed. One possible explanation may be that states with common ethnic ties hold more interconnected political relationships, i.e., given the common relationship amongst the peoples of these two states, the electorates may pressure their national representatives to pursue a more involved approach to neighboring states of this type. Moreover, states with a common ethnic majority may feature ethnic leaders with competing nationalist ideologies for the same transnational ethnic group, making such a dyad more dispute prone (Woodwell, 2004). This may explain why the democracy variable shares a positive and significant relationship with dispute initiation from this direction. In any event, these findings seem to reject the notion that the majority-minority dynamic significantly drives states to initiate disputes with states engaged in ethnic or non-ethnic wars.

    Last, models 1 and 2 indicate that the presence of a mutual alliance and increasing trade dependence decrease the probability of dispute initiation from the non-civil war state. As table 3 displays, the presence of an alliance between a civil war state and a non-civil war state modestly reduces the probability of conflict by about four percent. This effect is unsurprising, as allied states, by nature, tend to share a more collaborative relationship than non-allied states, and are therefore, less likely to initiate a dispute against one another. Trade dependence is also found to reduce the probability of conflict, decreasing it by about four percent when adding one standard deviation to the minimum level of trade dependence. The rationale behind this is likely similar to that surrounding alliances. All else being equal, governments would have more to lose by initiating a dispute against a state they depend heavily upon for trade, compared to a state they do not. Taken together, these results indicate a state is less likely to initiate a militarized dispute with a civil war state, given economic dependence and an alliance with the latter.

    Models 3 and 4 in table 1 detail the tests conducted on hypothesis two. Here, it was suggested that states engaged in an ethnic civil war would be more apt to initiate a dispute with a state dominated by an ethnic group with ties to a minority community in the former state, than it would be given the presence of a non-ethnic civil war. Once again, both the presence of a majority ethnic community in a bordering state with ties to a minority group in the civil war state and the type of civil war are found to be statistically insignificant. This evidence seems to undermine the assumptions of hypothesis 2, that states engaged in ethnic civil wars would feel threatened or retaliate against a state with majority community with ties to a domestic minority community. Still, like the previous models, the presence of a common, transnational ethnic group remains positively related to dispute initiation, with the presence of such a community raising the probability of dispute initiation by 12 percent. Compared to the previous hypothesis test, the effect of transnational ethnicity is slightly lessened, and the impact of democracy is no longer significant. This may potentially be due to the fact that a state engaged in a civil war is more concerned about national security interests, so the impact of popular perceptions is lessened. This preoccupation with national security may too diminish the perceived benefit of a diversionary conflict, as the perceived value of diversion is outweighed by the prospect of international conflict along with intrastate conflict (Leeds & Davis, 1997). Regardless, once again, what is being fought over appears to be insignificant, while which ethnicities are involved remains statistically relevant.

    Models 3 and 4 also proved insightful beyond the parameters of ethnicity. First, like the previous models, the presence of an alliance decreases the probability of dispute initiation. Similar to the primarily non-civil war states evaluated in models 1 and 2, a civil war state is found to be about three percent less likely to initiate a dispute against an ally than against a non-ally. Here again, this logic likely parallels that applied to the non-civil war state. Allied states are more likely to share collaborative relationships than non-allied states. For this reason, a civil war state may be less likely to suspect an ally of indirectly supporting a resistance group, and may be somewhat more likely to develop a common response to the said resistance group.

    Unlike the effects of alliances, however, trade dependence does not exert any significant effect on dispute initiation from civil war states, as it did upon non-civil war states. This too may relate to a civil war state’s pre-occupation with national security. From the perspective of a non-civil war state, any perceived benefit of initiating an international dispute is curbed with increasing trade dependence. In a civil war state, however, the very existence of the standing government is being called into uncertainty and conventional concerns of the state, such as trade relations, may be outweighed by what is necessary to conduct warfare against the given rebel group. This may include actions such as moving forces across the bordering regions of the state, where rebel groups are often located, or pursuing these rebels across international borders, each of which may trigger a militarized dispute with neighboring states (Salehyan, 2009). In short, trade dependence may be less significant for civil war states, as these governments are fighting for higher stakes than non-civil war governments, and are therefore more likely to sacrifice lesser goods such as stable relationships with trade partners.

    Next, table 2 details the results of models 3 and 4, which test hypothesis 3. Hypothesis three states that ethnic ties of any kind will positively relate to dispute initiation, as the presence of transnational ethnic ties combined with the presence of an ethnic civil war will increase the likelihood of an international dispute, rebel groups are more apt to coordinate with their transnational kin, establish bases outside the civil war state, and, in turn, increase the probability of transnational military operations and spillover. As this hypothesis does not state the inclination toward dispute initiation would vary from the civil war state to the non-civil war state, it is tested in models 3 and 4 using non-directional models, with the dependent variable being dispute initiation from either side of the dyad. Like the previous tests, these models find no significant relationship between civil war type and transnational ethnicity, nor are civil war types found to be independently significant. Consequently, hypothesis 3 appears to be disproved. Nevertheless, the presence of transnational ethnic ties are still found to be significant, increasing the probability of a militarized dispute by roughly 13 percent. As the majority-minority variable in this model remains insignificant, this may indicate that rebel groups are no more or less likely to establish bases in states with common ethnic majorities.

    Last, it is important to note that, like each of the previous models, alliances are found to decrease the probability of a militarized dispute by roughly three percent. Mutual trade dependence, this time measured as the average dependence each nation shares with the other, is found to be significant, but to a lesser extent than found in the first two models. Moving one standard deviation away from the minimum level of dependence in this measure is found to decrease the probability of conflict by roughly three percent, as demonstrated in table 3. These results are not surprising, as these non-directional models tend to aggregate the effect of the directional models.


    This study sought to determine whether an ethnic dimension of civil conflict had a significant impact on the initiation of international disputes either from or toward that state. Determining an answer to this question has important implications for both the study of civil conflict and the effects of transnational ethnicity in international relations. In terms of the study of civil conflict, this question provides insights as to whether certain types of civil conflict are more apt to become internationalized than others. In terms of the study of ethnicity in international relations, answering this question complements other studies finding a positive relationship between transnational ethnicity and international dispute initiation. That is, this study explores whether the presence of transnational ethnicity compounds with the presence of an internal ethnic struggle, or whether these two phenomena act independently of one another. Moreover, this study seeks to explain whether transnational ethnicity becomes more salient when composed differently, more specifically, whether a state with an ethnic minority will be more apt to defend an ethnic minority community in a state torn by ethnic civil war. In order to answer these questions, several logistic regression models were composed, incorporating the presence of each of these factors, from the presence of an ethnically-based civil war to the presence of different transnational ethnicity compositions, and whether these factors interacted in any significant way.

    Ultimately, it was found that an ethnic component to a civil war had no discernible impact on dispute initiation and failed to interact positively with any form of transnational ethnic grouping. This insignificance notwithstanding, these findings still provide several valuable insights. First, and most obviously, while states neighboring states engulfed by civil war may be more apt to initiate a dispute against the latter for reasons intertwined with the nature of the civil war itself (Gleditsch, Salehyan, & Schultz, 2008), an ethnic component of this civil war does not seem to be amongst these driving forces. Second, works in the past have often perceived transnational ethnic ties as being influential when an ethnic majority in one state shares ties with a minority in the other (Davis & Moore, 2003; Woodwell, 2004), under the assumption that the politically powerful former will seek to defend the status of the latter. Nevertheless, this study finds that, after adjusting for direction, that is, making sure the majority community is present in the state without a civil war, the significance of this dynamic is eliminated. This may suggest that the salience of transnational ethnicity must be reconceived beyond notions of a politically powerful group coming to the aid of their ethnic brethren abroad. Third, as no interaction was found between transnational ethnicity and ethnic civil war, it cannot be assumed that non-ethnic conflicts are less apt to diffuse into international conflicts due to ethnic ties. That is, transnational ethnicity seems to be salient beyond domestic issues primarily ethnic in nature.

    While this work does provide some valuable insights, as covered above, much remains to be studied regarding the issue of ethnic conflict within dispute initiation. First, while the test conducted in this study adjusts for what side of a state pairing an ethnic majority exists on, it fails to code for whether the corresponding minority group is actually engaged in a civil conflict within the civil war state. While it can be assumed that, given an ethnic civil war, most politically relevant ethnic minorities will be involved or subjugated in at least some capacity, this remains an indirect measure of minority uprising. In future work, this can be measured directly to better test the significance of the majority-minority dynamic. Moreover, this work could be expanded beyond the domain of civil war to test whether an ethnic component maintains significance at lower levels of civil disputes. In short, the tests conducted here take important steps toward a better understanding of conflict type and international dispute initiation; there remains much to be understood.

    These caveats notwithstanding, the presence of an ethnic dimension driving a civil war appears to have no significant bearing on international dispute initiation discernable from non-ethnic civil wars. As previously mentioned, this seems to suggest that ethnic conflicts may be no more apt to internationalize themselves than non-ethnic civil wars. Returning to the case of Bosnia, the ethnic dimension of the conflict may not have been the most prominent factor driving Croatian and Serbian leaders to become engaged within that dispute. Instead, it may have been the separable force of transnational ethnicity driving dispute initiation, or related attempts by these states to establish the most preferable geo-political makeup of the region before border and sovereignty disputes had been fully cemented. In any event, while the findings presented here provide progress toward a better understanding of ethnicity and civil war in international relations, they leave much unanswered in terms of how conflict within a state becomes conflict amongst states.  Nevertheless, it would appear as though civil wars with an ethnic dimension are no more or less prone toward international disputes.


    • Barbieri, K., Keshk, O., & Pollins, B. (2008). Correlates of war project trade data set codebook, Version 2.01. Online:
    • Brown, J.F. (1993) Turkey: Back to the Balkans. In G. Fuller, I.O. Lesser, P. Henze, J.F. Brown (Eds.), Turkey’s new geopolitics: From the Balkans to west China (p. 153). Boulder, Colo.: Westview Press.
    • Buhaug, H., Cederman, L., & Rod, J. K. (2008). Disaggregating ethno-nationalist civil wars. International Organization, 62, 531-551.
    • Buhaug, H., & Gleditsch, K. S. (2008). Contagion or confusion? Why conflicts cluster in space. International Studies Quarterly, 52(2), 215-248.
    • Central Intelligence Agency (2002). Office of Russian and European analysis. Balkan battlegrounds. Washington D.C.: Central Intelligence Agency
    • Collier, P., & Hoeffler, A. (1998). On economic causes of civil war. Oxford Economic Papers, 50(4), 563-573.
    • Collier, P., & Hoeffler, A. (2004). Greed and grievance in civil war. Oxford Economic Papers, 56: 563-95.
    • Davies, G. A. (2002). Domestic strife and the initiation of international conflicts. The Journal of Conflict Resolution, 46(5), 672-692. Retrieved June 16, 2010, from
    • De Soysa, I. (2001) Paradise is a bazaar? Greed, creed, grievance, and governance. United Nations University/World Institute for Development Economics Research Discussion Paper 2001/42.
    • Fearon, J. D., & Laitin, D. D. (2003). Ethnicity, insurgency, and civil war. The American Political Science Review97(1), 75-90. Retrieved June 22, 2010, from
    • Gates, S., Hegre, H., Jones, M. P., & Strand, H. (2006). Institutional inconsistency and political instability: Polity duration, 1800-2000. American Journal of Political Science50(4), 893-908.
    • Gelpi, C. (1997). Democratic diversions: Governmental structure and the externalization of domestic conflict. Journal of Conflict Resolution, 41, 255-282.
    • Gibler, D. M., & Sarkees, M. (2004). Measuring alliances: The correlates of war formal interstate alliance data set, 1816-2000. Journal of Peace Research, 41(2), 211-222.
    • Gleditsch, K. S., Salehyan, I., & Schultz, K. (2008). Fighting at home, fighting abroad: How civil wars lead to international disputes. Journal of Conflict Resolution, 52(4), 479-506. Retrieved June 14, 2010, from
    • Gleditsch, N. P., Wallensteen, P., Eriksson, M., Sollenberg, M., & Strand, H. (2002). Armed conflict 1946-2001. Journal of Peace Research, 39(5), 615-637.
    • Gurr, T. R. (1971). Why men rebel. Princeton: Princeton University Press.
    • Hegre, H., Elligsen, T., Gates, S., & Gleditsch, N. P. (2001). Toward a democratic civil peace? Democracy, political change, and civil war 1816-1922. American Political Science Review, 95(1), 33-38.
    • Hensel, P. (2000). Territory: Theory and evidence on geography and conflict. In J. Vasquez (Ed.), What do we know about war? (pp. 57-84). Lanham, MD: Rowman & Littlefield Publishers, Inc.
    • Humphreys, M., & Weinstein, J. M. (2008). Who fights? The determinants of participation in civil war. American Journal of Political Science52(2), 436-55.
    • Jones, D. M., Bremer, S. A., & Singer, J. D. (1996). Militarized interstate disputes, 1816-1992: Rational, coding rules, and empirical patterns. Conflict Management and Peace Science15(2), 163-213.
    • Keen, D. (1998). The Economic functions of violence in civil wars. London: Oxford University Press.
    • King, G., Tomz, M., & Wittenberg, J. (2000). Making the most of statistical analyses: Improving interpretation and presentation. American Journal of Political Science44(2), 341-355.
    • Lake, D. A., & Rothchild, D. (1998). Spreading fear: The genesis of transnational ethnic conflict. In D. A. Lake & D. Rothchild (eds.) The international spread of ethnic conflict (pp. 3-32). Princeton: Princeton University Press.
    • Le Billon, P. (2001). The political ecology of war: Natural resources and armed conflicts. Political Geography 20: 561-84.
    • Leeds, B. A., & Davis, D. R. (1997). Domestic political vulnerability and international disputes. Journal of Conflict Resolution41(6), 814-834.
    • Levy, J. S., & Thompson, W. R. (2010). Causes of war. India: Wiley-Blackwell (An Imprint of John Wiley & Sons Ltd).
    • Long, J. S. (1997). Regression models for categorical and limited dependent variables (Advanced quantitative techniques in the social sciences) (1 ed.). Thousand Oaks: Sage Publications, Inc.
    • Lujala, P., Gleditsch N.P., & Gilmore, E. (2005). A diamond curse: Civil war and a lootable resource. Journal of Conflict Resolution 49: 538-62.
    • Malone, D. M. & Nitzschke, H. (2005). Economic agendas in civil wars: What we know, what we need to know. United Nations University/World Institute for Development Economics Research Discussion Paper 2005/07.
    • Maoz, Z.  (2005). Dyadic MID Dataset (version 2.0): Retrieved from
    • Marshall, M. G. & Jaggers, K. (2002). Polity IV project: Political regime characteristics and transitions, 1800-2002. Retrieved August 11, 2010, from
    • Marshall, M. G., Gurr, T. R., Harff, B. (2009). Political instability task force–state failure problem set: Internal wars and failures of governance, 1955-2008. Political Instability Task Force. Retrieved August 11, 2010, from
    • McGillivray, F., & Smith, A. (2008). Punishing the prince: A theory of interstate relations, political institutions, and leader change. Princeton: Princeton University Press.
    • Minorities at Risk Project. (2009). Minorities at risk dataset. College Park, MD: Center for International Development and Conflict Management. Retrieved from on: [06/14/2010]
    • Mesquita, B. B., Morrow, J. D., Siverson, R. M., & Smith, A. (2004). The logic of political survival. London: The MIT Press.
    • Moore, W. H., & Davis, D. R. (1998). Transnational ethnic ties and foreign policy. In D.A.  Lake, D.A. & D. Rothchild, (Eds.), The International Spread of Ethnic Conflict (pp. 89-104). Princeton: Princeton University Press.
    • Mueller, J. E. (2009). War, presidents and public opinion. New York: The Educational Publisher/Zip Publishing.
    • Russett, B. M. & Oneal, J. (2000). Triangulating peace: Democracy, interdependence, and international organizations (The Norton Series in World Politics). New York: W. W. Norton.
    • Saideman, S. M. (2002). Overlooking the obvious: Bringing international politics back into ethnic conflict. International Studies Review, 4(3), 63-86.
    • Salehyan, I. (2007). Transnational rebels: Neighboring states as sanctuary for rebel groups. World Politics, 59 (January), 217-42.
    • Salehyan, I. (2009). Rebels without borders: Transnational insurgencies in world politics. Ithaca: Cornell University Press.
    • Sambanis, N. (2001). Do ethnic and nonethnic civil wars have the same causes? A theoretical and empirical inquiry (Part 1). Journal of Conflict Resolution45(3), 259-282. Retrieved June 27, 2010, from
    • Senese, P. D., & Vasquez, J. A. (2008). The steps to war: An empirical study. Princeton: Princeton University Press.
    • Singer, D. J., Bremer S., & Stuckey, J. (1972). Capability distribution, uncertainty, and major power war, 1820-1965. In B. Russett (Ed.) Peace, war, and numbers (pp. 19-48). Beverly Hills: Sage, 19-48.
    • Singer, D. J. (1987). Reconstructing the correlates of war dataset on material capabilities of states, 1816-1985. International Interactions, 14, 115-32.
    • Siverson, R. M., & Starr, H. (1991). The diffusion of war: A study of opportunity and willingness. Ann Arbor: University of Michigan Press.
    • Starr, H., & Most, B. A. (1976). The substance and study of borders in international relations research. International Studies Quarterly, 20, 581-621.
    • Tanja, E. (2000). Colorful community or ethnic witches brew? Multiethnicity and domestic conflict during and after the cold ar. Journal of Conflict Resolution44(2), 228-249.
    • Trumbore, P. F. (2003). Victims or aggressors? Ethno-political rebellion and use of force in militarized interstate disputes. International Studies Quarterly47(2), 183-201.
    • Vanhanen, T. (1999) Domestic ethnic conflict and ethnic nepotism. Journal of Peace Research 36: 55-73.
    • Walt, S. M. (1996). Revolution and war (Cornell Studies in Security Affairs). Ithaca: Cornell University Press.
    • Weinstein, J. M. (2005). Resources and the information problem in rebel recruitment.The Journal of Conflict Resolution 49(4), 598-624.
    • Woodwell, D. (2004). Unwelcome neighbors: Shared ethnicity and international conflict during the cold war. International Studies Quarterly, 48, 197-223.

    Table 1: Models 1-4, Coefficient Estimates, Whether a Militarized Interstate Dispute Is Initiated.

    Dependent Variables: MODEL 1    Non-Civil War State Dispute initiation: Ethnic civil war model MODEL 2    Non-Civil War State Dispute Initiation: Non-Ethnic Civil War Model MODEL 3    Civil War State Dispute Initiation: Ethnic Civil War Model MODEL 4    Civil War state Dispute initiation: Non-Ethnic Civil War Model
    Ethnic Civil War and Majority Ethnic Community in Bordering State .850  (.587) - .969  (.816) -
    Ethnic Civil War -.218  (.186) - .067  (.195) -
    Non-Ethnic Civil War - .118  (.175) _ -.143  (.187)
    Majority Transnational Ethnic Community in State Bordering Civil War State -.425  (.545) .285  (.213) -.466  (.789) .420  (.208)
    Transnational Ethnic Ties 1.337***  (.171) 1.322***  (.170) 1.046***  (.169) 1.037***  (.169)
    Capabilities Ratio -.031  (.258) -.069  (.5412) -.172  (.252) -.158  (.251)
    Level of Democracy  (Polity2) .030*  (.012) .030*  (.012) -.005  (.013) -.006  (.013)
    Trade Dependence -6.306**  (2.156) -6.210**  (2.121) -2.016  (1.514) -2.047  (1.60)
    Cold War .154  (.160) .139  (.159) .053  (.1541) -.048  (.154)
    Allied States -.790***  (.194) -.717***  (.187) -.537**  (.187) -.490**  (.183)
    * P<.05  ** p<.01  *** p<.001 N=1736  Pseudo R²=.0725 N=1736  Pseudo R²=.0708 N=1747  Pseudo R²=.0493 N=1747  Pseudo R²=.0481

    Table 2: Models 5 and 6, Coefficient Estimates, Whether a Militarized Interstate Dispute Is Initiated.

    Dependent Variables: MODEL 5  Dispute Initiation:  Non-Directional Ethnic Civil War Model MODEL 6  Dispute Initiation:  Non-Directional Non-Ethnic Civil War Model
    Transnational Ethnic Ties and Ethnic Civil War -.336  (.357) -
    Ethnic Civil War .023  (.288) -
    Non-Ethnic Civil War - .187  (174)
    Majority Transnational Ethnic Community in State Bordering Civil War State .138  (.227) .129  (.227)
    Transnational Ethnic Ties 1.595***  (.307) 1.346***  (.173)
    Capabilities Ratio -.502  (.562) -.587  (.555)
    Level of Democracy   (Polity2) .009  (.017) .011  (.017)
    Trade Dependence -5.412*  (2.280) -5.430*  (2.281)
    Cold War -.071  (.165) -.058  (.197)
    Allied States -.633**  (.198) -.637**  (.197)
    * P<.05  ** p<.01  *** p<.001 N=1596  Pseudo R2=.0686 N=1596  Pseudo R²=.0679

    Table 3: Predicted Effect of Significant Variables on Dispute Initiation.

      Probability Non-Civil War State Initiates a Dispute Probability Civil War State Initiates a Dispute Non-Directional Probability of Dispute Occurrence
    Base Probability of Dispute Initiation 7.91% 8.69% 8.54%
    Transnational Ethnic Ties are Present +16.48% +12.5% +12.72%
    Alliance is Present -3.76% -3.27% -3.26%
    Trade Dependence Increased One Standard Deviation -3.70% - -2.92%