Between indifference and coercion: Third-party intervention techniques in ongoing disputes

Abstract

Research on third parties’ conflict management has traditionally proposed a stark dichotomy between neutral mediators and non-neutral military joiners. Recent studies have blurred this dichotomy but have not investigated joiners’ use of techniques other than military action. Using data from Corbetta and Dixon (2005) on non-neutral interventions in post-Second World War interstate disputes, this paper explores non-neutral third parties’ choice of diplomatic, economic or military intervention techniques. It hypothesizes that such a choice is a function of third parties’ intensity of preferences for one side in conflict and antagonism toward the other side, which result from social proximity to the disputants.

Keywords

Bias conflict management technique intervention third-party states

Getting involved in ongoing conflicts of all kinds is one of the most complex foreign policy decisions that state leaders face. For example, at the time of this writing (Spring 2014), the Obama administration and European Union leaders find themselves in the uncomfortable position of having to decide whether and how to come to the aid of Ukraine and respond to Russia’s aggressive actions in Crimea and along Ukraine’s eastern border. The complexity of the situation is remarkable. The desire to help a new pro-Western government in Kiev must be balanced against the reality that the other disputant is a nuclear major power. Thus far, the US and European leaders have opted for a combination of diplomatic and economic assistance to Ukraine and economic sanctions against Russia.¹ In the rather unstable post-Cold War era, the tough choice of whether to intervene and with what strategy—or “technique”—presents itself quite frequently with regard to both interstate disputes—for example, US–Afghanistan in 2001, US–Iraq in 2003, Russia–Georgia in 2008, Russia and Ukraine in 2013—and intrastate conflicts—for example, Libya, Syria and Chad.

Despite the recursiveness of the “joining dilemma”, conflict scholars have paid comparatively more attention to mediation and other forms of neutral conflict management than non-neutral third parties that intervene in support of one side in an ongoing conflict and contribute to the conflict’s expansion.² While researchers of neutral interventions have investigated a variety of intervention techniques—from mediation to peacekeeping—research on non-neutral joining behavior has focused on one type of intervention only—military intervention. Third-party joiners have been assumed to have just two foreign policy options when facing ongoing conflicts: military intervention and abstention.³ This assumption is likely to produce biased results by underestimating the rate of third-party interventions (Corbetta, 2010) and by overlooking the possibility that combatants’ behavior will be affected by their expectations of specific types of intervention.⁴

In addition to a cursory view of international news, historical records suggest that third-party states intervene in both ongoing interstate and intrastate conflicts using a variety of techniques, covering the spectrum from diplomatic to economic to military. The question of what causes non-neutral third-party states to select a specific intervention technique over others has become more important as recent conflict management literature explores the possibility that mediators’ effectiveness may depend on their ability to apply one-sided coercive leverage just short of full-scale joining. This paper attempts to address this question by focusing on non-neutral third parties’ choice of military vs non-military—diplomatic and economic—intervention techniques and by proposing a theoretical model of joining behavior in interstate conflicts steeped in (sociological) structural theories of group formation and group interaction. Assuming that various third-party intervention techniques vary in regard to cost, the model suggests that joiners will choose more costly intervention techniques the closer they are in the social space to either combatant. This main hypothesis is tested using data on non-neutral third-party interventions in ongoing militarized interstate disputes (MIDs) from Corbetta and Dixon (2005) and ordered probit analysis with selection.

Joining behavior and intervention techniques

While students of peaceful conflict management use the generic term “mediation”, there is a general understanding that third-party states, intergovernmental organizations (IGOs) and individuals have a variety of conflict resolution techniques at their disposal (Frazier and Dixon, 2006; Owsiak, 2014). The same cannot be said of studies of joining behavior—that is, non-neutral interventions—which are characterized by the idea that decision-makers’ foreign policy menu contains only two options: non-intervention and military intervention. As early as 1969, Rosenau raised concerns about the vagueness and poor operationalization of the concept of intervention, stating that: “If one is interested in military intervention, operationalization may be accomplished in terms of the movement of a specified number of troops into or near the target society. Such a definition does not encompass other forms of intervention, but the omission of these does not mean that other forms do not occur” (Rosenau, 1969: 156).

Rosenau’s advice, however, has largely been neglected. Several factors motivate the narrow interest in military interventions. First, scholars of joining behavior have predominantly focused on intervention in conflicts of the highest magnitude—that is, wars. Second, there is a basic issue of data availability, as the most popular data collections provide information only about militarized behavior. Finally, in most cases attention to third-party joiners is a byproduct of other research questions that exclusively revolve around the use of military force—for example, balancing vs bandwagoning tendencies; alliance reliability; the geographical expansion of conflict; and the effectiveness of deterrence.

Scholarship is divided with regard to power considerations and third-party joining, with traditional realists theorizing that balance of power is a law-like occurrence in international politics (Morgenthau, 1985; Waltz, 1979), and others arguing that often states bandwagon with the source of the threat (Schweller, 1994). The empirical evidence is equally mixed, offering support for both balancing behavior in wars (Altfeld and Bueno de Mesquita, 1979) and interstate disputes (Werner and Lemke, 1997) and bandwagoning (Gartzke and Gleditsch, 2003). However, because existing works focus on military interventions only, the question of whether joiners may display different behaviors depending on the intervention technique they use is left unaddressed.

Others have explored joining behavior through the lenses of major power states’ foreign policy behavior. The general expectation is that great powers are more likely to become involved in other states’ conflicts militarily because they have enough capabilities and power projection to do so, and because of the breadth of their foreign policy interests (Volgy et al., 2011). Yamamoto and Bremer (1980) find that major powers are more likely to become involved in other great powers’ wars in a spiral-like dynamic that leads to the expansion of conflict. Huth (1998) provides evidence that major power states are more likely to intervene in crises in support of a protégé. Corbetta et al. (2013) find that status considerations affect major powers’ rate of intervention in both wars and militarized disputes. All of these authors, however, only consider militarized forms of intervention.

Similarly, studies on the expansion of conflict in size and space touch on the question of joining behavior, but mainly in its militarized form.⁵ As Starr (2005) eminently summarizes, geographical and environmental factors play a key role in the spatial diffusion of conflict. However, when conflict expands in space via addition of third parties, there is an unstated assumption that diffusion occurs only through military intervention. This assumption leaves open the possibility that, because non-military intervention techniques are not considered, many conflicts that are categorized as being limited in space may actually involve more states than the two (or more) originators. In addition, non-military interventions by third parties may spur preemptive military intervention by other joiners. In such instances, the omission of non-military interventions may be a source of sampling and modeling bias, as only a subset of all available interventions is considered and/or selection dynamics are ignored (Corbetta, 2010).

The study of third-party joining behavior naturally intersects with various research questions about alliances. The research on these topics is too vast to be adequately summarized here but, given that alliance reliability is measured by a state’s fighting on the same side as its allies, scholarly attention has gravitated toward militarized intervention. Early findings that alliances are unreliable (e.g. Sabrosky, 1980; Singer and Small, 1966; Siverson and King, 1979) have been superseded over time by strong evidence that, when we account for the specificity of alliance commitments, alliance partners do fight on the same side as their friends (Leeds et al., 2000). However, alliances also serve as signaling devices, and states involved in a conflict adjust their behavior based on their expectations about allies’ interventions (Smith, 1995, 1996; Werner, 2000a). While the literature on alliances focuses on military joining, it is possible that combatants develop their expectations about allied third parties’ intentions on the basis of interventions with other techniques—for example, economic interventions such as sanctions (Lektzian and Sprecher, 2007) or diplomatic signals (Corbetta, 2008).

Attention to joiners’ intervention techniques is also found in research about deterrence and the credibility of commitments. Again, the literature on this subject is too vast to be recapped here, but scholars of general or immediate extended deterrence—deterrence situations involving an aggressor, a third-party defender and a protégé—indirectly admit to the existence of two forms of intervention: coercive diplomatic threats and actual military intervention.⁶ The use of direct threats to use military force lies above the militarized threshold as canonically established by the MID project. However, as Fearon (2002) points out, there are various other forms of signaling in crisis bargaining that fall below the militarized threshold, and that can be considered forms of third-party intervention. Such forms of intervention, falling below the militarized threshold, are mostly neglected in the empirical literature on deterrence.

There are few studies addressing the possibility that non-neutral third-party states may employ multiple intervention techniques. Corbetta and Dixon (2005), in particular, present data on joiners’ biased interventions in ongoing MIDs and explore the frequency of diplomatic, economic and military intervention by third-party type—states, IGOs, non-governmental organizations, and individuals—in the post-Second World War era. Regan and Aydin (2006) and Regan et al. (2009) build a dataset of diplomatic and other forms of interventions in civil wars and investigate their impact on the duration of such conflicts. Aydin (2012) and Corbetta (2010, 2013) explore the distinction between military and non-military non-neutral interventions in ongoing interstate wars using the aforementioned Corbetta and Dixon (2005) data. Yet even the military/non-military distinction is not sufficiently fine-grained. Non-military interventions can take a variety of forms, as third parties can resort to different diplomatic or economic joining techniques. The following discussion largely builds on these works in order to develop a comprehensive explanation of third-party states’ interventions in ongoing militarized disputes.

Equally few are works that develop theoretical models of third-party interventions that may integrate and explain the choice of different intervention techniques. Most studies on militarized joining rely on some variant of the well-known opportunity and willingness framework (Joyce and Braithwaite, 2013; Joyce et al., 2014; Melin and Koch, 2010; Siverson and Starr, 1991). These studies share a focus on capability ratio and geographic distance as expressions of third parties’ opportunity for intervention, but are somewhat less specific about the sources of joiners’ willingness. Other works, instead, have concentrated on third parties’ willingness to intervene, as expressed by their biased preferences toward either disputant. Such preferences have been articulated in terms of strategic interests (Altfeld and Bueno de Mesquita, 1979; Bueno de Mesquita, 1989) or in terms of liberal state traits and foreign policy preferences, such as regime type and economic interests (Aydin, 2008, 2012; Reiter and Stam, 2002; Werner and Lemke, 1997).

Corbetta (2010, 2013) departs from these approaches by adopting a model steeped in sociological theories of group exchange that attributes strength of third parties’ preferences to their social distance with the disputant. This model explains third-party adoption of military techniques vs non-military techniques fairly well and, therefore, seems well suited to exploring further a more fine-grained distinction between various joining techniques. Assuming different costs between intervention techniques, the following section briefly reviews and builds on this model to explore under which circumstances third-party states choose to employ diplomatic, economic or military forms of intervention in ongoing interstate disputes.

Modeling third parties’ choice of intervention techniques

How do non-neutral third parties decide to intervene diplomatically, economically or militarily? Because most studies of joining behavior focus on military interventions, there is no ready-made model capable of generating predictions about third parties’ choice of other intervention techniques. Models that focus only on military interventions are only partially extensible because they inherently equate diplomatic or economic interventions with non-intervention. They do suggest, however, that third parties’ decisions are shaped by the strength of their preferences toward a combatant in a conflict. If preferences are sufficiently strong to push a third-party state beyond the indifference threshold, then a joiner will intervene militarily. More specifically, the stronger a potential third party’s preferences for a combatant, the greater the likelihood that the joiner will intervene, and that it will do so decisively—that is, militarily. This assumption represents a good starting point for theoretical development, although works on military joining have often neglected to clarify the source of variation in third parties’ preference strength.

In the joining literature, third parties’ preferences have been commonly expressed in terms of capabilities or alliance ties. Assuming that third-party states are guided by balance of power considerations, some scholars have suggested that joiners would prefer to intervene militarily in support of weaker combatants (Altfeld and Bueno de Mesquita, 1979; Bueno de Mesquita, 1989). An implication of this proposition is that, the greater the difference in capabilities between the combatants is, the greater the effort a joiner will be required to make. That is, as the difference in capabilities between enemies grows, the greater the likelihood is that a joiner will intervene with more intensive techniques—namely, military techniques. However, by itself, difference in capabilities does not fully explain why not all conflicts with mismatched combatants are equally strategically relevant to third parties.

Thus, joining scholars have also looked at alliance ties as an expression of third-party preferences. In their formal model of intervention and alignment in wars, Altfeld and Bueno de Mesquita (1979), for instance, use alliance ties to capture the utility a third party’s derives from participating in the war effort on the side of the preferred combatant. This type of utility expresses “the strategic importance that the third nation C attaches to each of the original belligerents” (Altfeld and Bueno de Mesquita, 1979: 97). This is contrasted with a third party’s utility for a specific outcome which is, instead, expressed by balance of power considerations. However, alliances are expressions of a variety of pre-existing preferences (see Cranmer et al., 2012; Lai and Reiter, 2000; Siverson and Emmons, 1991). In addition, because alliances are agreements about military cooperation, they are better predictors of military interventions alone. Alliance ties may not sufficiently reflect variations in preferences to allow us to explain the choice of lower-intensity techniques.

An alternative expression of strength of preferences is found in measures of regime similarity between joiners and combatants. Enterline (1999), Melin and Koch (2010) and Werner and Lemke (1997), for example, suggest that, the greater the similarity in regime type is between democratic joiners and democratic combatants, the shorter the time of their intervention and the greater the likelihood that they will side with their preferred side.⁷ Differences in regime type between states are expressed as differences in Polity scores (Marshall and Jaggers, 2002), and authors loosely appeal to the democratic peace in order to explain the sources of a democratic joiner’s preferences. Differences in degree between regime types are more granular expressions of preferences and may lead to more specific predictions about joiner’ choices of specific intervention techniques. However, the logic behind these explanations remains underspecified. The dynamic by which motivations for not fighting one another—that is, the democratic peace—translate into motivations to help another democratic combatant is not fully laid out. Appeals to the democratic peace logic may be further weakened by evidence suggesting that even nondemocratic states may also support one another in conflicts (Werner and Lemke, 1997).

An alternative, although less conventional, explanation of the origins of joiners’ preferences is found in structural theories of group formation and exchange. Sociological theories of group formation suggest that ties between social actors are the results of the existence of common traits between them. Actors with similar characteristics manifest the tendency to converge in the same social space—which some scholars have renamed “Blau space” from the work of Peter Blau (McPherson, 2004; Popielarz and McPherson, 1995). Convergence in the same area, or “niche”, of the latent social space increases the likelihood of tie formation and leads to the expectation that such ties will be positive and self-reinforcing (McPherson, Smith-Lovin and Cook, 2001). The principle stating that actors sharing similarity in key traits are more likely to share social ties has been labeled the homophily principle (Blau, 1977).

The homophily principle is easily summarized by the popular idea that “birds of a feather flock together”, but the notion is somewhat is more complex than that. As Blau (1977) suggests, not all shared traits among social actors result in homophilous ties. Rather, social actors must share a similarity in salient characteristics, or parameters. Parameters can be nominal or graduated. Nominal parameters express nominal characteristics that are either present or absent, shared or not shared. Graduated parameters, instead, are continuous. They allow for more variations and express degrees of similarity, or dissimilarity. Thus, homophily represents a multidimensional concept that allows us to place social actors in a multidimensional social space.

The homophily principle generates predictions about the nature of relations between social actors. Actors who are proximate to one another in the same niche of the social space as a result of shared discrete and graduated parameters are more likely to have denser, more positive relations (Blau, 1977: 36). The shorter the distance in the Blau space, the more alike two actors are, and the more likely to communicate they will be. Ultimately, “people from the same area of the social space will be similar in understandings, assumptions, and viewpoints” (McPherson, 2004: 270). The relationship between actors that are distant in the social space is, instead, considerably different.

In fact, the probability of the existence of ties and the density of communication between them decrease with social distance (Popielarz and McPherson, 1995). As a result, for actors who are in the same niche of the “Blau space”, the likelihood of experiencing conflictual relations increases with the distance along graduated parameters. The probability of conflict is even greater for actors who are in different niches of the social space, as they are not only distant according to graduated parameters but also unlikely to share nominal parameters (Blau, 1977).

Structural theorists have labeled the homophily resulting from shared characteristics between social actors as “value homophily” (Centola et al., 2007: 906). However, this is not the only manifestation of the homophily principle. Two or more actors occupying the same social niche not only share the same salient characteristics—that is, they have most of the same nominal parameters and are very similar in terms of graduated parameters—but they also tend to have a similar set of ties to other actors in the social space. The tendency of similar actors to interact with like-actors has been labeled “choice homophily” (McPherson and Smith-Lovin, 1987; McPherson et al., 2001). In a concept emerging from network analysis, two actors who have similar ties to other actors in the social space are said to be structurally equivalent, even if they may share few or no direct ties between one another. Two individuals i and j, for example, may not share a direct friendship tie with each other. However, they may have many friends in common, and many friends of their friends may be the same, etc. That is, the structure of the collection of their friendship ties may be such that i and j end up occupying similar (equivalent) positions in the social space. This form of indirect proximity leads to the expectation that the likelihood of positive relations between two actors will grow as their structural equivalence grows. Conversely, actors who are indirectly distant—they are not structurally equivalent—are more likely to have conflictual relations, or no relations.

Although caution must be exercised in applying these concepts to international politics (Hafner-Burton et al., 2009), both sociologists (for instance, Snyder and Kick, 1979) and international relations conflict scholars have shown that notions such as affinity, homophily and structural equivalence provide considerable leverage for explaining relations among states and broader social groups (Centola et al., 2007; Corbetta, 2013; Hafner-Burton et al., 2009; Kim and Barnett, 2007; Kinne, 2013; Maoz, 2006; Maoz et al., 2006, 2007; Maoz, 2011, 2012). The relative merits of homophily and structural equivalence vis-à-vis commonly used expressions of similarity in international relations such as tau-b (Bueno de Mesquita, 1981), S score (Signorino and Ritter, 1999) and United Nations General Assembly voting (Gartzke, 2000) have been amply discussed elsewhere (Corbetta, 2010, 2013; Maoz, 2011; Maoz et al., 2006). To summarize, there are two main differences between the approach involving homophily and structural equivalence and the approach used for other expressions of similarity. First, alliance portfolios or UNGA votes are expressions of underlying similar interests but do not specify the origin of such similarity. Homophily, instead, requires that one specify the process—convergence in the social space—from which similar policy preferences arise (Corbetta, 2010, 2013). Second, structural equivalence captures similarity in terms of ties to others in ways that alliance portfolios or S scores do not by considering not only direct ties, but also the structure of second-, third- and higher-order connections between the focal actor—a state, in our case—and other actors in a network (Maoz et al., 2006; Maoz, 2011).

Overall, the strength of the homophily approach is that it explains states’ preference formation in more general terms than traditional international relations theories. The general proposition emerging from this framework is that states with similar characteristics will want to preserve friendly relations with each other either because of their self-interest in cooperation, as liberalism would suggest, or because of interest in a stable and conflict-free strategic environment, as realism would propose.

This approach allows one to generate basic hypotheses about third-party states’ choices of specific intervention techniques. The model starts with the assumption that intervention is costly, and that costs vary by intervention technique.⁸ At one extreme are military interventions, which can impose reputational, political, economic and human costs. Many conflicts scholars have highlighted the reputational and credibility costs that states suffer from backing down from military threats (e.g. Guisinger and Smith, 2002; Sartori, 2002, 2005; Schultz, 2001). Such costs may be even higher if the threat is carried out unsuccessfully. Deployment in foreign conflicts is also difficult for policy-makers to sell to domestic audiences. Military mobilization is also expensive, and it is often used to produce the “sunk costs” necessary to communicate the credibility of a state’s intentions (Slantchev, 2005). Actual deployment of troops, of course, produces not only reputational and economic costs, but also human costs. Lives are likely to be lost in the fight. Families, and society as a whole, pay the price of participation in conflicts, often over the course of decades.

At the other extreme, there are diplomatic intervention techniques. These are diplomatic signals that may range from vague expressions of support or opposition, in which a third party issues a public statement expressing solidarity for a combatant or condemning another combatant’s actions, to diplomatic sanctions (Corbetta and Dixon, 2005). Such messages lack the specificity and visibility of military threats. Although they may generate reputational and credibility costs (Corbetta, 2008; Thyne, 2006, 2009), they remain signals within a larger bargaining interaction, aimed at assessing the opponent’s intentions or at communicating one’s type.⁹ The predominant view in conflict research is to treat such signals as cheap talk, although there is sparse evidence that they are not entirely costless and may increase the risk of escalatory interventions (Corbetta, 2008).

Economic interventions occupy a middle position between diplomatic and military threats. Interventions such as economic sanctions involve economic costs not only for the target, but also for the sanctioner. While there is debate about the actual effectiveness of economic interventions, and especially economic sanctions, there is growing consensus that the key feature of economic interventions lies in their ability to communicate intentions. Economic interventions involve sunk costs for the sanctioning state, and such costs increase the credibility of the sender’s signal. Empirical evidence further suggests that economic sanctions increase the likelihood of further military intervention (Lektzian and Sprecher, 2007). Yet backing down from economic sanctions cannot be considered as costly as backing down from military threats or as being militarily defeated on the battlefield.

Assuming then that different forms of intervention vary in costs, one can hypothesize that a potential joiner’s willingness to incur higher costs will vary with the strength of its preferences for a disputant. As preferences are shaped by proximity in the international social space, this paper explores the following key hypotheses:

H1: the greater the proximity in the international social space between a third party and a disputant, the higher the costs of the intervention technique a joiner will employ in support of that disputant.

Conversely,

H2: the greater the distance in the social space between a joiner and disputant, the higher the costs of the intervention technique a joiner will employ in opposing a disputant.

More specifically, the following analysis will test whether greater direct homophily and greater structural equivalence are effective predictors of the adoption of costlier intervention techniques. The null hypothesis is that no significant difference exists with regard to the effect of social proximity across different types of intervention techniques.

Explanations alternative to the social proximity account suggest that the choice of intervention technique is not a function of third parties’ preferences, but of security interests and opportunity. Given the presence of geostrategic concerns and the opportunity to intervene, third parties will choose more expensive forms of intervention if they have the opportunity to do so. Capabilities and geographical distance are usually said to influence this choice (Joyce and Braithwaite, 2013; Joyce et al., 2014). If intervention is exclusively a function of opportunity, joiners may be more likely to resort to resource-intensive techniques the more geographically proximate they are to the conflict.

The relationship between capabilities and intervention is somewhat more complex. Capabilities provide third parties with the opportunity to project their influence across vast spaces and to intervene in distant conflicts. Capabilities allow third parties to employ expensive intervention techniques. However, the need to intervene decisively and expensively may vary in relation to the capabilities of other actors in a conflict. It is reasonable to expect that, the greater the gap in capabilities in favor of a third party, the less likely the joiner will be to have to adopt costly techniques. Thus, somewhat paradoxically, capabilities increase a third party’s opportunity for intervention while reducing the need that such an intervention will be expensive. If these alternative accounts are correct, no significant relationship between homophily, structural equivalence and choice of intervention technique should emerge after controlling for capabilities and geographic distance. The empirical test of the main hypotheses concerning social proximity, along with a description of the measurement of the variables of interest, is outlined in the following section.

Research design

Testing hypotheses about third parties’ choices of non-neutral intervention techniques is complicated by the fact that most interstate conflict datasets only consider military interventions. For example, the MID dataset only allows for the exploration of third-party states’ preferences between military threats, military mobilization and actual use of military force. Although it covers a shorter time span—the 1946–2001 period—Corbetta and Dixon’s (2005) dataset on third-party non-neutral interventions in ongoing MIDs contains information about various categories of intervention. Corbetta and Dixon classify interventions into three key categories: diplomatic, economic and military. Diplomatic interventions include: official public statements of opposition or support for a disputant in an ongoing MID; appeals or demands to cease fire or withdraw troops exclusively directed to one side in the dispute; direct diplomatic assistance;¹⁰ and diplomatic sanctions. Economic interventions include: promise of economic aid to a disputant; one-sided threat of economic sanctions; actual provision of economic aid; and economic sanctions. Military interventions range from the offer and provision of military assistance short of direct fighting, such as the provision of military aid or military advisors, to the actual use of military force in support of one of the disputants.¹¹ Given the assumption of increasing costs of these techniques, the ordered choice of a diplomatic, economic or military intervention is the dependent variable of interest. Because states may intervene multiple times with different techniques in the course of a dispute, and because multiple interventions are not independent of one another (Corbetta, 2010), this analysis considers only the technique used during a third party’s first intervention.

However, the sample of third parties that intervene in post-Second World War disputes is not random. The choice of an intervention technique is likely to be dominated by selection effects, as joiners voluntarily select themselves into the sample. As a result, it is important to model the simultaneous process of intervention and choice of a specific intervention technique. For this purpose, a dataset matching each post-Second World War MID originator with a potential joiner was created.¹² For each potential joiner–MID originator dyad, the actual occurrence of an intervention on one side of an ongoing dispute was recorded dichotomously. In the following analysis, predictions about the adoption of a specific intervention technique were conditioned on the actual occurrence of an intervention. The resulting dataset contains approximately 630,000 potential joiner–MID originator dyads.¹³ Only 2874 of these dyads actually experienced an intervention. Among these, 1846 were diplomatic interventions, 182 were economic and 846 were military.

Corbetta (2013) points out that, in non-neutral third-party interventions, bias toward the disputants can take two forms. An intervening third-party state can have a strong preference for one of the disputants or a strong “dislike” of the other combatant. The strength and direction of a third party’s bias toward the conflict originators affects a third party’s decision to intervene and the choice of an intervention technique. Corbetta (2010), for example, shows that, in most interstate disputes, intervention is more likely to result from a third party’s “dislike” for one side in the conflict than from its affinity toward the side that is being supported. In essence, a third-party state can intervene in support of one of the combatants or against the other side. Different dynamics shape the intervention decision, depending on whether a third party wants to assist one side or antagonize the other side in a dispute. Thus, following Corbetta (2013), interventions are further differentiated between supportive interventions, in which a third-party state’s action is directed toward aiding a disputant, and antagonistic interventions, in which a third party’s action is directed against a combatant.¹⁴

The small ratio of actual to potential interventions creates computational complications for the convergence of the ordered probit models with selection used for this analysis. As an imperfect solution, a representative sample of potential interventions was randomly selected from the population of post-Second World War cases, while all of the actual interventions were retained. The final dataset used for the analysis contains 28,393 observations.¹⁵ Because of the assumption of growing costs involved in these intervention techniques, the dependent variable ranges from 0 (non-intervention) to 3 (military intervention) for both supportive and antagonistic interventions.

Operationalizing multidimensional concepts such as homophily and structural equivalence is challenging, but existing research offers some guidance as to which dyadic traits should be considered. Similarity in regime type has previously been found to affect joiners’ alignment choices (Corbetta, 2010, 2013; Werner and Lemke, 1997). Another oft-cited dimension of similarity among states concerns their cultural, religious and linguistic traits (Centola et al., 2007; Maoz, 2011, 2012). A third, less frequently used, dimension of interstate homophily is similarity in institutional structures. Institutional similarity seems to facilitate communication between states, and evidence suggests that it reduces both the occurrence and the duration of conflicts (Werner, 2000b). These traits certainly do not exhaust all the dimensions of similarity states. Some degree of personal discretion inevitably enters the operationalization of homophily. However, these shared traits have consistently proved to influence states’ behavior, bringing them together in a common social and policy space and leading to both absence of conflict and relative interstate solidarity. Absence or low levels of these traits, instead, have consistently been associated with conflictual relations at the dyadic level (Centola et al., 2007; Maoz, 2011, 2012; Werner, 2000b).

As stated earlier, similarity in key traits among social actors represents what sociologists have named “value homophily” (Centola et al., 2007: 906). However, there is another dimension of affinity, which can be called “choice homophily” (McPherson and Smith-Lovin, 1987; McPherson et al., 2001), and which reflects the tendency of two social actors to interact with the same set of alters, even if they do not share a direct connection with one another. That is, like actors tend to develop similar sets of social interactions. This notion is better captured by the concept of structural equivalence among states that belong to the same international institutions—such as alliances and IGOs—or have frequent interactions—for example, states that trade intensely with one another.

Homophily and structural equivalence can be operationalized with widely available data and known procedures. With regard to value homophily, similarity in regime type was measured with the Polity IV project’s index of difference between a state’s democratic attributes and its autocratic attributes (DEM-AUT), which ranges from −10 (least democratic) to +10 (most democratic) (Marshall and Jaggers, 2002). Dyads in which both states score less than −7 or more than +7 on this index were identified as representing “mature regime dyads” and were coded as 1. Dyads in which one or both member states fall between a score of +7 and −7 were coded as 0. Henderson and Tucker (2001) provide a measure of cultural similarity between states. This is a dichotomous indicator assuming a value of 1 if a joiner and a disputant share the same civilization traits.¹⁶ Finally, Werner (2000b) offers a measure of institutional homophily by employing various indicators from the Polity project to calculate the Euclidian distance in institutional structures between two states. Each dyad that is at least one standard deviation above the Werner’s indicator mean for a given year was coded as 1. Dyads below that threshold were coded as 0.¹⁷

Although there are numerous ways to estimate structural equivalence, a commonly used procedure involves the estimation of Pearson’s product-moment correlations (Hafner-Burton and Montgomery, 2006; Kim and Barnett, 2007; Maoz, 2011; Maoz et al., 2006). In a series of articles Maoz and his associates exhaustively discuss the process involved in estimating first-, second- and higher-order measures of structural equivalence (Maoz, 2006, 2011; Maoz et al., 2005, 2006, 2007).¹⁸ Following Maoz et al. (2006), three measures of structural equivalence are employed in order to capture the idea that states that share the same relations with other states in the international system should be socially proximate to one another and, therefore, more likely to experience positive interactions with each other. First, using Alliance Treaty and Obligation Provisions data (Leeds et al., 2000, 2002), Maoz et al. (2006: 676–677) derive an alliance-based measure of structural equivalence. Second, using trade data from Gleditsch (2002) and Oneal and Russett (2005), a trade-based measure of structural equivalence is estimated by entering in the sociomatrix the proportion of a state’s gross domestic product that comes from total trade with another country. Third, with data from Pevehouse et al. (2004) on IGO membership, Maoz et al. (2006) compute an IGO-based measure of structural affinity, which is based on the number of IGO memberships per year shared between countries. All of the measures from Maoz et al. (2006) are estimated as so-called first-order measures.

The adoption of a specific intervention technique may also depend on strategic or opportunistic calculations. Geographical proximity and material capabilities can affect whether a joiner can or must employ costlier techniques. The impact of a joiner’s capabilities on the choice of intervention techniques is captured through the Correlates of War project’s Cumulative Index of National Capabilities (Singer et al., 1972; Singer, 1988). Specifically, the ratio of capabilities between a disputant and a potential joiner was computed. Spatial distance, instead, was calculated as the logged capital-to-capital distance in miles between a joiner and a disputant.¹⁹

Testing the above hypotheses about social proximity and intervention involves modeling the interdependent dynamics of selection and intervention technique choice. Ordered probit regression with sample selection meets this requirement. This approach involves the simultaneous estimation of two probit equations—a selection equation, in which the dependent variable is the dichotomous occurrence of a third-party intervention, and an outcome equation, in which the dependent variable is the ordered choice of a diplomatic, economic, or militarized intervention technique. This approach tests for the existence of a significant correlation between the error term of the selection equation and the error term of the outcome equation. Following the theoretical model outlined earlier, it is hypothesized that potential joiners who are more socially proximate to either disputant will be more likely to have a stake in the outcome of the conflict and, therefore, will be more likely to intervene. Thus, the same covariates that appear in the outcome equation will also appear in the selection equation. Because the presence of the same regressors in both the selection and the outcome equation leads to weak model identification, an instrumental variable was added to the selection equation. This variable is a dichotomous indicator of whether a third party-disputant dyad is composed of two Middle Eastern countries.²⁰ The results from these models are presented below.

Results

Table 1 presents the results of an ordered probit regression with selection, with robust standard errors, in which both supportive and antagonistic interventions are considered in the outcome equation.²¹ From this model, the social proximity explanation of choice of intervention technique—as expressed by the three homophily variables and by the three structural variables—is only weakly supported. In the selection equation only regime similarity, cultural similarity and IGO structural equivalence increase the likelihood of intervention. In the outcome equation only mature regimes seem willing to invest greater resources to support similar regimes. Institutional similarity actually reduces the chances of adoption of costly techniques. Only IGO structural equivalence behaves as expected in explaining intervention, while none of the structural equivalence variables is significant in predicting the choice of specific intervention techniques.

Table 1.

Ordered probit (with selection) estimates of third parties’ choice of diplomatic, economic or military techniques in MID interventions, 1946–2001

	Coefficient	Standard error	p > z
Outcome
Regime similarity	0.236	0.066	0.000
Institutional similarity	−0.286	0.071	0.000
Cultural similarity	−0.084	0.086	0.328
IGO SE	0.479	0.388	0.217
Alliance SE	0.075	0.105	0.478
Trade SE	0.043	0.268	0.872
Capability ratio	−0.545	0.113	0.000
Log distance	0.016	0.060	0.794
Selection
Regime similarity	0.174	0.026	0.000
Institutional similarity	−0.029	0.025	0.252
Cultural similarity	0.134	0.035	0.000
IGO SE	1.602	0.078	0.000
Alliance SE	−0.150	0.053	0.005
Trade SE	−1.538	0.081	0.000
Capability ratio	−0.376	0.050	0.000
Log distance	−0.169	0.017	0.000
Middle East	−0.114	0.028	0.000
Constant	−0.706	0.171	0.000
Cut 1	1.348	0.775	0.082
Cut 2	1.523	0.766	0.047
Athρ	0.490	0.242	0.043
ρ	0.490	0.192
N	22,737
Log pseudolikelihood	−88,740.591
χ ²	56.50
p > χ²	0.000
χ² (ρ = 0)	4.10
p > χ² (ρ = 0)	0.043

SE, Structural equivalence.

The explanatory effectiveness of the social proximity model, however, could be watered down by the conflation of supportive and antagonistic interventions. In biased interventions it matters whether third parties are relatively more interested in supporting one side in the conflict, opposing the other side, or both. Relative social distance between a third party and the disputants may make a difference. Table 2 presents the results of two separate ordered probit equations with selection. The first model examines the choice of intervention technique as a function of a third party’s relationship with the side being supported in the conflict—supportive intervention. The second model looks at the relationship between a joiner and the side being opposed in the conflict—antagonistic interventions. If the social proximity model has any explanatory ability, we should observe opposite trends in the effect of homophily-based and structural equivalence preferences on the choice of intervention technique. Proximity in the international social space should increase the likelihood that a joiner will adopt a costlier technique, whereas distance in the social space should decrease that probability.

Table 2.

Ordered probit (with selection) estimates of third parties’ choice of diplomatic, economic, or military techniques in supportive and antagonistic MID interventions, 1946–2001*

	Supportive			Antagonistic
	Coefficient	Standard error	p > z	Coefficient	Standard error	p > z
Outcome
Regime similarity	0.252	0.059	0.000	−0.141	0.074	0.057
Institutional similarity	0.346	0.059	0.000	−0.576	0.072	0.000
Cultural similarity	0.062	0.065	0.341	−0.156	0.067	0.021
IGO SE	0.398	0.255	0.118	−0.471	0.205	0.022
Alliance SE	0.461	0.094	0.000	−0.462	0.107	0.000
Trade SE	0.369	0.205	0.072	−0.035	0.206	0.863
Capability ratio	−0.630	0.097	0.000	0.365	0.088	0.000
Log distance	0.035	0.038	0.358	0.011	0.040	0.779
Cut 1	1.135	0.491	0.021	−0.865	0.462	0.042
Cut 2	2.122	0.469	0.000	0.030	0.441	0.946
Cut 3	2.259	0.467	0.000	0.171	0.444	0.700
Athρ	0.318	0.155	0.040	−0.297	0.129	0.022
ρ	0.307	0.140		−0.288	0.119
N	22,737			22,737
Log pseudolikelihood	−9357.436			−9129.987
χ ²	363.06			385.74
p > χ²	0.000			0.000
χ² (ρ = 0)	4.21
p > χ² (ρ = 0)	0.0401			0.0219

Selection equation coefficients not reported.

This is what emerges from Table 2, and the results are largely consistent with Corbetta (2010, 2013). With regard to supportive interventions, the outcome equation shows that regime similarity, institutional similarity and equivalence in alliance structure correctly predict the adoption of more costly intervention techniques. Structural equivalence in trade relations is marginally significant (p < 0.10). Concerning antagonistic interventions, similar patterns emerge, but, as expected, they are in the opposite direction. As social distance in institutional similarity, cultural similarity and structural equivalence in alliance and IGO increases, a third party is less likely to resort to costly intervention techniques. In this case, similarity in regime type achieves borderline significance and, most importantly, is in the expected negative direction. These results hold after controlling for strategic considerations. Geographical distance has no significant effect on the adoption of progressively more costly techniques. Capabilities, instead, deter costly interventions of all types. As the capability ratio approaches parity, the likelihood of adopting costly intervention strategies diminishes.²²

It is possible to further test the validity of the social proximity model of intervention technique choice by creating cumulative indices of proximity in the international social space. Following Centola et al. (2007) and Corbetta (2013), a four-point composite homophily index was created. This ranges from 0, when a third party and a disputant share no relevant traits, to 3, when they are both mature regimes, have similar institutions and share the same cultural traits. Maoz et al. (2006) provide a procedure for integrating various measures of structural equivalence into a single indicator. Table 3 presents results from two ordered probit models with selection—one for supportive and one for antagonistic interventions—when the two cumulative indicators of international social proximity are used. Only the outcome equations are reported, as the selection equations are consistent with those in the previous tables.

Table 3.

Ordered probit (with selection) estimates of the effects of composite homophily and structural equivalence on third parties’ choice of diplomatic, economic, or military techniques in supportive and antagonistic MID Interventions, 1946–2001*

	Supportive			Antagonistic
	Coefficient	Standard error	p > z	Coefficient	Standard error	p > z
Outcome
Composite homophily	0.237	0.032	0.000	−0.298	0.066	0.000
Integrate SE	0.648	0.122	0.000	−0.665	0.153	0.000
Capability ratio	−0.615	0.100	0.000	0.334	0.131	0.011
Log distance	0.038	0.067	0.572	−0.029	0.125	0.815
Cut 1	0.771	0.261	0.003	−0.567	0.314	0.071
Cut 2	1.771	0.259	0.000	0.354	0.331	0.285
Cut 3	1.908	0.261	0.000	0.499	0.334	0.135
Athρ	0.211	0.315	0.504	−0.066	0.648	0.919
ρ	0.207	0.302		−0.066	0.645
N	22,737			22,737
Log pseudolikelihood	−9945.166			−9709.88
χ ²	315.20			192.59
p > χ²	0.000			0.000
χ² (ρ = 0)	0.45			0.01
p > χ² (ρ = 0)	0.5044			0.918

Selection equation coefficients not reported.

This last set of results is consistent with prior findings. As the direct and indirect social similarity—homophily and structural equivalence, respectively—between third parties and disputants grows, so does the likelihood that a third party will invest larger resources in support of a disputant. Confirming that intervention dynamics differ depending on whether a third party is interested in supporting a friend or in opposing an enemy, the direction of the likelihood of adopting more costly intervention techniques is reversed in antagonistic interventions. With regard to the side that is being opposed in an intervention, as the international social proximity between a third party and a disputant increases, the less likely a joiner is to employ more resource-intensive techniques.

Finally, it is possible to assess the substantive effect of direct and indirect social proximity between a third-party state and the disputants on the adoption of costly intervention techniques. The predicted probabilities of adopting specific intervention techniques, conditional on a third party’s decision to intervene in a dispute, were computed from the models in Table 3. The boxplots in Figures 1 and 2 display the distributions of predicted probabilities around their median estimate and across the four levels of the cumulative homophily index. Because intervention in an ongoing conflict is an infrequent phenomenon, the estimated predicted probabilities remain generally low. In all cases, the probability of economic interventions is considerably lower than the probability of diplomatic and military interventions. With regard to supportive interventions, the three panels in Figure 1 consistently show that the median probability of adoption of diplomatic, economic and military techniques increases positively across homophily levels. However, the impact of social proximity becomes more dramatic for more expensive intervention techniques. The first panel from the left indicates that the median predicted probabilities of less costly—that is, diplomatic—interventions remain relative high even at lower levels of homophily. Conversely, the panel for the median predicted probabilities of military interventions shows drastic changes as the cumulative homophily index reaches its maximum value of 3.

Figure 1.

Supportive intervention techniques: distribution of conditional predicted probabilities by homophily level.

Figure 2.

Antagonistic intervention techniques: distribution of conditional predicted probabilities by homophily level.

As expected, the opposite pattern is detectable with regard to antagonistic interventions in the three panels of Figure 2. Here the median predicted probability of adoption of hostile intervention strategies of any kind drops linearly as the social proximity between a third party and a combatant increases. The more “homophilous” a third party and a combatant are, the less likely a third party is to intervene against that combatant and to adopt a costly strategy in doing so. The adoption of antagonistic intervention techniques is substantively more likely when a third party and a disputant are socially distant from one another. Figure 2 shows that, again, the median predicted probability of antagonistic economic intervention is considerably lower than that of diplomatic and military intervention. The influence of social proximity is easily observable as we move from less costly (diplomatic) to more costly (military) intervention techniques. Greater homophily reduces the probability of diplomatic intervention techniques, but not nearly as much as it reduces the probability of antagonistic military interventions. At the highest level of social proximity between a third-party state and a disputant, the median predicted probability of an antagonistic military intervention approaches zero.

The present findings provide support for the general theme of the social proximity argument, but such support is far from unequivocal. The results are robust to the inclusion of other control variables, but not all of the indicators of direct and indirect homophily behave as expected. Both theoretically and empirically, modeling the choice of intervention strategy is a considerably more complex task than simply modeling intervention—whether neutral or non-neutral. The social proximity model investigated in this paper allows for more specific expectations than alternative explanations of intervention, which simply differentiate between non-intervention and military involvement. Further refinement of the underpinnings of the theoretical model presented above is necessary. Equally important is further refinement of the estimation techniques available to scholars of third-party conflict management. The existence of selection dynamics in interventions in conflicts is well known, but such dynamics complicate the empirical analysis of third parties’ choices once they have committed to intervention. Much progress in this area has been made, but the available econometric techniques are less stable and robust than the more familiar estimation routines where the dependent variable is dichotomous—that is, non-intervention/military intervention.

Conclusions

In the area of conflict management there is growing interest in the use of coercive techniques and the effectiveness of biased mediators. Until recently such topics were the purview of studies on joining behavior, which, for the most part, exclusively focused on military interventions. Military interventions are clearly easier to observe and record. The notion that an intervention, in order to be considered as such, should represent a radical break with the normal state of relations between states (Rosenau, 1969) influenced existing studies. The unintended result was the creation of an artificial separation between non-coercive conflict management and joining behavior, and between neutral and non-neutral forms of intervention. With regard to non-neutral interventions, emphasis on military joining led to the adoption of theoretical models that, for the most part, stressed third-party states’ security interests and opportunities for intervention—as reflected in direct alliance ties and state capabilities, respectively. Although effective and insightful, these models were unable to explain and predict the occurrence of non-neutral interventions below the military threshold. Few researchers have looked at the determinants of third-party states’ decision to adopt specific techniques beside military interventions. Theoretical models that incorporate a more nuanced view of third-party states’ interests, as expressed by the notion of social proximity between interveners and disputants, allow more specific explanations and predictions of the conditions under which third parties may resort to diplomatic, economic or military interventions. Such an approach is desirable and useful as both scholars of peaceful conflict management and scholars of joining behavior increasingly look at mediation and military coercion as extremes along a continuum, rather than mutually exclusive behaviors.

This paper attempts to shift the focus from the mediation–joining dichotomy to intervention as a more complex approach to conflict management—one that involves several different options for third parties. Because existing theories offer little guidance as to what may guide a joiner’s choice of intervention technique, this study offers a model based on well-established sociological theories of group formation and interaction. The driving proposition is that proximity in the international social space shapes the intensity of states’ preferences. States that are more proximate in the multidimensional social space—that is, have a higher degree of direct or indirect homophily—are more likely to adopt more costly intervention techniques to support each another. Conversely, states that are more proximate to each other are less likely to use costly intervention techniques when they oppose each other in the course of ongoing disputes.

The empirical test of the propositions emerging from this model exposes the many challenges involved in predicting whether and how a third-party state will enter an ongoing interstate dispute. The analysis provides some support for the social proximity model of joiners’ preferences but revealed inconsistencies that invite further work and revisions. This approach is more efficient at predicting the adoption of more costly techniques in supportive interventions than in antagonistic interventions. Some of the challenges involved in testing propositions about the adoption of specific intervention techniques are computational and built into the available econometric models. Most statistical models are based on the assumption that alternative outcomes of the dependent variable—different intervention techniques in this case—are independent of one another. This assumption is bound to be violated in the analysis of a choice between three interdependent options—diplomatic, economic and military intervention. However, the alternative assumption—that technique adoption choices depend on the intrinsic costs of each intervention strategy—is also problematic as it is based on the idea that the gaps in costs between diplomatic and economic interventions and between economic and military intervention are proportional. This is a difficult assumption to hold, because some states can attach different costs to different techniques, and because costs can only be assumed or evaluated ex post.

In spite of these challenges, the exploration of the many intervention techniques available to third parties is important to our understanding of conflict dynamics. The onset, evolution and termination of ongoing conflicts may be affected by the combatants’ own expectations about specific types of intervention, by the intervention techniques that third parties are willing to or can employ, and by the effectiveness of specific techniques. The traditional distinction between mediation and joining has produced great insights in the area of conflict management but, arguably, it has also produced two separate bodies of literature that have not sufficiently communicated with one another. The boundaries between these two lines of research are being increasingly blurred by studies that stress the effectiveness of biased mediators, coercive conflict management and the availability of multiple intervention techniques in both interstate and intrastate studies.²³ Ongoing data collection projects provide information about third parties’ varied intervention strategies (Corbetta and Dixon, 2005; Frazier and Dixon, 2006; Hogbladh et al., 2011; Melander et al., 2009; Regan et al., 2009). Still missing are theoretical models explaining the determinants of third parties’ choices of intervention techniques. This paper makes a first attempt in this direction by offering an alternative framework to more traditional strategic or bargaining models of conflict management, and is intended to stimulate further theoretical and empirical work on third parties’ conflict management choices and the intersection between neutral and non-neutral intervention.

Footnotes

Acknowledgements

The author would like to thank Molly M. Melin, Andy Owsiak, Paul Poast, Paul Diehl, Resat Bayer, Bill Dixon and the anonymous reviewers for their comments on earlier drafts.

Author’s note

A previous version of this paper was presented at the 2013 International Studies Association Convention, San Francisco, CA, USA, 3–6 April.

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Notes

References

Altfeld

Bueno de Mesquita

(1979) Choosing side in war. International Studies Quarterly 23: 87–112.

Aydin

(2008) Choosing sides: Economic interdependence and interstate disputes. Journal of Politics 70: 1098–1108.

Aydin

(2012) Foreign Powers and Interventions in Armed Conflicts. Redwood City, CA: Stanford University Press.

Beardsley

(2009) Intervention without leverage: Explaining the prevalence of weak mediators. International Interactions 35: 272–97.

Beardsley

(2013) Using the right tool for the job: Mediator leverage and conflict resolution. Penn State Journal of Law and International Affairs 2: 57–65.

Bennett

Stam

(2000) EUGene: A conceptual manual. International Interactions 26: 179–204.

Blau

(1977) Inequality and Heterogeneity: A Primitive Theory of Social Structure. New York: Free Press.

Bueno de Mesquita

(1981) The War Trap. New Haven, CT: Yale University Press.

Bueno de Mesquita

(1989) The contribution of expected utility theory to the study of international conflict. In: Midlarsky

(ed.) Handbook of War Studies. Boston, MA: Allen and Unwin.

10.

Centola

Gonzales-Avella

Eguiliuz

San Miguel

(2007) Homophily, cultural drift, and the co-evolution of cultural groups. Journal of Conflict Resolution 51: 905–929.

11.

Corbetta

(2008) Informal commitments and the escalation of intervention in ongoing conflicts. Typescript.

12.

Corbetta

(2010) Determinants of third parties’ intervention and alignment choices in ongoing conflicts, 1946–2001. Foreign Policy Analysis 6: 61–85.

13.

Corbetta

(2013) Cooperative and antagonistic networks: Multidimensional affinity and intervention in ongoing conflicts, 1946–2001. International Studies Quarterly 57: 370–384.

14.

Corbetta

Dixon

(2005) Danger beyond dyads: Third party participants in interstate disputes. Conflict Management and Peace Science 22: 39–61.

15.

Corbetta

Grant

(2012) Intervention in conflicts from a network perspective. Conflict Management and Peace Science 29(3): 314–340.

16.

Corbetta

Volgy

Rhamey

(2013) Major power status (in)consistency and political relevance in international relations studies. Peace Economics, Peace Studies, and Public Policy 19: 291–307.

17.

Cranmer

Desmarais

Kirkland

(2012) Toward a network theory of alliance formation. International Interactions 38: 295–324.

18.

De Luca

Perotti

(2011) Estimation of ordered response models with sample selection. Stata Journal 11: 213–239.

19.

Ellingsen

(2000) Colorful community or ethnic witches’ brew? Multiethnicity and domestic conflict during and after the Cold War. Journal of Conflict Resolution 44: 228–249.

20.

Enterline

(1999) Some other time: The timing of third party intervention in dynamic interstate conflicts, 1816–1992. Paper presented at the 33rd Annual Meeting of the Peace Science Society (International), Ann Arbor, MI, 8–10 October.

21.

Favretto

(2009) Should peacemakers take sides? Major power mediation, coercion, and bias. American Political Science Review 53: 248–263.

22.

Fearon

(1994) Signaling versus balance of power and interests: An empirical test of a crisis bargaining model. Journal of Conflict Resolution 38: 236–269.

23.

Fearon

(2002) Selection effects and deterrence. International Interactions 28: 5–29.

24.

Frazier

Dixon

(2006) Third party intermediaries and negotiated settlements, 1946–2000. International Interactions 32: 385–408.

25.

Garztke

(2000) Preferences and the democratic peace. International Studies Quarterly 44: 191–210.

26.

Gartzke

Gleditsch

(2003) Balancing, bandwagoning, bargaining, and war. Paper presented at the European Consortium of Political Research Joint Workshop Sessions, Edinburgh, 28 March to 2 April.

27.

Ghosn

Palmer

Bremer

(2004) The MID3 data set, 1993–2001: Procedures, coding rules, and description. Conflict Management and Peace Science 21: 133–154.

28.

Gleditsch

(2002) All International Politics Is Local: The Diffusion of Conflict, Integration, and Democratization. Ann Arbor, MI: University of Michigan Press.

29.

Guisinger

Smith

(2002) Honest threats: The interaction of reputation and political institutions in international crises. Journal of Conflict Resolution 46: 175–200.

30.

Hafner-Burton

Montgomery

(2006) Power positions: International organizations, social networks, and conflict. Journal of Conflict Resolution 50: 3–27.

31.

Hafner-Burton

Kahler

Montgomery

(2009) Network analysis for international relations. International Organization 63: 559–592.

32.

Henderson

Tucker

(2001) Clear and present strangers: The clash of civilizations and international conflict. International Studies Quarterly 45: 317–338.

33.

Hogbladh

Pettersson

Themner

(2011) External support in armed conflict, 1975–2009: Presenting new data. Presented at the 52nd Annual International Studies Association Convention, Montreal, 16–19 March.

34.

Huth

(1998) Major power intervention in international crises, 1918–1988. Journal of Conflict Resolution 42: 744–770.

35.

Jones

Bremer

Singer

(1996) Militarized interstate disputes, 1816–1992: Rationale, coding rules, and empirical patterns. Conflict Management and Peace Science 15: 163–213.

36.

Joyce

Braithwaite

(2013) Geographic proximity and third-party joiners in militarized interstate disputes. Journal of Peace Research 50: 595–608.

37.

Joyce

Ghosn

Bayer

(2014) When and whom to join: The expansion of violent interstate conflicts. British Journal of Political Science 44: 205–238.

38.

Kim

Barnett

(2007) A structural analysis of international conflict: From a communications perspective. International Interactions 33: 135–165.

39.

Kinne

(2013) Network dynamics and the evolution of international cooperation. American Political Science Review 107(4): 766–785.

40.

Kydd

(2003) Which side are you on? Bias, credibility, and mediation. American Journal of Political Science 47: 597–611.

41.

Kydd

(2006) When can mediators build trust? American Political Science Review 100: 449–462.

42.

Lai

Reiter

(2000) Democracy, political similarity, and international alliances, 1816–1992. Journal of Conflict Resolution 44: 203–227.

43.

Leeds

Mitchell

Long

(2000) Reevaluating alliance reliability. Journal of Conflict Resolution 44: 686–699.

44.

Leeds

Ritter

Mitchell

Long

(2002. Alliance treaty obligations and provisions, 1815–1944. International Interactions 28: 237–260.

45.

Lektzian

Sprecher

(2007) Sanctions, signals, and militarized conflict. American Journal of Political Science 51: 415–431.

46.

Maoz

(2006) Network polarization, network independence, and international conflict, 1816–2002. Journal of Peace Research 43: 391–411.

47.

Maoz

(2011) Networks of Nations: The Evolution, Structure, and Impact of International Networks, 1816–2001. Cambridge: Cambridge University Press.

48.

Maoz

(2012) Preferential attachment, homophily, and the structure of international networks, 1816–2003. Conflict Management and Peace Science 29: 341–369.

49.

Maoz

Kuperman

Terris

Talmud

(2005) International networks and international politics. In: Mintz

Russett

(eds) New Directions for International Relations. Lanham, MD: Lexington Books.

50.

Maoz

Terris

Kuperman

Talmud

(2006) Structural equivalence and international conflict, 1816–2000: A social network analysis of dyadic affinities and conflict. Journal of Conflict Resolution 50: 664–689.

51.

Maoz

Terris

Kuperman

Talmud

(2007) What is the enemy of my enemy? Causes and consequences of imbalanced international relations. Journal of Politics 69: 100–115.

52.

Marshall

Jaggers

(2002) Polity IV Project: Political regime characteristics and transitions, 1800–2002. University of Maryland: Typescript.

53.

McPherson

(2004) A blau space primer: Prolegomenon to an ecology of affiliation. Industrial and Corporate Change 13: 263–280.

54.

McPherson

Smith-Lovin

(1987) Homophily in voluntary organizations: Status distance and the composition of face-to-face groups. American Sociological Review 52: 370–379.

55.

McPherson

Smith-Lovin

Cook

(2001) Birds of a feather: Homophily in social networks. Annual Review of Sociology 27: 415–444.

56.

Melander

Moller

Oberg

(2009) Managing intrastate low-intensity armed conflict 1993–2004: A new dataset. International Interactions 35: 58–85.

57.

Melin

Koch

(2010) Jumping into the fray: Alliances, power, institutions and the timing of conflict expansion. International interactions 36: 1–27.

58.

Morgenthau

(1985) Politics Among Nations, 6th edn. New York: Knopf.

59.

Most

Starr

(1980) Diffusion, reinforcement, geopolitics, and the spread of war. American Political Science Review 74: 932–946.

60.

Most

Starr

Siverson

(1989) The logic and study of the diffusion of international conflict. In: Midlarsky

(ed.) Handbook of War Studies. Boston, MA: Allen and Unwin.

61.

Oneal

Russett

(2005) The rule of three, let it be: When more really is better. Conflict Management and Peace Science 22: 293–310.

62.

Owsiak

(2014) Conflict management trajectories in militarized interstate disputes: A conceptual framework and theoretical foundations. International Studies Review 16: 50–78.

63.

Owsiak

Frazier

(in press) The Conflict management efforts of allies in interstate disputes. Foreign Policy Analysis.

64.

Pevehouse

Nordstrom

Warnke

(2004) The Correlates of War 2 International Governmental Organizations Data version 2.0. Conflict Management and Peace Science 21: 101–119.

65.

Popielarz

McPherson

(1995) On the edge or in between: Niche position, niche overlap, and the duration of voluntary association membership. American Journal of Sociology 101: 698–720.

66.

Quackenbush

(2006) Not only whether but whom: Three-party extended deterrence. Journal of Conflict Resolution 50: 562–83.

67.

Regan

Aydin

(2006) Diplomacy and other forms of intervention in civil wars. Journal of Conflict Resolution 50: 736–56.

68.

Regan

Frank

Aydin

(2009) Diplomatic interventions and civil wars: A new dataset. Journal of Peace Research 46: 135–146.

69.

Reiter

Stam

(2002) Democracies at War. Princeton, NJ: Princeton University Press.

70.

Rosenau

(1969) Intervention as a scientific concept. Journal of Conflict Resolution 13: 149–171.

71.

Sabrosky

(1980) Interstate alliances: Their reliability and the expansion of war. In: Singer

(ed.) The Correlates of War II: Testing Some Realpolitik Models. New York: Free Press.

72.

Sartori

(2002) The might of the pen: A reputational theory of communication in international disputes. International Organization 56: 121–149.

73.

Sartori

(2005) Deterrence by Diplomacy. Princeton, NJ: Princeton University Press.

74.

Schultz

(2001) Looking for audience costs. Journal of Conflict Resolution 45: 32–60.

75.

Schweller

(1994) Bandwagoning for profit: Bringing the revisionist state back in. International Security 19: 72–107.

76.

Signorino

Ritter

(1999) Tau-B or not tau-B: Measuring the similarity of foreign policy positions. International Studies Quarterly 43: 115–144.

77.

Singer

(1988) Reconstructing the Correlates of War Dataset on material capabilities of states, 1816–1985. International Interaction 14: 115–132.

78.

Singer

Small

(1966) National alliance commitments and wars involvement, 1815–1945. Peace Research Society, Papers V: 109–140.

79.

Singer

Bremer

Stuckey

(1972) Capability distribution, uncertainty, and major power war, 1820–1965. In: Russett

(ed.) Peace, War, and Numbers. Beverly Hills, CA: SAGE.

80.

Siverson

Emmons

(1991) Birds of a feather: Democratic alliance choices in the twentieth century. Journal of Conflict Resolution 35: 285–306.

81.

Siverson

Starr

(1990) Opportunity, willingness, and the diffusion of war. American Political Science Review 84: 47–67.

82.

Siverson

Starr

(1991) The Diffusion of War: A Study in Opportunity and Willingness. Ann Arbor, MI: University of Michigan Press.

83.

Siverson

King

(1979) Alliances and the expansion of war. In: Singer

Wallace

(eds) To Augur Well: Early Warning Indicators in World Politics. Beverly Hills, CA: SAGE.

84.

Slantchev

(2005) Military coercion in interstate crises. American Political Science Review 99: 533–547.

85.

Smith

(1995) Alliance formation and war. International Studies Quarterly 39: 405–425.

86.

Smith

(1996) To intervene or not to intervene: A biased decision. Journal of Conflict Resolution 40: 16–40.

87.

Snyder

Kick

(1979) Structural position in the world system and economic growth, 1955–1970: A multiple network analysis of transnational interactions. American Journal of Sociology 84: 1096–1126.

88.

Starr

(2005) Territory, proximity, and spatiality: The geography of international conflict. International Studies Review 7: 387–406.

89.

Thyne

(2006) Cheap signals with costly consequences: The effect of interstate relations on civil war, 1945–1999. Journal of Conflict Resolution 50: 937–61.

90.

Thyne

(2009) International Relations and Civil Conflict. Lexington, KY: Lexington Books.

91.

Volgy

Corbetta

Grant

Baird

(eds) (2011) Major Powers and the Quest for Status in International Politics. New York: Palgrave Macmillan.

92.

Waltz

(1979) Theory of International Politics. New York: McGraw-Hill.

93.

Wasserman

Faust

(1994) Social Network Analysis: Methods and Applications. Cambridge: Cambridge University Press.

94.

Werner

(2000a) Deterring intervention: The stakes of war and third party involvement. American Journal of Political Science 44: 720–732.

95.

Werner

(2000b) The effects of political similarity on the onset of militarized disputes, 1816–1985. Political Research Quarterly 52: 343–74.

96.

Werner

Lemke

(1997) Opposites do not attract: The impact of domestic institutions, power, and prior commitments on alignment choices. International Studies Quarterly 41: 529–546.

97.

Yamamoto

Bremer

(1980) Wider wars and restless nights: Major power intervention in ongoing wars. In: Singer

(ed.) The Correlates of War II: Testing Some Realpolitik Models. New York: Free Press.

98.

Yuen

(2009) Target concessions in the shadow of intervention. Journal of Conflict Resolution 53: 745–773.