Title Menu - The International Journal of Peace Studies.  
Return to Table of Contents Return to I J P S Home



Jeffrey Pickering



In the analysis of interstate conflict, it is often assumed that two radically different systems have emerged in international politics.  The first is the zone of peace that cocoons the First World.  The second is the zone of turmoil in which the rest of the world labors, where interstate violence remains commonplace.  Recent research, however, argues that this dichotomous image of interstate conflict is insufficient.  Conflict revolves around regional processes to a greater degree than has often been appreciated.   In this study, logistic regression models demonstrate that the use of belligerent force varied significantly across regional subsystems over the 1946-1996 period.  Actors in some subsystems are more prone to employ belligerent force than actors in others.  The former subsystems stand less chance of developing into islands of tranquility, or nascent zones of peace, in the international system than the latter.


Over a decade ago, Kegley (1991: 8) observed the gradual and seemingly lasting development of Atwo systems [in international politics], a stable >central= system and an unstable >peripheral= system.@  The systems diverge most sharply when analysis turns to interstate peace and conflict.  A host of interlocking variables ensure peace in the central system, especially democratic governance, economic interdependence, changed popular attitudes, and a deeply ingrained fear of nuclear war.  Some scholars have even argued that the lessons of history do not apply to today=s center, because it represents a wholly new type of international order (Jervis, 1991/1992; Singer and Wildavsky, 1996).  It is a Azone of peace,@ which will remain tranquil for decades if not indefinitely.   In contrast, Aa lot of old fashioned history@ continues to grind on in the periphery (Singer and Wildavsky, 1996: 8).  The site of the vast majority of the world=s conflict over the past half century, it has been labeled the Azone of turmoil.@  The prospects for stability are often assumed to be dim in this unforgiving zone.  The international system thus appears to be segmented, with the bulk of the world=s population caught in a continuing cycle of interstate violence and retribution while a minority enjoys lasting peace. 
This dichotomous image of interstate conflict is prevalent in the literature (see Nagl, 2001).  A different perspective on the geographic spacing of conflict has nonetheless emerged in recent years.  It emphasizes the phenomenon=s local roots, arguing that unique patterns exist within different regional settings.  Advocates of this approach contend that regional variations must be fully understood to comprehend interstate violence in the late twentieth and early twenty-first centuries (Kupchan, 1998; Lake and Morgan, 1997; Solingen, 1998).  Kacowicz (1998), for example, asserts that conflict patterns in the periphery are far from uniform.  He maintains that while some regional subsystems in the developing world exhibit the types of violence associated with the zone of turmoil, others are gradually evolving toward greater stability.  They may even have the potential to develop into nascent zones of peace, with interstate relations to some degree comparable to those among center states.
This article tests these competing perspectives on contemporary interstate conflict.  It presents an empirical map of the use of military force in the world=s regional subsystems over the past five decades.  It focuses not on military force in the aggregate, but on force used over belligerent issues.  Belligerent issues have a high probability of sowing the seeds of future conflict among combatants and thus fostering broader regional instability.  The most widely studied belligerent issue is territory.  When military force is used in a dispute over land, it tends to generate lingering acrimony among the participants.  Hostilities frequently recur and conflict is often more severe than that over other issues (Hensel, 1996).  Troops deployed to oust the target government or to defeat its regional ambitions produce similar animosity, as do soldiers sent to seize control over natural resources or to undermine the target government=s foreign economic policies (see Huth, 2000).  All are considered episodes of military force being used for belligerent purposes in this article.    
Empirical analysis centers on a series of logistic regression models that employ regional subsystems as the primary independent variables and belligerent military force as the dependent variable.  If the center-periphery perspective is correct, the propensity to use force over belligerent issues should be roughly equal across the developing world.  At the very least, wide discrepancies in the use of such force would not be expected across regional subsystems.  The regional perspective assumes not only that regional differences exist, but that they are stark.  Obviously, these perspectives cannot be equally valid.  This study takes an initial step toward determining which is more faithful to the empirical record.  It also sheds light on the relative peacefulness of the world=s regional subsystems.
Before turning to the empirical record of the past fifty years, the theoretical framework of the study must be described.  The following section outlines the importance of regions and of belligerent issues.  The third section of the article operationalizes relevant variables and the fourth describes the study=s methodology.  The fifth section presents the empirical outcomes.  Conclusions are drawn in the final section. 

                                      Regional Subsystems and Belligerent Issues

Regional Subsystems
Studies of regional conflict patterns remain rare in the quantitative international politics literature (prominent exceptions include Gledistch and Ward, 2000; Lemke, 1996; Miller and Kagan, 1997).  This literature tends to focus analytically on either the system level or the dyad level.  It is increasingly apparent that neither is sufficient (see Crescenzi and Enterline, 1999; Thompson, 1996).  For one, regional dynamics vary dramatically across the international system and they cannot simply be deduced from global trends or power configurations.  Given the importance of contiguity for understanding patterns of international conflict, and the tendency for wars to spill over into nearby states, it should not be surprising that regions offer a useful analytical vantage point for the study of armed force (see Bremer, 1992; Siverson and Starr, 1991).  Unique regional patterns can be expected to emerge from the interaction of states within a common neighborhood, and this local context has increased in importance since the end of the Cold War.  Lake and Morgan (1997: 5) summarize this point succinctly, asserting that, AIn the foreseeable future, violent conflicts will mostly rise out of regional concerns and will be viewed by political actors through a regional, rather than a global, lens@ (see also Huth, 1996: 5; Miller, 2001: 199).
As this statement intimates, regions cannot be considered a sum of their dyadic parts either.  The probability of synergy in regional interstate interactions has long been appreciated, since at least the first studies of security communities by Deutsch and his associates (1957).  Different types of regional synergies can, of course, be envisioned.  One possibility might entail positive interactions that build on one another.  Regional actors might implement a series of confidence building measures that gradually transform the area into a stable zone of peace.  In this scenario, local states would no longer consider military force and especially war tolerable means of handling disputes.   A starkly different pattern will be present in regions where armed force is used to settle interstate disputes.  Cross-border violence may increase tensions among neighboring states and may lead to arms racing, long-term rivalry, and even war.  Whatever path a given regional subsystem takes, the central point is that interaction among states in the local context can, and often does, evolve into something more than can be understood by analyzing the dyadic connections among neighboring actors (Gleditsch and Ward, 2000; Kacowicz, 1998).

Belligerent Issues
Issues, or as they are often labeled, contentious issues, are defined as Awhat states choose to fight over@ (Diehl, 1992: 333).   International relations scholars increasingly recognize that Aissues are an essential part of international conflict@ that must be included in any convincing explanation of conflict processes (Mitchell and Prins, 1999: 172).  As Huth (2000: 96) explains, Aa focus on the kinds of issues in dispute proves very helpful . . . in understanding the reasons that state leaders might risk military conflict.@  This article examines how likely actors in different regional settings are to risk military conflict with their neighbors.  It does this by estimating their willingness to use military force over belligerent issues.
Belligerent issues are defined as those that pose a direct military challenge to the target government.  When armed force is used over such issues, it threatens the government=s vital national interests or its existence.  Territory, important natural resources, government legitimacy, and influence in the region are examples of the belligerent issues analyzed in this study.  All are highly salient (see Huth, 2000) and can produce hostile disputes among regional actors.  In the analysis that follows, attention is given both to belligerent issues as a general category and to perhaps the most irascible belligerent issue:  territory.  This is because, as Vasquez (1995: 285) asserts, Aterritorial issues are not like other issues.  They are special in their volatility and ability to give rise to collective violence.@  Territory has, in fact, been labeled the Amost contentious issue in world politics@ (Mitchell and Prins, 1999: 179).  Military force used over territory is significantly more likely to be reciprocated than military force used over other issues.  It also stands greater chance of escalating into more severe forms of conflict and of devolving into a pattern of recurrent interstate violence, even an enduring rivalry (Hensel, 1996; see also Huth, 1996, 2000; Kocs, 1995; Vasquez and Henehan, 2001).
This article compares the use of belligerent force across regional subsystems to determine if the conventional wisdom on the zone of turmoil is valid.  It will simultaneously provide an empirical map of any significant variation that exists across regions.  If regions are found to diverge sharply in their use of force over belligerent issues, it speaks volumes about how peaceful those locations have been over the past fifty years and how likely they are to be in the future.  It seems logical that regions where major actors have largely settled disputes over belligerent issues stand a greater chance of developing into pockets of stability in the international system than those that have not (see Kacowicz, 1998; Thompson, 1996).
To date, no study has analyzed actors= propensities to use military force over highly salient, belligerent issues in different geographic settings.  This article employs logistic regression models to estimate such proclivities.  It also provides regional frequency counts of belligerent deployments of force.  Of course, issues are not static.  Issues in dispute can not only explode into violence, they can be resolved peacefully.  Inevitable changes in issues= contentiousness and in the priority they are given by policymakers in different locales makes the analysis of a five decade time span difficult. Consequently, this article compares the use of military force over belligerent issues both across regions and across time.


                                           Operationalization and Measurement

Four pieces of information are required in this study.  Measures are needed for interstate military force, belligerent and territorial issues, regional subsystems, and the temporal demarcations used in this analysis.  Each is discussed separately.
One of the few interstate conflict data sets that codes contentious issues is used to operationalize both interstate force and belligerent issues (Diehl, 1992).  The Pearson-Baumann International Military Intervention data set records all verifiable instances when Atroops or forces . . . cross[ed] borders . . . in pursuit of political or economic objectives,@  or when forces already based in a country were employed for such objectives (Pearson, Baumann, and Pickering, 1994: 209).  In essence, it catalogs instances when national military personnel are purposefully and overtly dispatched into other sovereign states.  Originally spanning the years 1946 to 1988, the data set has recently been extended to 1996 (Pickering, 1999).  This study focuses on intervention initiations. 
Although the Pearson-Baumann collection is relatively unique in coding contentious issues, it does have one venial limitation.  Conceptual ambiguities continue to plague the term intervention itself.  It is often used to refer to any interference by one international actor into the affairs of another, be it by diplomatic, economic, or military means (Bull, 1984).  Given such problems with nomenclature, the events tallied by Pearson and Baumann are labeled the interventionary use of force in this article (see Pickering and Thompson, 1998).  The phenomenon being measured and cataloged remains the same.
Motivating issues are coded for 825 of the 827 interventions included in the data set.  For most interventions, multiple issues are listed.  In the Pearson-Baumann conceptualization, states use military force over one or more of the following eight issues: 1) insurgent issues, typically the pursuit of rebels across borders; 2) humanitarian issues, such as providing material aid to beleaguered or persecuted populations; 3) diplomatic and/or military protective issues, or protecting military and/or diplomatic property and personnel in foreign lands as well as nationals in such lands; 4) economic issues, or the protection of economic interests, such as preserving access to essential natural resources; 5) domestic issues, or those related to domestic disputes in the target state, with force most commonly being used in support of groups opposing the government; 6) ethnic/social protective issues, or protecting a social or ethnic minority in a state, often from the target state=s military apparatus; 7) strategic issues, such as regional power balancing or using force to preserve regional stability; and, finally, 8) territorial issues. 
These issues are not ranked in terms of their relative belligerence.  Such an exercise would be both awkward and subjective.  Rather, they are dichotomized to create the primary dependent variable, the use of belligerent force.  Only those issues that directly challenge the target government, its fundamental economic interests, or its territory are considered belligerent issues.  It is assumed that if the vital interests of the target government are challenged, such as the tenure of the government itself, the likelihood that force will be reciprocated increases.  And, if force is reciprocated, the conflict necessarily becomes more severe, and has a higher probability of broadening (see Jones, Bremer, and Singer, 1996; Siverson and Starr, 1991).  Regional stability will most likely suffer as a consequence.
The last four issues in the list provided above are classified belligerent issues.  All directly challenge the target government by employing force either to: provide military support to its domestic rivals; protect minorities in the target country; work against it in regional power balancing; or challenge it over territorial issues.  One further issue is included within this grouping.  If an intervention is undertaken for economic reasons, and it clashes directly with the target government=s policies on this issue, it is categorized a belligerent use of force.  Even though countries have become less willing to use armed force over economic issues in this century, attempts to gain control over critical resources still generate considerable hostility.  This is especially the case in the developing world where control over essential resources may be equated with national survival (Holsti, 1991: 285, 316).
On the other hand, if armed forces are deployed to support the target government=s economic interests, the intervention is coded non-belligerent.  The remaining three issues in the Pearson-Baumann categorization also fall into the non-belligerent category.  Pursuing rebels across borders is a transgression, to be sure, but not one typically designed or perceived to confront the target government.  Forces used to protect or to evacuate diplomatic or military personnel in non-combat situations are rarely of sufficient scale to pose a serious threat to the target government, and are rarely perceived as such.  Finally, humanitarian missions do not have hostile objectives, and are seldom considered to be the same type of affront to the target government as the belligerent issues listed above.  Although there will always be exceptions, it would be unusual for troops deployed over these non-belligerent issues to be considered a direct military challenge to the target government.  Consequently, the primary dependent variable, belligerent force, is coded A0@ when military force is used over one of these issues.   It is coded A1@ when force is used over at least one of the belligerent issues listed above.   The second dependent variable analyzed is also dichotomous.  If a use of interventionary force is triggered by territorial concerns, the dependent variable territorial force is coded A1.@  It is coded A0@ otherwise.
 The second preliminary task is to operationalize the independent variable, regional subsystems.  At first glance, demarcating regional boundaries seems an easy task.   It is, after all, commonly accepted that regional subsystems consist of clusters of geographically proximate states that tend to maintain substantial security, economic, and political relations (Buzan, 1991: 188; Kacowicz, 1998: 8; Kaiser, 1968: 86).  But, delineating the exact spatial boundaries of subsystems can be daunting, particularly given the time span of this study.  Regional boundaries have the potential to shift over time, as actors redefine bilateral relationships and their memberships in international coalitions change (Solingen, 1998).  To maintain consistency across this article=s five decade long sample, Thompson=s (1973) somewhat more relaxed conceptualization of the regional subsystem is employed.  In the Thompson formulation, a subsystem must, at a minimum, possess a common sense of neighborhood or propinquity.  More importantly, although it is recognized that regional boundaries could change slightly over lengthy spans of time, fairly regular, and sometimes dense, political and economic linkages are thought to keep actors in such neighborhoods connected to one another.  The perception of being bound within the same geopolitical locale is expected to endure.  The memberships of the thirteen regional subsystems analyzed in this study are identified in Appendix A.  When analyzing belligerent force in regional subsystems, both actions taken by regional actors and any uses of force within the subsystem by external actors are included.

Finally, temporal distinctions must be made.  This article=s periodization was designed to meet two criteria.  First, roughly equal periods of time were sought, since comparable segments allow for easier  interpretation.  Second, given the dramatic geopolitical changes of the last decade, the Cold War and the post Cold War periods were differentiated.  Both criteria were met by slicing the forty-four Cold War years included in the Pearson-Baumann collection into three approximately equal fifteen year segments: 1946-1959; 1960-1974; and 1975-1989.   The final period analyzed, the post Cold War era, is dated from 1990-1996.  Roughly half the length of the Cold War time periods, it remains readily comparable.  Except for the 1990 break-point, none of this study=s temporal markers represent watersheds in international politics.  Rather, this periodization has been fashioned to facilitate the comprehension of trends in belligerent and territorial force without the variability associated with annual or decadal measures.  To simplify their presentation, these four time periods are designated t1 (1946-1959), t2 (1960-1974), t3 (1975-1989), and t4 (1990-1996).


Logistic regression is employed because the dependent variables are dichotomous.  Also, this method has the advantage of placing as much emphasis on the effects of the independent variables as predicting the value of the dependent variable.  This study focuses on comparison across regions and temporal domains C that is, its principal concern is trends and variations among the independent variables.  Moreover, the logistic model has the practical attraction of offering a precise interpretation of independent variable effects.  This is accomplished by calculating the relationship between the variables as an odds-ratio.  In the logistic model, odds-ratios indicate the impact that an independent variable has on the dependent variable in comparison to a reference category.  Neutral reference categories are added to the logistic models in this study.  To analyze belligerent force, for example, a reference category that is evenly divided between belligerent and non-belligerent uses of force is included in the model.  It makes it easy to determine whether actors in a subsystem have a greater propensity to use belligerent or non-belligerent force.  Essentially, an odds ratio of 1 in this paper indicates that the subsystem in question (the independent variable) has an effect on belligerent force (the dependent variable) that is similar to the reference group.  Since the reference category is evenly divided between belligerent and non-belligerent uses of force, this means that interveners within that subsystem are statistically as likely to use belligerent as non-belligerent force.  An odds-ratio larger than 1 illustrates how much greater the odds are that the interventionary force used within that subsystem is belligerent.  For example, an odds ratio of 2.5 indicates that forceful interactions in that region are two and a half times more likely to involve belligerent than non-belligerent force.  If, on the other hand, the odds-ratio is below 1, non-belligerent uses of force are more likely than belligerent actions.  The odds-ratio is not affected by shifts in sample size and, by offering a readily interpretable comparison of the extent that force is motivated by belligerent or non-belligerent issues in different subsystems, it underscores the practical, substantive significance of the findings.


Reviewing the Empirical Evidence

This study=s findings are presented in three stages.  First, patterns of belligerent force and territorial force are traced for the international system=s center.  The center offers a useful test case for the issue dichotomy analyzed in this article.  If the center has grown more peaceful over the past half-century as is often suggested, center states= use of belligerent and territorial force should be decreasing.  Second, the proportion of interventionary uses of force launched over belligerent issues and over territory are mapped across regional subsystems in a series of charts.  These figures provide vivid images of the differences in the use of belligerent force across regions.  Third, logistic regression models estimate the statistical significance of these regional patterns, and whether the proportion of forceful interactions over either belligerent or territorial issues has changed significantly over time.
The utility of the issue dichotomy used in this study is apparent in Figure 1.  This figure illustrates the percentage of center uses of force motivated by non-belligerent and territorial issues.  Presenting non-belligerent rather than belligerent force makes the interpretation of Figure 1 and the graphs to follow easier.  As expected in the literature on the emergence of a zone of peace in the center, non-belligerent uses of force have been steadily rising over the past half-century (as a percentage of all interventions coded for belligerent force).  Since belligerent scores are by definition the inverse of non-belligerent outcomes (that is, the non-belligerent result of 87% for the center from 1990-96 means that 13% of reported interventions were over belligerent issues), belligerent force has correspondingly been on the wane.  When center states use military force, it is increasingly less likely that they do so over issues that directly challenge the recipient state.  In fact, almost one half (47%) of center military interventions in the post-Cold War era were motivated by humanitarian concerns, nearly all of which were supported by the host government.  Beyond this, First World states are also becoming less embroiled in squabbles over the most pugnacious issue in world politics, territory.  As Figure 1 illustrates, center deployments of interventionary force over territorial issues have trended downward since 1946, to the point that only 5% of center military interventions were driven by territorial issues from 1975-1989, and none were after 1990.  It must be emphasized that this figure charts all interventions by center actors, not just those that occurred within the center subsystem.  An analysis of the latter would include only two cases B the Cod War episodes between Britain and Iceland in 1958 and 1975.

Figure 1.

The fairly stark decline in belligerent and territorial deployments of force underscores that the center has not only become a place where force is rarely, if ever, used locally, center actors have also become much less likely to use bellicose military force abroad.  Their military actions therefore stand less chance of breeding resentment within the target country or of producing larger conflicts.  In short, center military deployments appear to have become more cautious and perhaps less conflict-inducing.
The reasons for this evolution are both numerous and largely beyond the scope of this article.  Nonetheless, a number of potentially overlapping explanations are evident.  The reluctance of the most prolific center intervener, the United States, to accept significant casualties in its post-Vietnam military deployments may partly explain the inclination to avoid belligerent force.  Particularly in the post-Cold War era, the U.S. has used unilateral military force only when it has had the ability to draw on overwhelming numbers of troops, not to mention sophisticated firepower (Pickering, 1999).  The domestic political costs of military missions abroad are also becoming more salient for center decision-makers.  Rally round the flag surges in support of military ventures tend to evaporate rapidly, and military actions that fail to produce quick and bloodless results can prove costly in later elections (Cotton, 1986; Gelpi and Grieco, 1998).  Concerns over casualties and battlefield embarrassments have only been heightened by the steady growth in the conventional and the non-conventional arsenals of developing countries over the past decade.
Figures 2-6 display patterns in the use of non-belligerent force in other regional subsystems.  It should be noted that some subsystems register too few uses of force for accurate graphing in the early Cold War period (when many states remained colonial appendages) and in the shorter post-Cold War period.  A mere glance at these figures is enough to see that patterns do diverge, sometimes markedly.  With the exception of South Asia, the proportion of military interventions motivated by non-belligerent issues was much lower in Asia and South America than in Africa.  The Middle East lies somewhat in between the other regions on this score, while the Eastern European figures vacillate wildly, a finding surely related to the political and economic transformations that have swept across this region the last decade and a half.  Similar results emerge in Figures 7-11, which portray the proportion of uses of force that were motivated by territorial issues.  South America and East Asia again stand out.  Both regions have a high percentage of interventions motivated by territorial concerns, although the South American pattern fluctuates considerably.  In contrast, actors in Central America, most all African subsystems, especially Central and Southern Africa, and to a lesser extent Eastern Europe and the Middle East use territorial force infrequently.
The prevalence of belligerent and territorial force in South America is explained by two well-documented rivalries over disputed territories.  The first is between Ecuador and Peru over their Andean border; the second is between Argentina and Britain over the Falklands/Malvinas Islands.  The rivalry between Ecuador and Peru accounts for nearly thirty-one percent of all uses of force within South America over the past fifty years, and that between Argentina and Britain accounts for a further nine percent.  Little wonder that when looking at the proportion of interventions that are generated by belligerent issues or territory that South America appears a rather bellicose regional subsystem.  Any such claims must be put into perspective, however.  Although the ratio of force used over belligerent and territorial issues to that used over non-belligerent and non-territorial issues is high in this subsystem, the overall rate of intervention is fairly low.  South America ranks roughly in the middle of the thirteen subsystems studied for the total number of belligerent interventions and the number of belligerent interventions per actor from 1946 to 1996.  Unresolved territorial disputes also fuel the bellicose use of force in a second contentious subsystem, East Asia.  The sources of these disputes are well known, namely the Chinese claim on Taiwan, the Korean divide, and the tensions that persist over island groupings in the South China Sea such as the Spratleys.  Over forty-four percent of all uses of force initiated by East Asian actors over the past half-century have involved China and one or more of its neighbors in a territorial dispute.
Other subsystems seem to be evolving away from the use of force over belligerent issues.  Sub-Saharan Africa is noticeable in this regard.  It has comparatively low proportions of both belligerent and territorial interventions.  The reasons for this seemingly propitious finding are evident.  One is that a relatively high percentage of external interventions into this region have been supportive, often taking the form of humanitarian operations to limit the effects of civil strife or natural disaster.  Perhaps more important is the norm against hostile intervention that was deeply embedded within most sub-Saharan African states in the years following independence.  This norm developed in response to the inherent fragility of many of the states in the region.  Decision-makers have been reluctant to use armed force across borders because doing so might result in external meddling among divisive factions at home -- a potentially explosive situation.  Thus, the volatile issue of territory has often been removed from interstate disputes in subSaharan Africa (see Jackson and Rosenberg, 1982). There is, however, evidence that the non-intervention norm may be eroding.  The increase in territorial force in t4 (see Figure 8) illuminates this possibility, as does the intervention of local states into the ongoing Congolese civil war (which began after 1996 and hence is not included in this analysis).  
Notable trends away from belligerent force are also present in Central America and Eastern Europe.  The use of force over territory is rare in the former, while the latter has witnessed some decline in the proportion of belligerent and territorial interventions since 1975.  Of all subsystems, the Middle Eastern pattern is perhaps least anticipated.  Actors in the Middle East have been less motivated by territorial issues than counterparts in a number of other regions, and they have employed comparatively low levels of belligerent force since 1975.  The absence of Arab-Israeli warfare and the increase in supportive center missions (notably during the Persian Gulf War) help explain the rise in non-belligerent force in t3 and t4.  Conspicuous patterns are not evident in other subsystems.
Figures 1-11 highlight regional trends in belligerent and territorial force, but they provide no indication of how significant regional variations are, or whether the propensity to intervene over belligerent issues or over territory has altered in the past half-century.  The logistic regression models in Table 1 provide initial answers to these questions.  To measure the variation among regional subsystems, these subsystems are employed as independent variables.  Belligerent force and territorial force are the dichotomous dependent variables.  As noted above, subsystems are not compared directly to one another.  They are compared to neutral reference groups.  In Model 1, the

Figures 2-6.  Percent of Military Interventions Motivated by Non-Belligerent Issues


  Figure 2.                                                                                        Figure 5.


 Figure 3.                                                                                          Figure 6.


 Figure 4.


Figures 7-11.   Percent of Military Interventions Motivated by Territorial Issues

    Figure 7.                                                                                         Figure 10.

    Figure 8.                                                                                         Figure 11.


    Figure 9.


reference group is split evenly between uses of force over belligerent and non-belligerent issues, while in Model 2 it is equally divided between uses of force over territorial and non-territorial issues.  These reference categories make it easy to determine if actors in a

subsystem have a greater proclivity to use belligerent as opposed to non-belligerent force and territorial as opposed to non-territorial force.  Independent variables t1 - t3 correspond to the time periods discussed above. T4 is used as the reference point for this set of indicators.
Regional variation is immediately apparent in Model 1.  Five of the thirteen subsystems analyzed have a statistically significant relationship with the dependent variable while the remaining eight do not.  Actors in the former have been more likely to use military force for either belligerent or non-belligerent issues (with odds ratios above or below one respectively), while actors in the latter have deployed troops for belligerent and non-belligerent reasons in roughly equal measure.   In other words, as was expected, some regional subsystems are prone to belligerent force, others are averse to it, and still others have no significant measurable tendency in either direction.
Model 1 confirms many of the regional trends in the use of belligerent force that are apparent in Figures 1-6.  Center states rarely use belligerent force.  Actors in the zone of peace are seventy-seven percent more likely to employ troops in a non-bellicose manner than a bellicose manner, a finding that is statistically significant.  As could be anticipated from the preceding figures, two subSaharan African subsystems have significant negative coefficients.  When force is employed by Central African actors or within that region, it is 75% more likely to be non-belligerent than belligerent.  The result for Southern Africa is 68%.  Yet, there are statistically significant findings that run in the other direction as well, underscoring those regions with a high propensity to use belligerent force.  South American interventions are nearly four times more likely to be belligerent than the reference group.  This finding is mitigated by the fact that interstate force is used relatively infrequently in the South American subsystem, as mentioned previously.  The results for East Asia are more sobering.  States in this subsystem are almost six times more likely to employ belligerent than non-belligerent force, a result that is highly statistically significant.

Table 1. Logistic Regression Estimates of Regional and Temporal Variation in Belligerent and
 Territorial Force 

                Model 1                                                    Model 2

         Belligerent Force                                     Territorial Force
Variables                              Coefficient            Odds Ratio                 Coefficient                 Odds Ratio  

Center                                   -1.45***                .23                         -1.56***                  .21
                       (.26)                                                     (.26)
Central America                       .04                    1.04                         -1.65***                  .19
                       (.50)                                                     (.50)
South America                       1.40*                  4.04                            .90                      2.46
                       (.65)                                                     (.52)
West Africa                             .12                    1.13                           -.77                        .46
                       (.53)                                                     (.50)
Central Africa                       -1.38**                  .25                         -1.02*                      .36
                       (.48)                                                     (.43)
East Africa                             -.53                      .59                           -.71                        .49
                       (.52)                                                     (.47)
Southern Africa                                -1.13*                 .32                                            -2.18***       .11
                       (.58)                                                     (.68)
Northern Africa                                  -.10                   .90                                              -.39             .68
                       (.49)                                                     (.38)
Middle East                            -.03                      .97                           -.36                        .70
                       (.39)                                                      .32
East Asia                                1.78***              5.96                            .67*                    1.94            
                       (.50)                                                     (.35)
South Asia                              -.57                      .57                           -.32                        .73
                       (.44)                                                     (.37)
Southeast Asia                        -.57                      .57                           -.09                        .91
                       (.44)                                                     (.37)
Eastern Europe                                    .68                   1.98                                            -.97*           .38
                       (.46)                                                     (.40)
t1                                             .59                    1.81                            .86**                  2.37
                       (.35)                                                     (.33)
t2                                             .64*                  1.90                            .49                      1.63
                       (.32)                                                     (.30)
t3                                             .28                   1.33                            .32                      1.38
                       (.32)                                                     (.31)
Constant                                 -.21                                                      -.64
                       (.45)                                                     (.3)
Chi-square                         114.70***                                           128.45***
Log-likelihood                  -301.27                                                -397.51
% Correctly Predicted               68%                                                   76%
N                                             5281                                                     811

* p < .05; ** p < .01; *** p < .001.  Two-tailed tests.  Standard errors in parentheses.

1 Model 1 Ns are lower because a number of interventions could not be accurately coded as either belligerent or non-belligerent.  The most numerous examples are interventions that have multiple, conflicting  issue codings.   Also, the information on sixteen confirmed interventions was too ambiguous to determine if they were motivated by territorial issues.

                Many of these same regions remain noteworthy when the analysis turns to territorial force.  Model 2 in Table 1 displays the outcomes.  Again, states in the center are much less prone to use force over territory, as are states in Central Africa, Southern Africa, Central America, and Eastern Europe.  All are at least 64% less likely to use force for territorial reasons than the neutral reference category.   Territorial issues have a 79% lower probability of motivating First World actors, while the result for Southern Africa is a striking 89%.   Only one subsystem produces a coefficient that is both positively signed and statistically significant.  Actors in East Asia are nearly twice as likely to deploy military force over territorial issues than the reference group.  The territorial disputes that haunt this subsystem have been described, but their significance for the future stability of the region cannot be overemphasized.  If territorial and belligerent issues continue to animate East Asian conflicts, the region may move even further from the stability and relative tranquility that characterizes the zone of peace, and may potentially characterize other peripheral subsystems in the future.
Temporal variations are apparent as well.  Belligerent force is almost two times more common in t2 than t4, a statistically significant result.  Also, at the .10 level of statistical significance, belligerent force is significantly higher in t1 than t4.  Territorial force is statistically more common in t1 than in t4, and at the .10 level, t2 varies significantly from t4 as well.  These results provide modest evidence that force was used less often for belligerent purposes and over territory in the latter half of the sample.  In other words, the likelihood of military force being employed in a manner that will prompt a forceful response, and possible conflict escalation, has decreased slightly over the past half-century.
Models 1 and 2 fit the data reasonably well, as the model chi-square results and the percentage correctly predicted indicate.  Moreover, likelihood ratio tests (King, 1989) show that the insignificant independent variables in both models can be safely omitted without affecting the remaining statistically significant relationships.  That is, the likelihood ratio test comparing Model 1 to a reduced model that excludes the independent variables for Central America, West Africa, East Africa, Northern Africa, the Middle East, South Asia, Southeast Asia, Eastern Europe, and t1 and t3 is not statistically significant at the .05 level.  Similarly, the likelihood ratio test comparing Model 2 to a reduced model omitting South America, West Africa, East Africa, Northern Africa, the Middle East, South Asia, Southeast Asia, t2 and t3 produces a statistically insignificant result at the .05 level.
Tables 2 and 3 present logistic regression models for each of the four time periods studied.  The findings generally mirror those in Figures 1-11 and Models 1 and 2 in Table 1.  As can be anticipated, though, the results are more varied when models are run for each time period under study.  For example, Table 2 confirms that the Center, Central Africa, and Southern Africa are subsystems where non-belligerent force has at times been significantly more likely to be employed than belligerent force, while the opposite is true

Table 2. Significant Results of Logistic Regression Estimates of Regional Variation in Belligerent Force

              by t   

        t1t2                                 t3                            t4       

Variables                              1946-1959            1960-1974                   1975-1989                 1990-1996  

Center                                -37.47***                                                                          -2.27***      
                     (1.10)                                                     (.57)
Central America                                                                                                              
South America                                                  2.59**
West Africa                                                      2.99*                         
Central Africa                                                                                   -2.37**
East Africa                              

Southern Africa                                                                                                            -1.88*
Northern Africa                      

Middle East                        -36.87***              2.54**                                                                       -2.50*
                       (.79)                    (.97)                                                    (1.22)
East Asia                                                           2.43*
South Asia                                                        2.32*                                                     2.54*
                                                (1.01)                                                    (1.33)
Southeast Asia                    -38.44***
Eastern Europe                                   

Constant                              37.34***             -2.24*                          .87                    -2.73*
                       (.72)                    (.96)                          (.84)                   (1.19)
Chi-square                           55.78***            26.17**                                                 35.90***       16.69*
Log-likelihood                    -39.04                 -80.81                        -90.08                 -44.50
% Correctly Predicted          77.55                  68.84                        71.60                    71.91

N                                               981                       138                              169                           89

* p < .05; ** p < .01; *** p < .001.  Two-tailed tests.  Standard errors in parentheses.

1 Ns reported in different time periods (t1-t4) mirror the total N=s for the given periods, adding a previously explained neutral control group.  In addition, likelihood ratio tests were run for each of the above models (t1-t4), all of which indicated that non-significant variables can be safely excluded from the models (that is, none of the likelihood ratio results were significant at the .05 level).  

Table 3. Significant Results of Logistic Regression Estimates of Regional Variation in Territorial Force     

             by t   

t1                           t2                                 t3                           t4        

Variables                              1946-1959             1960-1974                   1975-1989                1990-1996  

Center                                  -1.54***             -1.60***                   -1.92***                 
                       (.45)                    (.49)                          (.59)
Central America                                                                               -2.17**                                     
South America                                                                                   1.80*                    5.01**
                                                                                   (.84)                   (1.80)
West Africa                                                     -2.31*                                                     3.26*
                                                (1.19)                                                    (1.39)
Central Africa                                                  -2.80**                                                    
East Africa                                                                  

Southern Africa                                                                                                            -2.95**
Northern Africa                                                 
Middle East                                                                                                                    3.47**
East Asia                                                                                             
South Asia                                                                                                                      3.32*
Southeast Asia                                                 -1.61*
Eastern Europe                                                            -2.42**                                                            2.36**
                                                  (.93)                                                    (1.40)
Constant                                  .33                      .60                           -.60                    -4.32***
                       (.75)                    (.60)                          (.66)                   (1.31)
Chi-square                           19.60*                46.16***                  76.74***              15.56
Log-likelihood                    -92.21               -131.60                            -94.69             -43.63
% Correctly Predicted          66.67                  73.56                        85.06                    78.95

N                                             1531                       261                                     262                          95

* p < .05; ** p < .01; *** p < .001.  Two-tailed tests.  Standard errors in parentheses.

1 Ns reported in different time periods (t1-t4) mirror the total N=s for the given periods, adding a previously explained neutral control group.  In addition, likelihood ratio tests were run for each of the above models (t1-t4), all of which indicated that non-significant variables can be safely excluded from the models (that is, none of the likelihood ratio results were significant at the .05 level).

for East Asia and South America.  But it also highlights statistically significant results absent from Table 1.  In specific time periods, actors in West Africa, the Middle East, and South Asia have been significantly more prone to use belligerent force than the neutral control group.  While such findings are intriguing, they are not highly statistically significant and they do not indicate a consistent trend toward belligerent force in these regions.  The light that such results shed on the propensity of actors in these regions to use belligerent force is consequently limited.  Table 3 similarly supports the evidence from Table 1 that the Center, Central America, Southern Africa, and Eastern Europe have been at times significantly less likely to use force over territory than the neutral reference category.  Table 3 also seems to highlight one notable finding that Table 1 conceals.  It appears that territorial force increased significantly in many peripheral subsystems in t4, the post-Cold War era.  Five subsystems report positive, statistically significant coefficients.  It must be emphasized, however, that the logistic regression model for t4 in Table 3 does not fit well, as the model chi-square underscores.  These findings must be treated with caution.     


An Alternative Measure:  The Volume of Belligerent Force

The models in Tables 1-3 estimate the proclivity of actors in different regions to use belligerent rather than non-belligerent force and territorial rather than non-territorial force.  This measure has limitations.  Most notably, a subsystem might experience a large number of belligerent and non-belligerent interventions simultaneously.  Hostile acts could cause a surge in non-belligerent interventions, to shore up allies= borders, to protect minorities, to evacuate nationals, or to provide safe passage for refugees.  Non-belligerent interventions could likewise spark belligerent attacks.  The ratio of belligerent to non-belligerent interventions might be low in these circumstances even if disputes over belligerent issues regularly turn violent in the region.  It would thus be wise to examine the total volume of belligerent force used in regional subsystems in addition to this ratio.
Table 4 presents regional frequency counts for the past five decades.  Regions are listed from those with the least belligerent interventions per regional actor to those with the most.  The results correspond to those presented in Table 1.   The Center, Southern Africa, Central Africa, and Central America are among the regions where actors have been least likely to deploy hostile military forces into neighboring countries.  East Asia is among the regions where belligerent force has been used most often.
An examination of intervention frequencies does produce one notable result absent from Tables 1-3.  Over the past five decades, actors in Southeast Asia and the Middle East have used belligerent force more often than their counterparts in East Asia.  This outcome is masked in the graphs and the logistic regression models presented in this article because non-belligerent interventions were prevalent in the former regions.  In

                        Table 4. The Volume of Belligerent Force Used in Regional Subsystems, 1946-1996

                   Belligerent Uses of Force                 Belligerent Uses of Force
                        Subsystem                           Per Actor                                            Total N

                        Center                                      .08                                           2                     
                        Southern Africa                       .26                                           4
                        Central Africa                          .27                                           3
                        Northern Africa                       .40                                           2         
                        Central America                       .41                                           9
                        West Africa                             .50                                           7
                        South America                         .66                                           8         
                        Eastern Europe                        .73                                           19
                        East Africa                               1.16                                         7
                        South Asia                               1.20                                         16
                        East Asia                                  1.70                                         11
                        Southeast Asia                         2.10                                         21                   
                        Middle East                              2.67                                         19

other words, force has been used for a multitude of reasons in these areas, whereas it was deployed primarily for hostile purposes in East Asia.  If future research finds that the high levels of non-belligerent force recorded in Southeast Asia and the Middle East were largely in response to the threats posed and the humanitarian emergencies created by belligerent violence, these locales must also be considered among the least stable of the past fifty years.  



A regional approach provides a great deal more nuance, and is more faithful to the empirical evidence, than a framework that distinguishes center from periphery.  The center-periphery dichotomy conceals the fact that some regional subsystems in the periphery are less prone to belligerent force than others, and some of these appear to be becoming even less so over time.  These areas may be slowly evolving past the types of bitter conflicts that result in recurrent interstate force.  Central America and Southern Africa are prominent examples, for force is rarely used over territorial issues in these regions, and belligerent uses of force have been uncommon in Southern Africa.  At least one regional subsystem seems to be headed in the opposite direction.  Actors employ belligerent force much more frequently than non-belligerent force in East Asia, and they have recorded a comparatively high number of belligerent interventions per actor over the past fifty years.  This subsystem seems likely to be a tension-filled pocket within the current international system, with lingering animosities over basic issues like territory and considerable potential for future conflict.
Such variance across regional subsystems is the most important finding of this study.  Tracing regional differences provides insight not only into the present stability of the world=s subsystems, but also their prospects for the future.  What=s more, doing so demonstrates that while umbrella concepts like the periphery and the zone of turmoil capture an important kernel of truth concerning conflict processes across the globe, they also have inherent liabilities.  The use of such blanket concepts limits our ability to explain and predict significant regional variation.
Further research is needed to provide a more detailed map of the use of force across regions.  At a minimum, such studies should control for a range of variables that have the potential to influence decisions to intervene.  States= military capabilities, alliances, economic strength, and  regime type might be among the variables considered.  Regional characteristics could also be included, such as the concentration of power in the region, the average age of the states in the region, and the total number of alliances there.  Future studies might analyze the military force that non-state actors, such as rebel and mercenary forces, use across borders as well.  Much remains to be done.  Nonetheless, this article provides a first cut at understanding patterns of military force across regions.  It demonstrates that significant regional differences do exist, and at times are marked.
Moreover, an appreciation of regional variance may grow increasingly important for the study of conflict in the first decades of the new century.  The propensity to use force over belligerent issues and over territory seems to have declined over the past five decades in the international system as a whole.  Given the fact that overall levels of interventionary force have remained steady over that same period (Pickering, 1999), this finding provides at least tentative grounds for optimism.  If belligerent and territorial force continue to wane in some of the world=s more stable regional subsystems, the probability that new, embryonic zones of peace might emerge necessarily increases.  Other considerations will certainly be important in the process of forging regional stability.  But, if present trends persist, a handful of peripheral subsystems may well continue along paths that diverge from our standard conception of the zone of turmoil.  Differences in the conflictual behavior of regional subsystems will thus become even more glaring.  In such an international context, regional analysis will be not only important for a thorough understanding of global conflict patterns, it will be essential.


Appendix A

States have been assigned to regional subsystems primarily on a subjective basis as developed in Pearson and Bauman (1989), Pearson, Baumann, and Pickering (1994), and Pickering and Thompson (1998).  A few states not found in the Pearson and Baumann (1993-94) data collection’s codebook have been added.  The subsystemic assignments are more likely to be overly inclusive rather than overly exclusive.  In other words, several of the subsystems could be further subdivided into smaller regional systems.

Center- Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Japan, Luxembourg, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, Turkey, U.K., and U.S.
Central America - Cuba, Haiti, Dominican Republic, Jamaica, Trinidad/Tobago, Barbados, Mexico, Guatemala, Honduras, Nicaragua, Panama, El Salvador, Costa Rica, Bahamas, Dominica, Grenada, St. Lucia, St. Vincent, Antigua, St. Christopher (Kitts)-Nevis, Anguilla, and Belize
South America - Columbia, Venezuela, Guyana, Ecuador, Peru, Brazil, Bolivia, Chile, Paraguay, Argentina, Uruguay, and Suriname
West Africa - Gambia, Mali, Senegal, Benin, Mauritania, Niger, Ivory Coast, Guinea, Burkina-Fasso (Upper Volta), Liberia, Sierra Leone, Ghana, Togo, Nigeria, Guinea-Bissau, and Cape Verde
Central Africa - Cameroon, Gabon, Central African Republic, Chad, Congo, Democratic Republic of Congo (Zaire), Burundi, Rwanda, Equatorial Guinea, Sao Tome, and Principe
East Africa - Uganda, Kenya, Tanzania, Somalia, Ethiopia, Djibouti
Southern Africa - Zambia, Malawi, South Africa, Lesotho, Botswana, Madagascar/Malagasy Republic, Angola, Mozambique, Zimbabwe, Swaziland, Comoros, Mauritius, Seychelles, Reunion
Northern Africa - Morocco, Algeria, Tunisia, Libya, Sudan
Middle East - Iran, Turkey, Iraq, Egypt, Syria, Lebanon, Jordan, Israel, Saudi Arabia, Yemen Arab Republic, Yemen (PDR), Kuwait, Abu Dhabi, Bahrain, Qatar, United Arab Emirates, Oman
East Asia - China, Mongolia, Japan, Taiwan, North Korea, South Korea, Tibet
South Asia - Afghanistan, India, Junagadh, Kashmir, Hyderabad, Sikkim, Pakistan, Myanmar (Burma), Sri Lanka, Maldive Islands, Nepal, Bangladesh, Bhutan
Southeast Asia - Thailand, Cambodia, Laos, North Vietnam, South Vietnam, Malaysia,
Singapore, Philippines, Indonesia, Brunei
Oceania - Papua New Guinea, Vanuatu, Solomon Islands, Fiji, Marshall Islands, Micronesia, Western Samoa.  Note: this subsystem is not included in this article’s estimates because of its small intervention N.
Eastern Europe - Croatia, Bosnia-Herzegovina, Moldova, Ukraine, Belarus, Hungary, Poland, Czech Republic, Slovakia, Slovenia, Albania, Bulgaria, Romania, Estonia, Latvia, Lithuania, Russia, Georgia, Armenia, Azerbaijan , Turkmenistan, Tadzhikistan, Kirgizia, Uzbekistan, Kazakhstan



Adler, Emanuel and Michael N. Barnett. 1998. Security Communities. Cambridge: Cambridge University Press.
Bremer, Stuart. 1992. ADangerous Dyads: Conditions Affecting the Likelihood of Interstate War, 1816-1965.@ Journal of Conflict Resolution, Vol. 36, No. 2, pp. 309-341. 
Buzan, Barry. 1993. AFrom International System to International Society: Structural Realism and Regime Theory Meet the English School.@ International Organization, Vol.47, No. 3, pp. 327-352.
Buzan, Barry. 1991. People, States and Fear: An Agenda for International Security Studies in the Post-Cold War Era. 2nd edition. Boulder, Co: Lynne Rienner.
Bull, Hedley, ed. 1984. Intervention and World Politics. Oxford: Clarendon Press.
Cotton, Timothy Y.C. 1986. AWar and American Democracy: Voting Trends in the Last Five American Wars.@ Journal of Conflict Resolution, Vol. 30, No. 2, pp. 616-635.
Crescenzi, Mark J.C. and Andrew J. Enterline. 1999. ARipples from the Waves? A Systemic, Time-Series Analysis of Democracy, Democratization, and Interstate War.@ Journal of Peace Research. Vol. 36, No. 1, pp. 75-94.
Deutsch, Karl W., Sidney A. Burrell, Robert A. Kann, Maurice Lee, Jr., Martin Lichterman, Raymond E. Lindgren, Francis L. Loewenheim, and Richard W. Van Wagenen. 1957. Political Community and the North Atlantic Area. Princeton: Princeton University Press.
Diehl, Paul F. 1992. AWhat are They Fighting For? The Importance of Issues in International Conflict Research.@ Journal of Peace Research, Vol.29, No. 3, pp. 333-344.
Gelpi, Christopher and Joseph M. Grieco. 1998. ADemocracy, Crisis Escalation, and the Survival of Political Leaders, 1918-1992.@ Paper presented at the Annual Meeting of the American Political Science Association, Boston, MA, USA.
Gledistch, Kristian S. and Michael D. Ward. 2000. AWar and Peace in Time and Space: The Role of Democratization.@ International Studies Quarterly, Vol. 44, No. 1, pp. 1-29.
Hensel, Paul R. 1996. ACharting a Course to Conflict: Territorial Issues and Interstate Conflict, 1816-1992.@ Conflict Management and Peace Science, Vol. 15, No. 1, pp. 43-73.
Holsti, Kalevi J. 1991. Peace and War: Armed Conflicts and the International Order, 1648-1989. Cambridge: Cambridge University Press.
Huth, Paul K. 2000. AWhy are Territorial Disputes Between States a Central Cause of Conflict?@ In John A. Vasquez, ed., What do we Know About War? Lanham, MD: Roman Littlefield.
Huth, Paul K. 1996. Standing Your Ground: Territorial Disputes and International Conflict. Ann Arbor: University of Michigan Press.
Jackson, Robert H. and Carl G. Rosenberg. 1982. AWhy Africa=s Weak States Persist: The Empirical and Juridical in Statehood.@ World Politics, Vol. 35, No. 1, pp. 1-24.
Jervis, Robert. 1991/1992. AThe Future of World Politics: Will it Resemble the Past?@ International Security, Vol. 16,  No. 3, pp. 39-73.
Jones, Daniel M., Stuart A. Bremer, and J. David Singer. 1996. AMilitarized Interstate Disputes, 1816-1992: Rationale, Coding Rules, and Empirical Patterns.@ Conflict Management and Peace Science, Vol. 15, No. 2, pp. 163-213.
Kacowicz, Arie M. 1998. Zones of Peace in the Third World: South America and West Africa in Comparative Perspective. Albany: SUNY Press.
Kaiser, Karl. 1968. AThe Interaction of Regional Subsystems.@ World Politics, Vol.21, No. 1, pp. 84-107.
Kegley, Charles W., Jr. 1991. AExplaining the Great Power Peace: The Sources of Prolonged Postwar Stability.@  In Charles W. Kegley, Jr., ed., The Long Postwar Peace. New York: Harper Collins.
King, Gary. 1989. Unifying Political Methodology: The Likelihood Theory of Statistical Inference. Cambridge: Harvard University Press.
Kocs, Stephen. 1995. ATerritorial Disputes and Interstate War, 1945-1987.@ Journal of Politics, Vol.57, No. 1, pp. 159-175.
Kupchan, Charles A. 1998. AAfter Pax Americana: Benign Power, Regional Integration, and the Sources of Stable Multipolarity.@ International Security, Vol. 23, No. 2, pp. 40-79.
Lemke, Douglas. 1996. ASmall States and War.@ In Jacek Kugler and Douglas Lemke, eds., Parity and War: Evaluations and Extensions of the War Ledger.  Ann Arbor: University of Michigan Press.
Lake, David A. and Patrick M. Morgan. 1997. AThe New Regionalism in Security Affairs.@ In David A. Lake and Patrick M. Morgan, eds., Regional Orders: Building Security in a New World.  University Park: Pennsylvania State University Press.
Miller, Benjamin. 2001. AThe Global Sources of Regional Transitions from War to Peace.@  Journal of Peace Research, Vol. 38, No. 2, pp. 199-225.
Miller, Benjamin and Korina Kagan. 1997. AThe Great Powers and Regional Conflict: Eastern Europe and the Balkans from the Post-Napoleonic Era to the Post-Cold War Era.@ International Studies Quarterly, Vol. 41, No. 1, pp. 51-85.
Mitchell, Sarah M. and Brandon C. Prins. 1999. ABeyond Territorial Contiguity: Issues at Stake in Democratic Militarized Interstate Disputes.@ International Studies Quarterly, Vol. 43, No. 1, pp. 169-183.
Nagl, John A. 2001. AHitting Us Where We Don=t Expect It: Asymmetric Threats to U.S. National Security.@ National Security Studies Quarterly, Vol. 7, No. 4, pp. 113-121.
Pearson, Frederic S. and Robert A. Baumann. 1989. AInternational Military Intervention in Sub-Saharan African Subsystems.@ Journal of Political and Military Sociology, Vol. 17, No. 1, pp. 115-150.
Pearson, Frederic S. and Robert A. Baumann. 1993-1994. AInternational Military Intervention.@ Inter-University Consortium for Political and Social Research, Data Collection 6035, University of Michigan.
Pearson, Frederic S., Robert A. Baumann, and Jeffrey J. Pickering. 1994. AMilitary Intervention and Realpolitik.@ In Frank W. Wayman and Paul F. Diehl, eds., Reconstructing Realpolitik. Ann Arbor: University of Michigan Press.
Pickering, Jeffrey. 1999. AThe Structural Shape of Force: Interstate Intervention in the Zones of Peace and Turmoil, 1946-1996.@ International Interactions, Vol. 25, No. 4, pp. 363-391.
Pickering, Jeffrey and William R. Thompson. 1998. AStability in a Fragmenting World: Interstate Military Force, 1946-1988.@ Political Research Quarterly, Vol. 51, No. 1, pp. 241-263.
Ray, James L. 1995. Democracy and International Conflict. Columbia: University of South Carolina Press.
Russett, Bruce M. 1993. Grasping the Democratic Peace: Principles for a Post-Cold War World. Princeton: Princeton University Press.
Singer, Max and Aaron Wildavsky. 1996. The Real World Order: Zones of Peace/Zones of Turmoil. Revised edition. Chatham, NJ: Chatham House.
                        Siverson, Randolph and Harvey Starr. 1991. The Diffusion of War: A Study of Opportunity  and Willingness. Ann Arbor: University of Michigan.
Solingen, Etol. 1998. Regional Orders at Century=s Dawn. Princeton: Princeton University Press. 
Thompson, William R. 1996. ADemocracy and Peace: Putting the Cart before the Horse?@             International Organization, Vol. 50, No. 1, pp. 141-174.
Thompson, William R. 1973. AThe Regional Subsystem: A Conceptual Explication and Propositional Inventory.@ International Studies Quarterly, Vol. 17, No. 2, pp. 89-117.
Vasquez, John A. 1995. AWhy Do Neighbors Fight? Proximity, Interaction, and Territoriality.@ Journal of Peace Research, Vol.32, No. 3, pp. 277-293.
Vasquez, John A. and Marie T. Henehan. 2001. ATerritorial Disputes and the Probability of War, 1816-1992.@ Journal of Peace Research, Vol. 38, No. 2, pp. 123-138.


Model 1 was also estimated with a more restrictive dependent variable.  Interventions launched over territory were coded A0@ along with non-belligerent interventions. Those launched over other belligerent issues were coded A1.@  Coefficient signs remain the same, but some variables slip from statistical significance at the .05 level.  East Asia remains positive and statistically significant, while the Center and Central Africa are negative and statistically significant.  Also, each of the logistic models presented was re-estimated using two different measures of subsystemic interventions.  Subsystemic interventions were considered to be either 1) any interventions undertaken by regional actors or 2) interventions regional actors initiated against other regional states.  Estimating these alternative measures did not change the findings reported in Table 1 significantly.   


1 Such an iterative process of establishing and deepening peace in regional settings has been studied from varying theoretical approaches, including constructivist analyses of security communities (Adler and Barnett, 1998), English school interpretations of Amature anarchy@ (Buzan, 1993) and zones of peace (Kacowicz, 1998), quasi-realist approaches to differently defined zones of peace (Singer and Wildavsky, 1996), and neo-liberal and realist concern with the democratic peace (Russett, 1993; Ray, 1995).

It must be noted that although t4 is half the length of the other time periods, it contains roughly the same number of uses of force as t1, when there were fewer sovereign actors.  The distribution of military interventions is roughly bell-shaped across the four time periods.  The totals are: t1 = 154, t2 = 261, t3 = 272, t4 = 140.  Any significant differences measured between time periods do not result from unusually small or large Ns.