The Geography of Civil War
Summary and Keywords
Attention to geography in the study of civil war has risen dramatically in recent years. Beginning with county-level data in the fields of classical political geography, international relations, and comparative politics, a vast body of conflict research is now dedicated to sub-national analysis. This later turn is itself geographical. Innovations in the geographical study of civil war have dramatically improved our collective understanding of violence and continue to call for modifications of conflict theory.
While a turn toward geography has therefore proved valuable for academic research that is most often dominated by political science, there remain fundamental differences within the research community about what constitutes geographical inquiry. An example of such a difference is the attention that physical geography (such as forest cover, mountainous terrain) has received in civil war research over investigations of the nuanced social composition of regions and localities, which tends to dominate for conflict research within the discipline of human geography.
The spatial dependencies among conflict locations and events need to be highlighted for their importance. These patterns can reveal important underlying social forces that are interesting to scholars in various disciplines, as well as to the study of key geographical processes and the shift toward spatial disaggregation. This localization of violence studies is necessary and concerns the notion of hierarchical scale, which is a conceptual foundation of human geography. In studying the geography of civil war, there are methodological tools that can outline some risks associated with geospatial analysis of violence.
The study of intrastate violent conflict (especially civil war) has changed dramatically in recent years. In large part this change has taken place because researchers increasingly adopt a geographical lens for understanding the onset, dynamics, and trends of conflict events within civil wars. The advances made in the study of domestic political violence through this geographic focus will be discussed. The field of conflict research has significant breadth and depth—to cast a wide net for pertinent research, we avoid restricting the scope of political violence under consideration to include only cases where the government is an engaged actor (this is an important element of some civil war definitions). After all, while the majority of civil war research is oriented toward rebel-government interactions, some of the most severe contemporary violence takes place among non-state actors. This dynamic actually makes the studying of many ongoing civil conflicts quite complicated because distinctions among actors are fuzzy.
Understanding the “geography of civil war” entails much more than visual, static interpretations of conflict incidents displayed on a map.1 While observing a geospatial pattern is part of any researcher’s task, we believe explaining such distributions should be the ideal achievement in any geographical analysis of civil war. A geographical analysis of political violence concentrates on the variation in causes, correlates, and consequences of conflict across space created through institutional exclusion, vacillating political economies, and multifaceted identity relationships; it seriously considers how scales within states (e.g., local, regional, and national) produce distinctly different contexts and motivations for the emergence of animosities; it focuses on relationships within and between actors, groups, and localities. Geography assumes a position within the study of political violence where complex relationships can certainly include technical and theoretical elements that are proprietary to the discipline; however, the potential explanatory mechanisms for observed spatial patterns of violence are necessarily interdisciplinary, as is the broad field of human geography itself.
The most fundamental concept in the geographical study of civil war is the heterogeneity of spaces of conflict and peace within states. This principle means that at various levels, locations or regions are inter-dependent as units of observation. This is meaningfully observed in most conflict research as spatial dependencies or a process of spatial diffusion. The spread of international and civil war violence has been well documented country-by-country in the international system (among the first examples, see Most & Starr, 1980; Houweling & Siccama, 1985; Hill & Rothchild, 1986; O’Loughlin, 1986; Kirby & Ward, 1987; O’Loughlin & Anselin, 1991) but also within countries at sub-national levels (some recent examples include Toft & Zhukov, 2012; Schutte & Weidmann, 2011; Linke, Witmer, & O’Loughlin, 2012; Schutte & Donnay, 2014). However, identifying positive spatial correlations among conflict incidents (a spatial trend) is only a step toward understanding that pattern. Hence, a positive and statistically significant correlation coefficient is insufficient: to extend the analysis further, geographical and temporal dependencies must be viewed as observable signals emanating from a social process that it may be impossible to directly observe empirically. Using some of the methodologies reviewed in “Data and Methods for Geographical Analysis of Civil Wars” is a critically important component of the effort to understand civil war, but the mission of conflict researchers should be to uncover why conflicts cluster in certain places, not only that they do cluster in space. Recent research has advanced the nuanced understanding of how geographical clustering emerges, for example by identifying the differences between receiver, sender, dyad, and network forces (Metternich, Minhas, & Ward, 2015). Such efforts to isolate a mechanism driving observed relationships is a gold standard and represents a substantial improvement upon classical descriptive research in this field, some of which started in the 1980s. For sub-national analysis of civil war violence in Bosnia, Kosovo, Burundi, and Rwanda, Schutte and Weidmann (2011) argue that the most plausible reason for finding “relocation” diffusion in certain cases and “escalation” diffusion in others is the type of conflict for a given case. The relocation diffusion signal is more likely in settings of regular war whereas irregular war is characterized by “escalation” diffusion. Similarly, although for less intense civil conflict, Mesev, Shirlow, and Downs (2009) identify segregated communities of Belfast (over 90% either Catholic or Protestant) as especially likely to witness fatalities during Northern Ireland’s conflict, indicating a key inter-group behavioral politics explanation for observed clustering of violence.
Each section here reviews numerous examples of geographical civil war research in detail. Spatial dependencies, diffusion processes, and other human geography phenomena related to civil war and political violence are discussed. Inherently geographical concepts and principles that explain patterns of civil war violence but that are fundamentally different from the majority of conflict studies research are considered. For instance, network analysis is used by few civil conflict researchers but can be extremely informative. Another set of research is more qualitatively oriented and identifies the construction of social space in ways that can contribute to violent cycles of conflict.
Advancements of localization and spatial disaggregation—now preferred by most conflict researchers—in terms of hierarchical scale, which is an inherently geographical lens for studying violence, are examined. Our position is that many conflict scholars are implicitly and necessarily engaging with geographical methodological tools tacitly as means to other ends. Traditionally in the cross-national comparative and political science traditions, investigations of correlates of civil war used nation-states as units of analysis (the most frequently cited examples include Collier & Hoeffler, 2004; Fearon & Laitin, 2003). It is arguably quite difficult, however, to make a conceptual leap from the structural country-level conditions (e.g., Gross Domestic Product per capita, GDP) under which civil war breaks out to the individual-level motivations that influence combatants and rebel group leadership (e.g., as explained in Buhaug & Rød, 2006). Poorer countries may be prone to civil wars, but this does not necessarily mean that poorer people within a country are those who decide to organize against a government and use deadly violence to achieve their goals. Nevertheless, for years this is the empirical data that has been used to test the “greed vs. grievance” motivations for civil war onset. Conceivably, in contrast to the assumptions researchers must make using these coarse data, influential political leaders in the wealthiest region within a country may actually be those who start a secessionist war. Even within a poor region, the most savvy entrepreneurs of violence may be rich and well-connected politicians.
To learn more about the dynamics of civil war violence, researchers have increasingly concentrated on the compositional incidents of conflict at sub-national levels within countries. Tracking the individual violent events that take place within a civil war reveals a wealth of information about actor behavior and the sub-national structural conditions associated with conflict. Violence within a war may take place in a limited extent of a given country’s territory; understanding why it happens in those key areas is impossible if we assume that war takes place homogenously across large countries. While studying the “microfoundations” of conflict has some precedent in political science (e.g., Gates, 2001 names “geography” as primary determinant of rebel group success), Kalyvas (2006) applied a localized lens to understand civil war. He concludes that levels of violence observed across towns during the Greek civil war are a function of degrees of governmental or rebel territorial control and the varying motivations that individuals have to inform on militants. Other scholars preceding him who probed variation in civil war violence location-by-location have studied Spain (Tone, 1994), Mozambique (Finnegan, 1992), Indonesia (Cribb, 1991), and India (Varshney, 2002), among other countries. In the anthropological and ethnographic tradition, students of civil war violence find many in-depth accounts of individual villages. Nevertheless, the adoption of Kalyvas’s localized conflict research within comparative politics and international relations, fields where large N statistical analysis is common, was unprecedented.
The dramatically improving access to geographical data and spatial analysis tools for studying the geography of civil wars indicates great potential for future research; some classical methods for point-pattern and geostatistical descriptive analyses in nontechnical terms are outlined along with spatial regression analysis. Some of the potential issues that geographical data present for statistical analysis and testing theoretical propositions are briefly outlined.
Interdependence of Observations and Spatial Diffusion
Across a diverse set of disciplinary approaches, it is increasingly accepted that Tobler’s “first law of geography”—where “close” things are more directly related than those that are distant—applies to the study of civil wars. Spatial dependencies among incidents of conflict exist in nearly every observational events–based study of civil war. With great regularity, some form of geostatistical and point-pattern detection methods reveals positive statistical correlations between countries experiencing civil war in the international system (Black, 2013; Braithwaite, 2010; Buhaug & Gleditsch, 2008; Enterline, 1998; Flint, Diehl, Scheffran, Vasquez, & Chi, 2009; Gates, 2001; Gleditsch, 2007; Gleditsch & Ward, 2000; Hegre & Sambanis, 2006; Houweling & Siccama, 1985; Joyce & Braithwaite, 2013; Maves & Braithwaite, 2013; Most & Starr, 1980; Murdoch & Sandler, 2002; O’Loughlin & Anselin, 1991; Radil, Flint, & Chi 2013; Ward & Gleditsch, 2002; Weidmann, 2015; Zhukov & Stewart, 2013) and also for spatially disaggregated violent events within civil conflicts (Braithwaite & Johnson, 2012; Do & Iyer, 2010; Linke, Witmer, & O’Loughlin, 2012; Linke, Schutte, & Buhaug, 2015; Mesev, Shirlow, & Downs, 2009; O’Loughlin & Witmer, 2012; Schutte & Weidmann, 2011; Toft & Zhukov, 2012; Weidmann & Ward, 2010; Zhukov, 2012, 2013). The reasons that violence leads to subsequent and nearby violence are many and the explanation depends on the type of conflict, as well as the political and economic contexts of and within the war.
But spatial clustering at a cross-section in time (t) reveals relatively limited information about how the cluster of conflict came to exist. This is true for both cross-national country-level studies and sub-national research. As has been articulated, our purpose should be identifying the reasons that any pattern emerged, which requires introducing a temporal dimension (t-1) to any study. To use the epidemiological language of “contagion” and “infection” (e.g., Braithwaite, 2010; Toft & Zhukov, 2012), the ideal goal can be described as understanding how conflict spread into a given location. For example, anecdotal accounts and academic research have shown that the Afghanistan-Pakistan border is particularly volatile and deadly for residents. But why? As a line on a paper map, a border means little. Even a casual observer might at first point out that violence rarely spilled over into neighboring Tajikistan. More nuanced considerations of the political contexts surrounding violence in the region reveals how a transmission of instability takes place from Afghanistan’s eastern districts into the Federally Administered Tribal Areas (FATA) in Pakistan. Research suggests that the qualities of political regimes contribute to the spread of violence (Braithwaite, 2010): the inability of authoritarian regimes (or their lack of will) to manage the externalities of conflict has been shown to facilitate the diffusion of conflict in cross-national research (Maves & Braithwaite, 2013). Sambanis (2001) and Gleditsch (2002) earlier reached similar conclusions about how regime characteristics influence civil war. Gleditsch (2002, p. 109) notes that countries that are “located among relatively democratic neighbors have significantly lower risks of experiencing civil war than do countries located in a zone of more autocratic and less constrained politics.”
There are much more direct propositions as to what causes conflict to relocate across international borders and that consider the particular contexts of that environment: militant migration or shared Pashtun identity across the border may explain part of the spread of Afghanistan’s violence into FATA as militants can easily blend into the population. This and similar explanations are based in the rejection of what Gleditsch (2007) called the “closed polity” approach to conflict research, where countries are viewed as physically sealed from outside influences instead of integrated economically (de Soysa, 2002), politically (Regan, 2000), and socially (Davis & Moore, 1997) with nearby regions in ways that contribute to conflict risk.
Despite the recent appreciation for diffusion processes occurring at sub-national scales (zooming in), much research in the militant-mobility tradition actually takes place with supra-national-level considerations (zooming out). For example, Saleyhan and Gleditsch (2006) attribute the spread of political instability to refugee overflows from country to country. Some suggested but untested explanations for the pattern include: “the transnational spread of arms, combatants, and ideologies conductive to conflict” in addition to the fact that immigration changes ethnic demographic compositions of arrival areas and distorts labor markets in the host area. Others suggest that ethnic ties to nearby or neighboring countries help spread civil conflict through logistical and other material support (Forsberg, 2014). For example, violence in Democratic Republic of Congo during the mid-late 1990s is strongly tied to the aftermath of the Rwandan genocide; it is often presented as a perfect typology of conflict diffusion and escalation in severity and scope (Prunier, 2008; Autesserre, 2011). But the DR-Congo war case, being a long and complex conflict, is also an example of how diffusion is a function of political competition at the international level; how rebellions are funded and financially beneficial to multiple actors; and the fact that refugee-related diffusion mechanisms are not universal (e.g., refugees from DR-Congo and Rwanda did not have the same effect in either neighboring Uganda or Tanzania—see Whitaker, 2003).
Within countries experiencing civil conflict, the domestic diffusion patterns of conflict can reveal theoretically informative trends about political processes. Toft and Zhukov’s (2012) “epidemic” model of conflict dynamics tests whether “punishment” strategies of the Russian government inhibit site-recovery of conflict, or “denial” options, which are designed to halt the spatial spread of conflict to new sites, are more effective in fighting insurgents in the North Caucasus from 2000 to 2008. Their sub-national geographic research design, which is similar in character to that of Schutte and Weidmann (2011), suggests that punishment may have unintended negative consequences, but that efforts to deny insurgent mobility can be effective. In an expected relationship, Zhukov (2013) uses explicitly spatial methods and finds that insurgents’ use of road networks also conditions violence patterns in the North Caucasus region (a factor also noted by Straus, 2006, in relation to war in Rwanda). The precise distances at which Iraqi “tit-for-tat” reprisal acts of violence between insurgents and counterinsurgents fade offer Linke, Witmer, and O’Loughlin (2012) valuable insights into the way that particular social and economic contexts shape trends in violence. Crost and Felter (2015) similarly test proposed explanations for contagion of conflict within the Philippines and identify the opposition New People’s Army’s efforts to exploit gaps in army force strength as the most likely cause. These latter works are innovative in using the spatial contexts in which conflict occurs to situate diffusion processes vis-à-vis social conditions that allow them to emerge.
Geographical Concepts and Analysis
While many geographical influences on civil war violence (border politics, infrastructure, population distributions, etc.) are tested in spatial diffusion or dependency or spatially disaggregated analyses, other studies use unique definitions of analysis units or investigate research questions that are necessarily geographical in character (e.g., territorial disputes). As zones of international interaction and policing, borders are especially conflict prone for several reasons. Fighting near borders sometimes takes place over the location of the border itself, which has occurred in and near Ethiopia on both the Eritrean and Somali frontiers. According to long traditions in political geography (e.g., Newman, 2006) and international relations (e.g., Buhaug & Gleditsch, 2008), borders have been and will continue to be sites of contention due to their cultural importance but also because of their strategic political value and regulatory (e.g., trade) role. Who lives in border regions, and their relationship to the state, is also critical. Buhaug and Gates (2002) and Buhaug (2006) show that contests over establishing a new territory (i.e., secession) differ in character and actions from those intended to displace a sitting government. Yet, groups situated in border regions, or the fringes of state control, sometimes engage in conflict against distant governments over not only secession, but also representation. The border and distance findings are unique to civil wars and do not always hold for more recent conflicts, urban contests, and violence perpetrated by militias and community agents (see Raleigh, 2016).
In cases where physical distance is associated with higher risks of civil wars, it is often so because of how social and physical distance are correlated. “Social” distance refers to the degree of similarity in culture, identity, and representation, and groups who base their political standing on ethnic, regional, class, or religious identity (over programmatic political stances) often have large social distances from other groups. Often, the difference in identity becomes the basis upon which people assume political inclusion and exclusion occurs (see Cederman, Weidmann, & Gleditsch, 2011 for an example of this framework).
In a political ecology tradition, Le Billon (2004) in Political Geography argued for using a territorial framework for understanding the role of natural resources in civil war violence. Le Billon identifies four typologies of physical geography or natural resource endowment that lead to distinct forms of violence: coups d’etat, rebellion/rioting, secession, and “warlordism” all emerge from combinations of point-proximate (to the capital), diffuse-proximate, point-distant, and diffuse-distant geographies of resource distributions, respectively.
Population distributions within countries can also have important implications for civil war violence. Raleigh and Hegre (2009), for example, find that population centers are prone to conflict but specifically in remote regions; they link this likelihood to the demographic opportunity that coincides with identity politics previously discussed. Schutte (2014) explains the severity of casualties during civil war using relative geographical proximities of population centers to the capital city. Capital cities are generally expected to see elevated risks of violence compared with other areas because they are sites of national governance, and are therefore prone to attempted capture by groups aspiring to take control of a country during a civil war (Buhaug, 2006; Buhaug & Gates, 2002). Large population sizes are related to conflict risk according to some, but several qualities of rural areas makes certain mechanisms for elevated conflict risk more robust (Weidmann, 2015). Raleigh (2016) considers why a wider variety of political violence occurs in urban areas, while civil war studies typically focus on rural cases: she finds that the multiple, low-intensity forms of urban violence—including militia attacks, communal contests, riots, and protests—indicate a change in the collective-action capabilities and goals of modern conflict agents. These goals are themselves shaped by the experience of mounting urban grievances, but the ethnic and regional heterogeneity in urban areas prevents substantial collective action to counter the “rural bias” practiced by many developing governments, particularly in Africa.
Above we made brief note of territorial politics as a potential cause of conflict (Huth, 1996; Vasquez, 1993) with the caveat that many of the scholarly investigations of this topic tend to focus on inter-state disputes instead of civil war. Even within the study of international conflict, geographers have identified regional clustering effects and heterogeneous estimates of how territorial disputes explain war on the global stage; there are important exceptions to the universal rule (Chi & Flint, 2013). Walter (2006) has suggested that territorial concessions to a rebel group engaged in civil war (e.g., in a peace agreement) may lead to a knock-on effect based on the expectations of other opposition forces that they will experience the same. In a global analysis, Forsberg (2013) questions whether such a diffusion takes place.
An important body of research has emerged from disciplinary human geography that is not tied to locational geography (e.g., proximity to a border or city). The central concern in these studies is the political power to claim and control space with the goal of population expulsion—the goal being to “remake” space (Dahlman & O’Tuathail, 2005; Lunstrum, 2009; Oslender, 2007; Porteous & Smith, 2001; Tyner, 2008; Wood, 2001). In describing the “geographic aspects of genocide” in Bosnia and Rwanda, Wood (2001) focuses on ethnic cleansing campaigns where conflict actors sometimes rely on a strategy of “domicide” (a term of Porteous & Smith, 2001) to evict populations by destroying homes. Death is not always the primary goal, and domicide can entail the destruction of houses and all cultural and historical symbols of a population such as churches or religious sites and graveyards or other monuments. Similar lenses for interpreting civil war violence have been applied to Bosnia’s post–Dayton Accords political geography (Dahlman & O’Tuathail, 2005), to Mozambique in identifying civil war legacies of Massigner district (Lunstrum, 2009), and to politically motivated violence in Colombia (Oslender, 2007). A substantial theme of this body of research is the identification of how influential leaders produce geopolitical knowledge. Tyner (2011), for example, provides a vivid example of such practices in showing how the Khmer Rouge redrew political maps in school textbooks as part of the creation of Democratic Kampuchea. In the eyes of many at the time of Cambodia’s civil war, these changes provided partial justification for supporting or participating in violence.
Much of this research emanating from human geography is firmly dedicated to acknowledging and appreciating the nuances of locality and the particularities of individual places (e.g., Agnew, 1987) where conflict has taken place. Because of these firm convictions—and despite other epistemological and methodological differences this community has with the majority of research avenues in political science—these studies are sometimes designed for a thematic or theoretical generalization, but not always. Unfortunately the divisions between these communities of researchers interested in conflict geography (on one hand striving for deep knowledge and on the other international generalization) often seem too vast to bridge.
Within the study the civil conflict, network research methods are gaining popularity of late. Networks are inherently geographical because they define the connections between actors (countries, companies, armed groups, individuals, or any others) in relative space instead of defining their location by locational coordinates on the globe. The principle of this work is that a position in a network can be as influential as position in territorial place. “ConflictSpace,” for example, is one lens for viewing the interactions among actors and the degree to which each is embedded in international relationships (Flint, Diehl, Scheffran, Vasquez, & Chi, 2009). Innovative actor-centric network research on interstate war and alliances is becoming widely available (e.g., Dorff & Ward, 2013; Radil, Flint, & Chi, 2013) but sub-national analysis and the application of network analysis to the geography of civil wars is less common. As an exception to the rule, König, Rohner, Thoenig, and Zilibotti (2015) illustrate how strong networks among armed groups in the Second Congolese war intensified fighting dramatically. Metternich, Dorff, Gallop, Weschle, and Ward (2013) find that connections between anti-government opposition networks in Thailand translate into elevated levels of conflict during the country’s tumultuous recent years. One of the benefits of applying network analysis methods to the study of conflict is that an actor-centric focus dramatically expands the possibilities for incorporating ephemeral and complex sets of groups that fight in contemporary violent conflicts and “negotiate statehood” within civil wars, to use the term of Hagmann and Péclard (2010). Radil and Flint (2013) blend actor-centric networks, and both civil war and interstate wars in Africa to illuminate shifting characteristics of sovereignty regimes. Network analysis is also the basis for studies of terrorist group activity within the Sahel and Sahara regions (Walther & Christopoulos, 2012) and across the globe (Medina & Hepner, 2011).
The physical environment is suggested by some as critical in creating a fertile environment for insurgency (Do & Iyer, 2010; Fearon & Laitin, 2003). In particular, mountainous terrain—measured by many as the percentage of a state at high elevation—is perceived as an important causal force in driving civil war onset. More recent research has called the finding into question empirically and also conceptually (Buhaug & Gates, 2002; Rustad, Rod, Larsen, & Gleditsch, 2008). At a country level and also at sub-national scales, several anecdotal case studies (e.g., Burma or Cambodia) have driven the narrative on this topic rather than the nuanced analysis of the locations of fighting and movement of actors. This variable remains a common addition to statistical models, despite some obvious questions regarding whether the conflicts actually occurred in mountainous terrain (or whether it is possible to measure and observe actor mobility effectively within the conflict); what role terrain plays in conflicts that are not rural or peripheral; and what explains insurgency in states without significant mountainous terrain. Daly (2012, p. 473) has directly called into question the prominent role of mountainous terrain in Colombia’s civil conflict, instead linking violence to the “human and social geography that determines if rebellion is organizationally feasible.”
Finally, the possibility that climate change will lead to conflict has gained recent scholarly and public-policy interest. Any effort to understand the impact of changing environmental conditions for violent conflict involves an engagement with physical geography data, which may include rainfall estimates, freshwater availability (e.g., well or watering hole locations), or land use change over time. The vast majority of this research lacks an effort to understand the mechanisms operating upon violent conflict, including some articles in high-profile general science journals (Hsiang, Burke, & Miguel, 2013; for concerns with this research agenda see Raleigh, Linke, & O’Loughlin, 2014). Recent exceptions to the poor quality of most climate change and conflict research include articles that isolate explanatory conditions tying shortages of rainfall or pasture directly to conflict via harvest loss in India (Wischnath & Buhaug, 2014), livestock market prices in Somalia (Maystadt & Ecker, 2014), and migration in Sudan (de Juan, 2015). Some research also specifically tests the potentially moderating role of local-level social variables such as the presence of institutional rules for natural resource management (Linke, Schutte, & Buhaug, 2015).
Geographical Scale and Sub-National Analysis
Hierarchical scale is a critically important conceptual element of human and political geography. Recent major trends in conflict analysis—resulting in substantial advancements in the study of civil war violence—have been implicitly or tacitly geographical in character. The localization of data below the boundaries of the nation-state is necessarily a methodological engagement with hierarchical scale even if for a given researcher this is a geographical means to another end. Consider that a binary measure of civil war occurring (=1) or not (= 0) tells us very little about the conditions under which conflict develops, escalates, and spreads before becoming a country-wide catastrophe. A firm critique of (nation-)state centric conceptions of power in the international system was offered in disciplinary political geography by Agnew (1994) and Agnew and Corbridge (1995), who warned researchers of succumbing to the “territorial trap” that privileged a single level of analysis (and thereby a restricted set of actors) in the study of international affairs.
The many recent disaggregated civil war studies have different explanatory emphases, naturally, but they all have one quality in common: the conclusions could not be found using crude data that eclipse local social forces. In Bosnia, Weidmann (2011) isolates forms of ethnic conflict emanating “from above” and “from below” with regard to narratives of resentment and motivating influences for mobilization, showing that the each occurs under distinct conditions across municipalities. For counterinsurgency campaigns in the North Caucasus Lyall (2009) uses data for Russian shelling of Chechen villages to test whether indiscriminate violence deters militant activity and reaches the counterintuitive conclusion that it does: subsequent violence by insurgents drops. Electoral motivations for political violence have long been a focus of conflict research but in Afghanistan the recent election took place within the setting of active insurgency. Weidmann and Callen (2013) show that patterns of violent conflict provide opportunities for fraud, which is a critical obstacle to institutional development. Do and Iyer (2010) find that poorer districts in Nepal are “drawn into the insurgency earlier” and that there are variable effects of structural conditions and social contexts (affecting the onset vs. intensity of conflict, respectively). Poverty and inequality is a crucially important explanation for certain patterns of civil war violence, which is shown by Buhaug, Gleditsch, Holterman, Østby, and Tollefsen (2011).
Fine spatial resolution civil conflict research is also of course easily paired with computational (agent-based) modeling methods, allowing researchers to use simulations in experimental analyses of scenarios for the contexts that may lead to violence. Using an agent-based model Weidmann and Saleyhan (2012) show that sectarian polarization and levels of policing have shaped the ebb-and-flow of conflict within Iraqi districts. Urban-level ethnic segregation or separation of populations in Jerusalem has reduced violence according to Bhavnani, Donnay, Miodownik, Mor, and Helbing (2014), but in their spatial analysis of potential futures the authors argue that their observed effect hinges on levels of inter-community social distance.
These disaggregated studies have been single-country case studies, but increasingly cross-national studies are conducted sub-national spatial resolutions. Pierskalla and Hollenbach (2013), for example, use a half-degree GIS grid structure to investigate the links between cell phone coverage and violence across African states. Dowd and Raleigh (2013) consider the territories that Islamist actors control in northwestern Africa and take issue with overly simplistic narratives of Muslim militants sweeping the entire region. Linke, Schutte, and Buhaug (2015) base their investigation of how population attitudes may lead to the spread of conflict using survey data within level one administrative units across 16 countries of sub-Saharan Africa. Raleigh (2014) uses a local grid of 10kmsq to consider how multiple forms of conflict co-occur within states but are spatially distinct; each form is motivated by the type of local relationship with the central government. Therefore, a topography of conflict risk can be found within states, subject to the degree and type of governance practiced by regimes. Raleigh, Choi, and Kniveton (2015) use subnational units, internal relationships, and local feedback to observe the inter-effects of climate, food price, and conflict across 25 African states. For an analysis of temperature and precipitation effects across sub-Saharan Africa O’Loughlin, Linke, and Witmer (2014) use sub-national 50km grid cells.
In conclusion, the use of geographical techniques in conflict studies has advanced researchers’ understandings of how various social factors operate. However, there are vast unchartered areas of study in the disaggregated framework for understanding political violence and conflict within civil wars. Across the board, these require rigorous, robust, generalizable frameworks informed by geographic theory and methods.
Data and Methods for Geographical Analysis of Civil Wars
In recent years, the number of tools available for researchers to conduct spatial analysis of violence has grown dramatically. Not every option available for each conceivable variable researchers are interested in will be listed. However, important changes in conflict research have come about as a result of work using some of the following data resources. Many of the methods researchers use for analysis of civil war violence and its explanatory forces have not changed fundamentally from examples of traditional work in spatial sciences and geographical analysis decades ago. Nevertheless, there has been substantial improvement made in the flexibility and accessibility of the methods, and several examples are provided. With the great potential of data and methods innovations comes the risk of hastily conducting analysis that leads to false conclusions. Fortunately, many possible issues can be avoided and at the very least acknowledged by researchers.
At fine spatial resolutions, civil war and conflict researchers now have access to conflict data sets that are in Geographic Information Systems (GIS)–ready platforms or formats. In addition to the standard Armed Conflict Dataset (ACD) (Gleditsch et al., 2002), the Upsalla Conflict Data Program/Peace Research Institue Oslo civil war data set, PRIO has the Conflict Site dataset (Hallberg, 2012), which records the locations of violent confrontations between rebels and government forces within civil wars. Buhaug and Lujala (2005) use a zone defined around centerpoints of ACD in their research. Alongside the shift toward sub-national analyses in civil war research, the number of spatially disaggregated and publically available conflict data sets has risen. The Armed Conflict Location and Event Data Project (ACLED) was the first example of conflict data identifying the precise location, timing, type, and involved actors (among other information) for violent events across all African states and South and Southeast Asia. These data are available from 1997 and are being updated in real time. This data and analysis project produces information on the specific dates and locations of political violence, the types of event, the groups involved, fatalities, and changes in territorial control. Information is recorded on battles, the killing of civilians, riots, and recruitment activities of rebels, governments, militias, armed groups, and protesters. UCDP–Georeferenced Event Data collects conflict data, covering 1989 through 2010 (Sundberg & Melander, 2013). These data are more restrictive than ACLED and other databases by definition, including only events where an “incompatibility” between identifiable actors led to at least one death during a calendar year. The UCDP-GED polygons data set outlines the spatial extent of interactions between conflict actors as it is defined by the location of these events.
Other political, economic, and social data are also increasingly available. For political boundaries the C-Shapes data set (Weidmann, Kuse, & Gleditsch, 2010) standardizes the country borders that researchers use in national-level studies. The ethnic composition of regions and their exclusion from state governance can also be captured in Geo-referencing Ethnic Power Relations (GeoEPR; Wucherpfennig, Weidmann, Girardin, Cederman, & Wimmer, 2011) and Geo-Referencing of Ethnic Groups (GREG; Weidmann, Rød, & Cederman, 2010). Sub-national income surrogates are provided by G-ECON data (Nordhaus, 2006), which can be an important component of civil war research either as a variable of substantial interest or as a control for the influences of poverty. New sub-national poverty data are also under development by geographers (Tatem, Gething, Bhatt, Weiss, & Pezzulo, 2013). Population patterns are an important influence on conflict dynamics and data for local-level variation in demographics can be found in the Gridded Population of the World (GPW) data set (Balk & Yetman, 2013). As an alternative to GPW, Afripop raster images are available at very fine (1km) spatial resolutions but with a static temporal dimension (Afripop, 2015).
There have been valuable recent efforts to offer tools for aggregating socioeconomic and conflict event data into a common platform for researchers to use “off the shelf.” The most recent and flexible example is SpatialGridBuilder (Pickering, 2015). SpatialGridBuilder is a standalone program that incorporates bounding coordinates for a study area (from a country, to a continent, to the globe) and defines a gridded cell structure for joining layers of data into a common panel time-series data set that can be used for statistical analysis. These layers can be existing political data sets at a country level or sub-national resolution information in raster or vector formats. The most unusual quality of SpatialGridBuilder is that the user may define any spatial resolution for the grid cell dimensions; whether a statistical association that is observed at 50-by-50km resolution exists at 10-by-10km (or 100-by-100km) can be verified by the researcher. The work of Pickering (2015) is similar to the important work of Tollefsen et al. (2012), who developed PRIO-GRID, but has additional options for spatial variation.
As has been discussed, the spatial interdependence of cities or towns, regions, and countries is the foundational element of a geographical analysis of civil war violence. One can identify the space-time dependencies of conflict within a data set using several spatial analysis tools that are specifically designed for either locations (e.g., town or city) and, separately, territorial (e.g., administrative or grid cell) units of analysis. All of these can loosely be described as geostatistical methods, but the two are discussed separately (some applications can be used for both areal units and locations, such as SatScan, explained below).
For identifying distance-decay effects among conflict observations the key is finding a range where relationships among places cease to be meaningful. There are a number of specific variance measurements that can achieve this (or more technical variations of the basic method): nearest neighbor analysis, Ripley’s K statistics, and semivariograms are all options that reveal patterns across space (e.g., Bivand, Pebesma, & Gómez-Rubio, 2013; Brunsdon & Comber, 2015; Brundson, Fotheringham, & Charlton, 1998). Depending on the format of the data at hand (locations versus areal units), Getis-Ord Gi*, Moran’s I, and their related Local Indicators of Spatial Association (LISA) can be valuable tools (Cliff & Ord, 1981; Anselin, 1995).
One of the best examples of an advanced and informative cluster identification method is the space-time clustering SatScan statistic (Kulldorff, Heffernan, Hartman, Assuncao, & Mostashari, 2005), which was designed for epidemiological research and disease mapping. Martin Kulldorff’s method can be applied to areal unit data or locations, making it especially flexible. SatScan has been applied to Afghanistan-Pakistan conflict research in O’Loughlin, Witmer, Linke, and Thorwardson (2012) to isolate trends along the border frontier. A similar statistical approach using independent methods was applied in Iraq (Linke, Witmer, & O’Loughlin, 2012; Braithwaite & Johnson, 2012) and the North Caucasus (O’Loughlin & Witmer, 2012).
The point-pattern analysis methods and geostatistical tests just identified are necessarily descriptive. While the results may indicate spatial clustering within civil wars (or even the location of clustering within civil wars), we cannot know from these methods the cause of that hot spot of violence. Scholars dedicate entire textbooks to the topic of spatial regression analysis and spatial econometrics, tools that can isolate the effects of various independent variables upon violence in the presence of Independent and Identically Distributed (i.i.d) assumption violations (Anselin, 1988; Brundson, Fotheringham, & Charlton, 1998; Ward & Gleditsch, 2007). In nontechnical terms and in cursory form, we will note two general families of models that can be used to (a) substantively analyze and (b) control away the effects of i.i.d violations in civil war research that applies large-N analysis. First, in any estimation of conflict as dependent variable (DV)—whether onset, duration, or intensity, depending on the research question at hand and the specifics of the regression model—a “spatial lag” term for the values of the DV in neighboring observations (an autoregressive term) can be added into the estimation alongside any other independent variables (IV) of interest, such as population size, ethnic polarization, or income. This analysis approach will give coefficients of the impact of income (or any other variable) upon conflict and the associated standard error. It will also return the effect of the spatial lag term, which can be understood as the strength of a spillover effect between units of observation. In this way researchers can isolate the explanatory influence of geographical influences of conflict across space. The second alternative is most appropriate for ensuring that significance levels of IVs of interest are not affected by i.i.d violations, in contrast to an estimation of how spatial relationships shape conflict as an outcome. A “spatial error” model will adjust the standard errors of all coefficients to take into account the spatial clustering among observations. This reduces the chances of finding statistical significance for an IV like income level when in fact it has no relationship to conflict. An example of this standard adjustment error strategy can be found in Conley (1999).
The optimal solution for modeling conflict data would be a combined approach that identifies the direct spillover effects of conflict in nearby administrative units and also adjusts standard errors according to a neighboring weights matrix. These two general approaches describe generalized differences in modeling strategies, but advanced statistical estimators exist for particular data distributions and underlying assumptions. Franzese and Hays (2008) review the many cases of interdependence in international relations and comparative politics research, which could include civil war, offering insights into the many possible solutions for estimating parameters of interest (the advantages demonstrated empirically in Franzese & Hays, 2007, among other works). Examples of applied approaches in spatial panel data analysis include Geyer and Thompson (1992) for the explanation of a dependent autologistic model applied to the study of civil war in Ward and Gleditsch (2002). In the sub-national level Bosnian civil war example previously described, Weidmann and Ward (2010) use a similar technique to predict conflict event occurrence over time. For non-binary conflict outcomes, including the sum of violent event incidents that capture intensity rather than onset across observations, event count models (spatial generalized linear mixed models) that have been developed in other disciplines include potential applications in the field of civil war studies (e.g., Christensen & Waagepetersen, 2002; Wakefield, 2007). A multitude of options for estimating spatial econometric models are outlined in the free text of LeSage (1999).
Finally, it would be a mistake to conclude without making note of a central problematic issue in the quantitative geographical analysis of civil war: The dimensions of an areal unit can influence correlation analysis. This concern applies to geostatistical and predictive models that were previously reviewed. Such an effect of units’ dimension was first identified by Gehlke and Biehl in 1934, who found “certain effects of grouping upon the size of the correlation coefficient in census tract material” in the Journal of the American Statistical Association. More recently, Openshaw and Taylor (1979) wrote of “a million or so correlation coefficients” that one can report by varying the spatial dimensions of units of observation in statistical analysis. Openshaw (1983) later called this effect the “Modifiable Areal Unit Problem” (MAUP). The MAUP principle is simple: for one scale of analysis you may report a relationship between A and B that does not exist in another scale.2 Geographers have illustrated in great empirical detail such variability for the study of civil war (see Linke, Witmer, Holland, & O’Loughlin, forthcoming). A direct relation to the key geographical concept of hierarchical scale and the recent trend in disaggregating conflict analysis is clear. Some researchers previously cited (Linke & O’Loughlin, 2015; Schutte & Donnay, 2014), have tried to isolate the dimensions where a “neighborhood” or “context” effect is found, but this is difficult and decisive conclusions are absent in the literature to date.
The role of geography within the study of war and peace is increasingly popular, and “space” is now more often taken seriously as an essential component of understanding violence. Research areas in which spatial concepts have particular resonance include the location of conflict, the location(s) of violence within conflicts, diffusion and contagion patterns of violence, and as a context through which to understand the likelihood of conflict occurring.
Each conflict researcher and research agenda should complement wider academic and public aims, and be accessible to other geographers and those interested in violence across disciplines. There is a middle ground between quantified, seemingly reductive external interpretations of violence and idiosyncratic applications of deep theory to limited cases or with limited ability to generalize. For our purposes, the link between two styles of research and disciplinary agendas is a renewed dedication to the tenets of political geography. Some ongoing dilemmas in the study of civil war geographies can be solved by a reinvestment in the roots of our disciplinary inquiry. Contributions to civil war analysis could be substantially improved through deeper investigations into the “topography” of state power, to use the terms of Boone (2003); nuanced understandings of developing country statehood (Hagmann & Péclard, 2010); collective action by territorially defined actors (Weidmann, 2013); political ecology of civil war forms (Le Billon, 2004); social histories (Daly, 2012); and the transnational behavior of groups and states (Flint, Diehl, Scheffran, Vasquez, & Chi, 2009).
Agnew, J. (1987). Place and politics: The geographical mediation of state and society. London: Allen and Unwin.Find this resource:
Agnew, J. (1994). The territorial trap: The geographical assumptions of international relations theory. Review of International Political Economy, 1(1), 53–80.Find this resource:
Agnew, J., & Corbridge, S. (1995). Mastering space: Hegemony, territory and international political economy. London: Routledge.Find this resource:
Anselin, L. (1988). Spatial econometrics: Methods and models. Dordrecht, The Netherlands: Kluwer Academic Publishers.Find this resource:
Anselin, L. (1995). Local indicators of spatial association—LISA. Geographical Analysis, 27(2), 93–115.Find this resource:
Autesserre, S. (2011). The trouble with the Congo: Local violence and the failure of international peacebuilding. Ithaca, NY: Cornell University Press.Find this resource:
Bailey, T. C., & Gatrell, A. C. (1995). Interactive spatial data analysis. Essex, UK: Pearson Education Limited.Find this resource:
Balk, D., & Yetman, G. (2013). Gridded Population of the World (GPW), version 3, Center for International Earth Science Information Network (CIESIN), International Food Policy Research Institute (IFPRI). Available: http://sedac.ciesin.columbia.edu/data/collection/gpw-v3.
Bhavnani, R., Donnay, K., Miodownik, D., Mor, H., & Helbing, D. (2014). Group segregation and urban violence. American Journal of Political Science, 58(1), 226–245.Find this resource:
Bivand, R., Pebesma, E., & Gómez-Rubio, V. (2013). Applied spatial data analysis with R (2d ed.). New York: Springer.Find this resource:
Black, N. (2013). When have violent civil conflicts spread? Introducing a dataset of substate conflict contagion. Journal of Peace Research, 50(6), 751–759.Find this resource:
Boone, C. (2003). Political Topographies of the African State. New York: Cambridge University Press.Find this resource:
Braithwaite, A. (2010). Resisting infection: How state capacity conditions conflict contagion. Journal of Peace Research, 50(6), 751–759.Find this resource:
Braithwaite, A., & Johnson, S. (2012). Space-time modeling of insurgency and counterinsurgency in Iraq. Journal of Quantitative Criminology, 28(1), 31–48.Find this resource:
Brunsdon, C., & Comber, L. (2015). An introduction to R for spatial analysis and mapping. London: SAGE.Find this resource:
Brundson, C., Fotheringham, S., & Charlton, M. (1998). Geographically weighted regression. Journal of the Royal Statistical Society: Series D (The Statistician), 47(3), 431–443.Find this resource:
Buhaug, H. (2006). Relative capability and rebel objective in civil war. Journal of Peace Research, 43(6), 691–708.Find this resource:
Buhaug, H., & Gates, S. (2002). The geography of civil war. Journal of Peace Research, 39(4), 417–433.Find this resource:
Buhaug, H., & Rød, J. K. (2006). Local determinants of African civil wars, 1970–2001. Political Geography, 25(3), 315–335.Find this resource:
Buhaug, H., & Gleditsch, K. (2008) Contagion or confusion? Why conflicts cluster in space. International Studies Quarterly, 52(2), 215–233.Find this resource:
Buhaug, H., Gleditsch, K., Holterman, H., Østby, G., & Tollefsen, A. F. (2011). It’s the local economy stupid! Geographical wealth dispersion and conflict outbreak location. Journal of Conflict Resolution, 55(5), 814–840.Find this resource:
Buhaug, H., & Lujala, P. (2005). Accounting for scale: Measuring geography in quantitative studies of civil war. Political Geography, 24(4), 399–418.Find this resource:
Burke, M. B., Miguel, E., Satyanath, S., Dykema, J. A., & Lobell, D. B. (2009). Warming increases the risk of civil war in Africa. Proceedings of the National Academy of Sciences USA, 106(49), 20670–20674.Find this resource:
Cederman, L-E., Weidmann, N., & Gleditsch, K. (2011). Horizontal inequalities and ethno-nationalist civil war: A global comparison. American Political Science Review, 105(3), 478–495.Find this resource:
Chi, S. H., & Flint, C. (2013). Standing different ground: The spatial heterogeneity of territorial dispute. Geojournal, 78(3), 553–573.Find this resource:
Christensen, O. F. & Waagepetersen, R. (2002). Bayesian prediction of spatial count data using generalized linear mixed models. Biometrics 58, 280–286.Find this resource:
Cliff, A. D., & Ord, J. K. (1981). Spatial processes: Models and applications. London: Pion.Find this resource:
Collier, P., & Hoeffler, A. (2004). Greed and grievance in civil war. Oxford Economic Papers 56, 563–595.Find this resource:
Conley, T. G. (1999). GMM estimation with cross sectional dependence. Journal of Econometrics, 92, 1–45.Find this resource:
Cribb, R. (1991). Gangsters and revolutionaries: The Jakarta people’s militia and the Indonesian Revolution, 1945–1949. Honolulu, HI: University of Hawaii Press.Find this resource:
Crost, B., & Felter, J. (2015). Is conflict contagious? Evidence from a natural experiment. Working Paper 197. Households in Conflict Network, Institute of Development Studies, University of Sussex, UK.
Dahlman, C., & O Tuathail, G. (2005). Broken Bosnia: The localized geopolitics of displacement and return in two Bosnian places. Annals of the Association of American Geographers, 95(3), 644–662.Find this resource:
Daly, S. Z. (2012). Organizational legacies of violence: Conditions favoring insurgency onset in Colombia, 1964–1984. Journal of Peace Research, 49(3), 473–491.Find this resource:
Davis, D., &Moore, W. (1997). Ethnicity matters: Transnational ethnic alliances and foreign policy behavior. International Studies Quarterly, 41(1), 171–184.Find this resource:
De Juan, A. (2015). Long-term environmental change and geographical patterns of violence in Darfur, 2003–2005. Political Geography, 45, 22–33.Find this resource:
de Soysa, I. (2002). Paradise is a bazaar? Greed, creed, and governance in civil war, 1989–99. Journal of Peace Research, 39(4), 395–416.Find this resource:
Diggle, P. J. (1983). Statistical analysis of spatial point patterns. New York: Oxford University Press.Find this resource:
Dittrich Hallberg, J. (2012). PRIO conflict site 1989–2008: A geo-referenced dataset on armed conflict. Conflict Management and Peace Science, 29, 219–232.Find this resource:
Do, Q. T., & Iyer, L. (2010). Geography, poverty, and conflict in Nepal. Journal of Peace Research, 47(6), 735–748.Find this resource:
Dorff, C., & Ward, M. (2013). Networks, dyads, and the social relations model. Political Science Research and Methods, 1(2), 159–178.Find this resource:
Dowd, C., & Raleigh, C. (2013). The myth of Islamic terrorism and local conflict in Mali and the Sahel. African Affairs, 112(148), 498–509.Find this resource:
Enterline, A. (1998). Regime changes, neighborhoods, and interstate conflict, 1816–1992. Journal of Conflict Resolution, 42(6), 804–829.Find this resource:
Fearon, J., & Laitin, D. (2003). Ethnicity, insurgency, and civil war. American Political Science Review, 97(1), 75–90.Find this resource:
Finnegan, W. (1992). A complicated war: The harrowing of Mozambique. Berkeley: University of California Press.Find this resource:
Flint, C., Diehl, P., Scheffran, J., Vasquez, J., & Chi, S. H. (2009). Conceptualizing conflict space: Toward a geography of relational power and embeddedness in the analysis of interstate conflict. Annals of the Association of American Geographers, 99(5), 827–835.Find this resource:
Forsberg, E. (2013). Do ethnic dominoes fall? Evaluating domino effects of granting territorial concessions to separatist groups. International Studies Quarterly, 57(2), 329–340.Find this resource:
Forsberg, E. (2014). Transnational transmitters: Ethnic kinship ties and conflict contagion 1946–2009. International Interactions, 40(2), 143–165.Find this resource:
Franzese, R. J., & Hays, J. C. (2007). Spatial econometric models of cross-sectional interdependence in political science panel and time-series-cross-section data. Political Analysis, 15(2), 140–164.Find this resource:
Franzese, R. J., & Hays, J. C. (2008). Interdependence in comparative politics: Substance, theory, empirics, substance. Comparative Political Studies, 41(4–5), 742–780.Find this resource:
Gates, S. (2001). Recruitment and allegiance: The microfoundations of rebellion. Journal of Conflict Resolution, 46(1), 111–130.Find this resource:
Geyer, C. J., & Thompson, E. A. (1992). Constrained Monte Carlo maximum likelihood for dependent data. Journal of the Royal Statistical Society. Series B (Methodological), 54(3), 657–699.Find this resource:
Ghelke, C. E., & Biehl, K. (1934). Certain effects of grouping upon the size of the correlation coefficient in census tract material. Journal of the American Statistical Association, 29(185), 169–170.Find this resource:
Gleditsch, K. (2002). All international politics is local: The diffusion of conflict, integration, and democratization. Ann Arbor: University of Michigan Press.Find this resource:
Gleditsch, K. (2007). Transnational dimensions of civil war. Journal of Peace Research, 44(3), 293–309.Find this resource:
Gleditsch, K., & Ward, M. (2000). War and peace in space and time: The role of democratization. International Studies Quarterly, 44(1), 1–29.Find this resource:
Gleditsch, K., & Weidmann, N. (2012). Richardson in the Information Age: GIS and spatial data in international studies. Annual Review of Political Science, 15, 461–481.Find this resource:
Gleditsch, N. P., Wallensteen, P., Eriksson, M., Sollenberg, M., & Strand, H. (2002). Armed conflict 1946–2001: A new dataset. Journal of Peace Research, 39(5), 615–637.Find this resource:
Hagmann, T., & Péclard, D. (2010). Negotiating statehood: Dynamics of power and domination in Africa. Development and Change, 41(4), 539–562.Find this resource:
Hallberg, J. D. (2012). PRIO Conflict Site 1989–2008: A geo-referenced dataset on armed conflict. Conflict Management and Peace Science, 29(2), 219–232.Find this resource:
Hegre, H., & Sambanis, N. (2006). Sensitivity analysis of empirical results on civil war onset. Journal of Conflict Resolution, 50(4), 508–535.Find this resource:
Henderson, J. D. (1985). When Colombia bled: A history of the violencia in Tolima. Tuscaloosa: University of Alabama Press.Find this resource:
Hill, S., & Rothchild, D. (1986). The contagion of political conflict in Africa and the world. Journal of Conflict Resolution, 30(4), 716–735.Find this resource:
Houweling, H. W., & Siccama, J. G. (1985). The epidemiology of war, 1816–1980. Journal of Conflict Resolution, 29(4), 641–663.Find this resource:
Hsiang, S., Burke, M., & Miguel, E. (2013). Quantifying the influence of climate on human conflict. Science, 31(6151), 1235367.Find this resource:
Huth, P. (1996). Enduring rivalries and terroritorial disputes, 1950–1990. Conflict Management and Peace Science, 15(1), 7–41.Find this resource:
Joyce, K., & Braithwaite, A. (2013). Geographic proximity and third party joiners in militarized interstate disputes. Journal of Peace Research, 50(5), 595–608.Find this resource:
Kadera, K. (1998). Transmission, barriers, and constraints: A dynamic model of the spread of war. Journal of Conflict Resolution, 42(3), 367–387.Find this resource:
Kalyvas, S. (2006). The logic of violence in civil war. New York: Cambridge University Press.Find this resource:
Kirby, A., & Ward, M. (1987). The spatial analysis of peace and war. Comparative Political Studies, 20(3), 293–313.Find this resource:
König, M., Rohner, D., Thoenig, M., & Zilibotti, F. (2015). Networks in conflict: Theory and evidence from the great war of Africa. UBS Center Working Paper Series Working Paper No. 14. Available: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2704419.
Kulldorff, M., Heffernan, R., Hartman, J., Assuncao, R., & Mostashari, F. (2005). A space-time permutation scan statistic for disease outbreak detection. PLoS Medicine, 2(3), 216–224.Find this resource:
Le Billon, P. (2004). The political ecology of war: Natural resources and armed conflicts. Political Geography, 20(5), 561–584.Find this resource:
LeSage, J. (1999). Spatial econometrics. University of Toledo.Find this resource:
Linke, A.M., Witmer, F., Holland, E., & O’Loughlin, J. (forthcoming). The geography of conflict dynamics and mountainous terrain in the Caucasus region. Annals of the American Association of Geographers.Find this resource:
Linke, A. M., O’Loughlin, J., McCabe, T., Tir, J., & Witmer, F. (2015). Rainfall variability and violence in rural Kenya: Investigating the effects of drought and the role of local institutions with survey data. Global Environmental Change, 34, 35–47.Find this resource:
Linke, A. M., Schutte, S., & Buhaug, H. (2015). Population attitudes and the diffusion of political violence in sub-Saharan Africa. International Studies Review, 17(1), 107–125.Find this resource:
Linke, A. M., & O’Loughlin, J. (2015). Reconceptualizing, measuring, and evaluating distance and context in the study of conflicts: Using survey data from the North Caucasus of Russia. International Studies Review, 17(1), 1–19.Find this resource:
Linke, A. M., Witmer, F., & O’Loughlin, J. (2012). Space-time Granger analysis of the war in Iraq: A study of coalition and insurgent action-reaction. International Interactions, 38(4), 402–425.Find this resource:
Lunstrum, E. (2009). Terror, territory, and deterritorialization: Landscapes of terror and the unmaking of state power in the Mozambican “civil” war. Annals of the Association of American Geographers, 99(5), 884–892.Find this resource:
Lyall, J. (2009). Does indiscriminate violence incite insurgent attacks? Evidence from Chechnya. Journal of Conflict Resolution, 53(3), 331–362.Find this resource:
Maves, J., & Braithwaite, A. (2013). Autocratic institutions and civil conflict contagion. Journal of Politics, 75(2), 478–490.Find this resource:
Maystadt, J. F., & Ecker, O. (2014). Extreme weather and civil war: Does drought fuel conflict in Somalia through livestock price shocks? American Journal of Agricultural Economics, 96(4), 1157–1182.Find this resource:
Medina, R., & Hepner, G. (2011). Advancing the understanding of sociospatial dependencies in terrorist networks. Transactions in GIS, 15(5), 38–72.Find this resource:
Mesev, V., Shirlow, P., & Downs, J. (2009). The geography of conflict and death in Belfast, Northern Ireland. Annals of the Association of American Geographers, 99(5), 893–903.Find this resource:
Metternich, N., Dorff, C., Gallop, M., Weschle, S., & Ward, M. (2013). Antigovernment networks in civil conflicts: How network structures affect conflictual behavior. American Journal of Political Science, 57(4), 892–911.Find this resource:
Metternich, N., Minhas, S., & Ward, M. (2015). Firewall or wall on fire? A unified framework of conflict contagion and the role of ethnic exclusion. Journal of Conflict Resolution.Find this resource:
Most, B., & Starr, H. (1980). Diffusion, reinforcement, geopolitics, and the spread of war. American Political Science Review, 74(4), 932–946.Find this resource:
Murdoch, J. C., & Sandler, T. (2002). Economic growth, civil wars, and spatial spillovers. Journal of Conflict Resolution, 46(1), 91–110.Find this resource:
Newman, D. (2006). The lines that continue to separate us: Borders in our ‘borderless’ world. Progress in Human Geography, 30(2), 143–161.Find this resource:
Nordhaus, W. D. (2006). Geography and macroeconomics: New data and new findings. Proceedings of the National Academy of Sciences of the USA, 103(10), 3510–3517.Find this resource:
O’Loughlin, J. (1986). Spatial models of international conflicts: Extending current theories of war behavior. Annals of the Association of American Geographers, 76(1), 63–80.Find this resource:
O’Loughlin, J., & Anselin, L. (1991). Bringing geography back to the study of international relations: spatial dependence and regional context in Africa, 1966–1978. International Interactions, 17(1), 29–61.Find this resource:
O’Loughlin, J., Linke, A., & Witmer, F. (2014). The effects of temperature and precipitation variability on the risk of violence in sub-Saharan Africa, 1980–2012. Proceedings of the National Academy of Sciences, 111(6), 16712–16717.Find this resource:
O’Loughlin, J., & Witmer, F. (2012). The diffusion of conflicts in the North Caucasus of Russia, 1999–2010. Environment and Planning A, 44(10), 2379–2396.Find this resource:
O’Loughlin, J., Witmer, F., Linke, A., & Thorwardson, N. (2012). Peering into the fog of war: the geography of the Wikileaks Afghanistan War Logs, 2004–2009. Eurasian Geography and Economics, 51(4), 472–495.Find this resource:
Openshaw, S. (1983). The modifiable areal unit problem. Norwich, UK: Geo Books.Find this resource:
Openshaw, S., & Taylor, P. (1979). A million or so correlation coefficients: Three experiments on the Modifiable Areal Unit Problem. In N. Wrigley (Ed.), Statistical applications in the spatial sciences. London: Pion.Find this resource:
Oslender, U. (2007). Spaces of terror and fear on Columbia’s Pacific coast: The armed conflict and forced displacement among black communities. In D. Gregory & A. Pred (Eds.), Violent geographies: Fear, terror, and political violence (pp. 111–132). New York: Routledge.Find this resource:
Pickering, S. (2015). Introducing Spatial Grid Builder: A new dataset for creating geo-coded datasets. Conflict Management and Peace Science.Find this resource:
Pierskalla, J., & Hollenbach, F. (2013). Technology and Collective Action: The Effect of Cell Phone Coverage on Political Violence in Africa. American Political Science Review, 107(2), 207–224.Find this resource:
Porteous, J. D., & Smith, S. E. (2001). Domicide: The Global Destruction of Home. Montreal & Kingston, Canada: McGill-Queen’s University Press.Find this resource:
Prunier, G. (2008). Africa’s World War: the Rwandan Genocide and the Making of a Continental Catastrophe. New York: Oxford University Press.Find this resource:
Radil, S., & Flint, C. (2013). Exiles and arms: The territorial practices of state making and war diffusion in post–Cold War Africa. Territory, Politics, Governance, 1(2), 183–202.Find this resource:
Radil, S., Flint, C., & Chi, S.H. (2013). A relational geography of war: Actor-context interaction and the spread of World War I. Annals of the Association of American Geographers, 103(6), 1468–1484.Find this resource:
Raleigh, C. (2016). Pragmatic and promiscuous: Explaining the rise of competitive political militias across Africa. Journal of Conflict Resolution, 60(2), 283–310.Find this resource:
Raleigh, C., & Hegre, H. (2009). Population size, concentration, and civil war: A geographically disaggregated analysis. Political Geography, 28(4), 224–238.Find this resource:
Raleigh, C., Linke, A., Hegre, H., & Karleson, J. (2010). Introducing ACLED: An Armed Conflict Location and Event Dataset. Journal of Peace Research, 47(5), 651–660.Find this resource:
Raleigh, C., Linke, A., & O’Loughlin, J. (2014). Extreme temperatures and violence. Nature Climate Change, 4(Jan.), 76–77.Find this resource:
Raleigh, C., Choi, H. J., & Kniveton, D. (2015). The devil is in the details: An investigation of the relationships between conflict, food price and climate across Africa. Global Environmental Change, 32, 187–199.Find this resource:
Regan, P. (2000). Civil war and foreign powers. Ann Arbor, MI: University of Michigan Press.Find this resource:
Rustad, S., Rod, J. K., Larsen, W., & Gleditsch, N. P. (2008). Foliage and fighting: Forest resources and the onset, duration, and location of civil war. Political Geography, 27(7), 761–782.Find this resource:
Sack, R. D. (1986). Human Territoriality: Its Theory and History. Cambridge, U.K.: Cambridge University Press.Find this resource:
Saleyhan, I., & Gleditsch, K. (2006). Refugees and the spread of civil war. International Organization, 60(2), 335–366.Find this resource:
Sambanis, N. (2001). Do ethnic and nonethnic civil wars have the same causes? A theoretical and empirical inquiry (Part 1). Journal of Conflict Resolution, 45(3), 259–282.Find this resource:
Schutte, S. (2014). Geography, outcome, and casualties: A unified model of insurgency. Journal of Conflict Resolution, 59(6), 1101–1128.Find this resource:
Schutte, S., & Donnay, K. (2014). Matched wake analysis: Finding causal relationships in spatiotemporal event data. Political Geography, 41, 1–10.Find this resource:
Schutte, S., & Weidmann, N. (2011). Diffusion patterns of violence in civil war. Political Geography, 30, 143–152.Find this resource:
Singer, J. D., & Small, M. (1972). The wages of war, 1816–1965: A statistical handbook. New York: John Wiley.Find this resource:
Straus, S. (2006). The order of genocide: Race, power, and war in Rwanda. Ithaca, NY: Cornell University Press.Find this resource:
Sundberg, R., & E. Melander. (2013). Introducing the UCDP georeferenced event dataset. Journal of Peace Research, 50(4), 523–532.Find this resource:
Tatem, A., Gething, P., Bhatt, S., Weiss, D., & Pezzulo, C. (2013). Pilot high resolution poverty maps. University of Southampton/Oxford.Find this resource:
Toft, M., & Zhukov, Y. (2012). Denial and punishment in the North Caucasus: Evaluating the effectiveness of coercive counter-insurgency. Journal of Peace Research, 49(6), 785–800.Find this resource:
Tollefsen, A. F., Strand, H., & Buhaug, H. (2012). PRIO-GRID: A unified spatial data structure. Journal of Peace Research, 49(2), 363–374.Find this resource:
Tone, J. (1994). The fatal knot: The guerrilla war in Navarre and the defeat of Napoleon in Spain. Chapel Hill, NC: University of North Carolina Press.Find this resource:
Tyner, S. (2008). The killing of Cambodia: Geography, genocide, and the unmaking of space. Burlington, VT: Ashgate.Find this resource:
Tyner, S. (2011). “Genocide as reconstruction: The political geography of democratic Kampuchea. In S. Kirch, & C. Flint (Eds.), Reconstructing conflict: Integrating war and post-war geographies (pp. 49–66). Burlington, VT: Ashgate.Find this resource:
Vasquez, J. A. (1993). The War Puzzle. Cambridge, U.K.: Cambridge University Press.Find this resource:
Varshney, A. (2002). Ethnic conflict and civic life: Hindus and Muslims in India. New Haven, CT: Yale University Press.Find this resource:
Wakefield, J. (2007). Disease mapping and spatial regression with count data. Biostatistics, 8(2), 158–183.Find this resource:
Walter, B. (2006). Information, uncertainty, and the decision to secede. International Organization, 60(1), 105–135.Find this resource:
Walther, O., & Christopoulos, D. (2012). A social network analysis of Islamic terrorism and the Malian rebellion. CEPS/INSTEAD Working Paper No 2012-38.
Ward, M., & Gleditsch, K. S. (2002). Location, location, location: An MCMC approach to modeling the spatial context of war and peace. Political Analysis, 10, 244–260.Find this resource:
Ward, M., & Gleditsch, K. S. (2007). An introduction to spatial regression models in the social sciences. Thousand Oaks, CA: SAGE.Find this resource:
Weidmann, N. B. (2011). Violence “from above” or “from below”? The role of ethnicity in Bosnia’s civil war. Journal of Politics, 73(4), 1178–1190.Find this resource:
Weidmann, N. B. (2015). Communication networks and the transnational spread of ethnic conflict. Journal of Peace Research, 52(3), 285–296.Find this resource:
Weidmann, N. B., & Callen, N. (2013). Violence and election fraud: Evidence from Afghanistan. British Journal of Political Science, 43(1), 53–75.Find this resource:
Weidmann, N. B., Kuse, D., & Gleditsch, K. (2010). The geography of the international system: The CShapes dataset. International Interactions, 36(1), 86–106.Find this resource:
Weidmann, N. B., Rød, J. K., & Cederman, L. E. (2010). Representing ethnic groups in space: A new dataset. Journal of Peace Research, 47(4), 491–499.Find this resource:
Weidmann, N. B., & Saleyhan, I. (2012). Violence and ethnic segregation: A computational model applied to Baghdad. International Studies Quarterly, 57(1), 52–64.Find this resource:
Weidmann, N. B., & Ward, M. (2010). Prediction conflict in space and time. Journal of Conflict Resolution, 54, 883–901.Find this resource:
Whitaker, B. (2003). Refugees and the spread of conflict: Contrasting cases in central Africa. Journal of Asian and African Studies, 38(2–3), 211–231.Find this resource:
Wischnath, G., & Buhaug, H. 2014. On climate variability and civil war in Asia.” Climate Change, 122(4), 709–721.Find this resource:
Wood, W. (2001). Geographic aspects of genocide: A comparison of Bosnia and Rwanda. Transactions of the Institute of British Geographers, 26(1), 57–75.Find this resource:
Wucherpfennig, J., Weidmann, N. B., Girardin, L., Cederman, L.E., & Wimmer, A. (2011). Politically relevant ethnic groups across space and time: Introducing the GeoEPR dataset. Conflict Management and Peace Science, 28, 423–437.Find this resource:
Zhukov, Y. (2012). Roads and the diffusion of insurgent violence: The logistics of conflict in Russia’s North Caucasus. Political Geography, 31, 144–156.Find this resource:
Zhukov, Y. (2013). An epidemic model of violence and public support in civil war. Conflict Management and Peace Science, 30(1), 24–52.Find this resource:
Zhukov, Y., & Stewart, B. M. (2013). Choosing your neighbors: Networks of diffusion in international relations. International Studies Quarterly, 57, 271–287.Find this resource:
(1.) The “geography of civil war” also means more than focusing on terrain (e.g., mountains and forest cover) or environment (e.g., drought) as explanatory forces, even though these topics are noteworthy facets of conflict research (Fearon & Laitin, 2003; Burke, Miguel, Satyanath, Dykema, & Lobell, 2009, respectively). We also briefly address some geographical issues or dilemmas that may lead to war, such as territorial disputes (e.g., Vasquez, 1993; Huth, 1996, but this area of research is dominated by a focus on inter-state (rather than intra-state, or civil) war and may or may not be spatial research methodologically.