This paper investigates the effect of colonial legacy on brutality seen during post-colonial civil wars. Specifically, I seek to find strong relationships between the type of colonizer and types of violence perpetrated during later civil wars, looking to instances of mass killing, genocide, and politicide. This work expands upon existing literature in a novel way, introducing colonial legacy and colonial policies as underlying causal mechanisms for future brutality seen during conflict. Results suggest there are strong relationships between former colonies and instances of mass killings, genocide, and politicide. Former British colonies are most significant across all three tests, as are Belgian colonies, indicating perhaps a mix of colonial policies leading to greater likelihoods of violence among former colonies. Results also suggest that keeping civilian support for rebel groups low is an important factor in reducing the likelihood of mass killings, and geno-/politicide. This paper empirically expands current thought on brutality seen in civil war, tying together history with the present.
Table of Contents:
“I and the public know
What all schoolchildren learn,
Those to whom evil is done
Do evil in return.” – W.H. Auden
Following the end of World War II, many former colonies gained their independence after years of foreign rule. While this should have been a time of great jubilation within these former colonies, oftentimes violence immediately sprang forth. The Democratic Republic of Congo has seen decades of bloody and brutal violence with estimates of more than 5 million people killed in the conflict still raging. Syria is currently witnessing its own civil war, with reports of more than 100,000 deaths in the 27 months of fighting (New York Times, June 27, 2013), and no immediate end in sight. Why does such violence occur in these states? Why are so many killed in a civil war involving a former colony? Certainly, all civil wars see violence and death, but the degree and brutality of those in former colonial holdings seems consistently higher than non-colonies. There are many factors that determine when, how, or why a civil war occurs and why one may be bloodier than another—but could the colonial history of a state truly be a factor in the brutality seen during civil war? Does a colonial legacy impact instances of this extreme brutality?
Despite its importance, little attention has been given to this question. Most academic research thus far has focused on colonial styles and civil war in general (Blanton, Mason, and Athow 2013; Henderson and Singer 2000)or brutality in general (Heger and Salehyan 2007; Valentino, Huth, and Balch-Lindsay 2004; Harff 2003; Gurr et al. 2005; Mason and Krane 2013; R. M. Wood 2010), without taking the other into account. Understanding the impact of colonial legacy on the severity of civil war is of great importance, particularly at this point in time with so many civil conflicts in Africa, the Middle East, Asia, and South America flaring up—many of which are former colonies.
During Belgian colonization of the Congo, the maiming and killing of indigenous persons by Belgian colonizers was common practice (New York Times, Oct 25, 1909), with reports of entire villages being wiped out. Frequently, hands or arms would be cut off by the Belgians as punishment for those who would not work (New York Times, Nov. 24, 1909). Life under colonial rule was undoubtedly harsh. Once free from their colonizers, a new state was formed, but peace would not last. By the 1990’s, violence erupted with rebel factions perpetrating brutal acts of violence similar to that seen during colonial rule (Keesing's Record of World Events, April 2003), vying for power against multiple warlords. This type of violence is seen time and again in these former colonies through Africa, the Middle East, South Africa, and Asia. Put simply, civil wars in states which have a colonial legacy are more likely to see higher rates of brutality.
This paper expands upon the gaps left in previous literature that examines the effect of colonial legacy on brutality by identifying reasons why certain types of violence occur in former colonies. I demonstrate that, due to lasting colonial legacy, there is an increased likelihood of post-colonial states experiencing higher rates of brutality (mass killings, genocide, and politicide). Analyses of my results strongly support the proposition that brutality is generally more prevalent across former colonies. With ongoing conflict in the Democratic Republic of Congo, Central African Republic, Syria, and Myanmar to name a few, I find that this is an important and timely addition to the literature, which will shed light upon and add to the discourse of brutality in civil wars.
I begin this research with a review of literature, focusing on work addressing brutality—specifically instances of mass killing, genocide and politicide. I then discuss my theoretical framework, proposing that colonial legacy has an impact on instances of mass killings in general, and subsequently that certain types of colonial legacies should lead to higher instances of politicide/genocide. Research methods will then be discussed, and I will test this novel theory using logistic regressions. Finally, I discuss the results of this analysis and conclude with areas of future research and the impact this study has on current literature.
Scholars have pointed to a decrease in the overall bloodiness of civil war in the post-Cold-War era (Melander, Oberg, and Hall 2009; Lacina 2006). I do not set out to refute this argument, nor augment it. The focus of this research involves brutality—specifically mass killings, genocide and politicide—and the effect on former colonial holdings.
Many scholars have focused on types of violence perpetrated by the state against its civilian population (Heger and Salehyan 2007; Eck and Hultman 2007). While few have specifically looked to colonial history as a source for future brutality, there is much overlap and reason to theoretically link the two, which I present throughout this section. This work is imperative to determining the degree to which extreme violence occurs in certain states, which will then be analyzed against those states’ colonial legacy.
“Death squad violence,” according to Mason and Krane (1989), and the targeting strategy utilized by the regime, is shown to drastically change the levels of civil violence. They find that repressive violence does, in fact, decrease oppositional support, while the increase in that violent repression increases oppositional support, while indiscriminate killings turn those who may not have been involved into active dissidents. This finding is important in the context of colonial legacy and mass killings in that it sets a precedent for the cyclical nature of the conflict, and giving rise to the idea that—through policies of ethnic stratification and repressive tactics utilized by colonial regimes—more may be thrust into the conflict and fight more violently than they otherwise would have.
State violence levels drastically change the way oppositional forces act against civilians (R. M. Wood 2010). As Wood points out, increases in rebel strength actually decrease levels of violence upon civilians whereas foreign assistance tends to increase levels of violence. Since post-colonial state structures are largely set-up according to colonial rule, any opposition to those structures pre-independence should, ceteris peribus, remain post-independence, and cause further instability. Hegar and Salehyan (2007) find that the same rebel strength is associated with a higher overall number of battlefield deaths, as the rebel groups increase their attack capacity and intensity. This is a highly significant finding in that the regime in power will then have fewer options to win: negotiate or step up their own violence against rebel groups and those associated with them, as noted in other work by Wood, Kathman, and Gent (2012). The idea that brutality against civilians in this instance is imperative to victory is then further solidified (Kalyvas, 2006; Wood, 2010).
Another factor pointed to by some scholars that may help to explain extreme brutality is regime type. Highly democratic states are less likely to use mass killings, genocide, or politicide in order to win (Valentino, Huth, and Balch-Lindsay 2004), whereas semi-democracies are most likely to use these tactics in order to fulfill their objectives (Henderson and Singer 2000). When state capacity is low and violent action is taken, many governments may have no other choice than to institute violence against their citizens, particularly if there is a direct threat to the regime (Demeritt; Valentino, Huth, and Balch-Lindsay 2004).
Work by Lange and Dawson (2009) began to look at the effect of colonialism on post-colonial civil violence, but fell short of explaining the correlation. Interestingly, they found that there is little evidence that colonialism is a cause of civil violence, but that there are some forms of colonial violence that do appear to increase the risk and intensity of post-colonial violence. These findings do tend to fall in line with those of other scholars—both looking at colonial legacy and violence—but refute the proposition that the two are correlated (Blanton, Mason, and Athow 2013; Harff 2003; Reed M. Wood and Gent 2012; Reed M. Wood 2010; Henderson and Singer 2000).
Previous work has shown a greater preponderance of civil war in post-colonial states, though much of this attribution has been focused on the degree or level of democracy within said countries, with semi-democratic institutions bringing about more conflict (Henderson and Singer 2000; Blanton, Mason, and Athow 2013). These findings are important, giving credence to the idea that there may be a more direct correlation between colonial rule and subsequent civil war, and providing the basis to which many other works have been published looking into greater detail at the ways in which colonial legacy shapes the future of civil war.
Others have focused on the effect of colonial style on ethnic conflict following independence, concentrating on ethnic stratification that formed out of British or French styles of governance (Blanton, Mason, and Athow 2013). This is a truly important link to make in studying the effect of colonial legacy on civil war. Finding that the style of governance, indeed, has an impact on post-colonial violence sets the stage for a portion of my theory—the ethnic stratification as defined by Blanton et al. changes the types of violence seen—described below. Harff (2003) further expands upon this concept by measuring specifically the regime type in relation to incidences of mass murder, pointing to a higher probability in autocratic regimes with high rates of ethnic stratification. Though neither point to colonial legacy as a cause for increased violence, it is reasonable to test the likelihood that the two are, in fact, related.
Thinking through the proposition that colonial legacy has an effect (either positive or negative) on brutality seen during post-colonial civil wars, I believe that there will be greater instances of brutality in former colonies than in states with no colonial legacy. This will be further expanded in the next section.
Motivations for colonization affect how post-colonial regimes use brutality. For this research, I propose that there are two primary motivations for colonization, and thus types: 1) extractive (Blanton, Mason, and Athow 2013; Reed M. Wood and Gent 2012), and 2) assimilative (Blanton, Mason, and Athow 2013; Clapham, 1985). Regional, or internal colonization (Lange and Dawson 2009)has occurred frequently throughout history, but is not included in this study, as the effects are arguably much different than those by foreign powers.
Extractive colonization occurs when a party invades, colonizes, and extracts natural resources, goods, or services on a large scale with the intent or interest of furthering the foreign state’s interests above those of the indigenous population being colonized. This type of colonization seeks to extract maximum effect at the expense of the colonized people, with little regard for their wishes or well-being, setting up a slave-master relationship between colonizer and the colonized. This was seen with the Belgian colonization of the Congo, or British colonization of the Gold Coast (now Ghana). It is important to make this distinction first, as the type of colonization will affect the way the colony was governed, as described below.
In assimilative colonization (colonization with little economic gain or that done for the expansionist “land-grab”), colonizer-colonized relationships resemble that of occupier-occupied. As such, a higher military/police presence should reasonably be expected with some forms of repression, though large-scale and wanton violence against the population by the colonizers themselves should not regularly occur unless the colonizer is facing indigenous opposition, and thus, an insurgency. With the cooption of the population, the colonizer was better able to exert control. This should translate to indigenous cooperation without direct violent action against the populace. Such was the case in the French colonization of Lebanon and Algeria.
The reason then, for colonization, sets the stage for the colonial regime praxis. Arguably, we can assume that each colonizer will operate as fits in its respective rationales. For this study, I chose to analyze British, Belgian, French, Spanish, Portuguese, and Italian colonization styles and make generalized assumptions on their regime types utilizing case studies. Each colonizer ruled consistently—such as the British who ruled through ethnic stratification, or the French who ruled largely through incorporation and assimilation (Blanton, Mason, and Athow 2013)—and are thus grouped: British and Belgian colonies under Extractive, and French, Spanish, and Portuguese under Assimilative.
Brutality here refers to cases where civilian populations are specifically targeted and systematically murdered, by state-sponsored or by militant opposition groups with the aim of achieving their goal—winning the war. The levels of brutality are closely tied to those seen during colonization, in that they will mirror the violence perpetrated on the group by the colonizers. One such example was mentioned previously; the Democratic Republic of Congo. Another such example is that of French Lebanon. In Lebanon, those most like the French—Maronite Christians—were the ones largely placed in the seat of power. Once French rule ended, however, heavy fighting that lasted for decades broke out, with an estimated 150,000 killed (Los Angeles Times, March 08, 1991; Keesing's Record of World Events, June, 1976). Due in large part to the structure of the colonial regime, this violence was fostered seemingly before it began. Thus, because of the colonial legacy, future civil wars fought post-colonial independence should be more bloody and violent with mass killings, genocide, and politicide than those which have not seen a colonial history due to the factors surrounding the colonial legacy to begin with.
Based upon these arguments, I develop three interrelated hypotheses about colonial legacy and heightened brutality in civil war. Although there are differences between colonizers and colonies, I seek to support generalizable results regarding major European colonial powers and their effect on mass killings in civil war.
H1: Former colonies are more likely to see higher rates of mass killings.
H2a: Former colonies which saw extractive regimes are more likely to see higher incidences of genocide.
H2b: Former colonies which saw assimilative regimes are more likely to see higher instances of politicide.
My empirical work here uses unit of analysis of civil war country-year, as a pooled cross sectional data set. I use the data-sets on insurgency violence in civil war compiled by Valentino, et al. (2004), colonial history compiled by Hensel (2009) and operational definitions of genocide and politicide as posited by Harff (2003). My dependent variable 1, mass killing, comes from the insurgency violence in civil war (Valentino, Huth, and Balch-Lindsay 2004), and variables 2—genocide—and 3—politicide—come from Harff (2003). These data provide binary variables for each incidence, based on operationalization by the respective authors. Harff describes geno-/politicide as a violent act taken by a regime against an identifiable group designed to cause mass suffering and a threat to the existence and survival of that group. The data are merged and coded according to instance. This allows me to empirically test my initial hypothesis (H1) and subsequent hypotheses (H2a and H2b), using controls which have been previously tested.
I operationalize colonial legacy by coding first for the major historical European colonizers; Britain, Spain, Belgium, France, and Portugal from the colonial history dataset (Hensel, 2009). I chose to omit individual coding for both Italy and the Netherlands, as there were too few cases and their addition as independent variables would likely skew my results. I attempted multiple variations on coding and found that placing Italy and Netherlands in the Other Colony list produced a better model for this study and omitted fewer observations.
Others listed as colonizers in the data do not coincide with those listed above, as they are oftentimes referred to as internal colonies (Lange and Dawson 2009)—meaning they are conquered by neighboring powers and ruled as though they were integral parts of that state’s territory, such as Austro-Hungary, Ottoman-Turk, Russia (formerly USSR), Korea, the United States, Yugoslavia, and others—and do not coincide with my definition of colonialism here.
I will follow work laid out by Valentino, et al (2004), Henderson and Singer (2000), and others who begin looking at state violence in civil wars after 1945. This start date accurately reflects not only the shift in global power, but the handover of many former colonies to their inhabitants. Following these handovers, the preponderance of post-colonial civil wars was more likely (Henderson and Singer 2000; Blanton, Mason, and Athow 2013), and thus we have more instances to measure.
I use the operational definition of mass killing as “the intentional killing of a massive number of noncombatants,” and use the same definition for “massive number” as 50,000 intentional deaths over the course of five or fewer years (Valentino, Huth, and Balch-Lindsay 2004). Using this definition reduces my overall number of observations, but should lead to more accurate results, as I seek not to find each instance where civilians were targeted at low levels during post-colonial conflict, but the instances where they were systematically targeted.
I also make the distinction here, unlike Valentino, et al., of instances of genocide and politicide during post-colonial civil wars in order to better account for heightened brutality during civil wars. I operationalize these terms using Harff’s (2003) operational guidelines, and applying two additional analyses of observations of genocide and politicide to the first of mass killings. Valentino et al. (2004) did not find this distinction necessary, as they noted great similarity in mass killings and geno/politicide. For this study, it is important to utilize the operational definitions, code them as dummy variables, and run an analysis on them as well. This will give a better idea as to which types of violence are more likely to be seen by former colonies during post-colonial conflict, based on colonial type. I thus coded two additional variables—gcide and pcide—and, using data from Harff (2003) matched up instances of genocide and politicide with the data from Valentino et al. (2004). Since the temporal measures for each varied slightly, I aggregate incidences of genocide and politicide with instances of mass killings where there was a perfect temporal match or when there was an occurrence of geno-/politicide prior to the end date within the mass killings data. Some instances which did not match up with the Valentino et al. criteria, such as the politicide in Iraq from ’88-’91, were not coded as having seen geno-/politicide. While this reduces again my overall number of observations, it most accurately reflects incidences being studied.
To account for the persistence of colony legacy and its effect on civil war brutality, I operationalize my primary independent variable in three ways: 1) a post-independence up to 50 year measure, 2) a 50 to 100 year post-independence measure, and 2) a 100+ year post-independence measure. By using these temporal criteria, the persistence of colonial legacy can be estimated to determine both short and long term effects of colonial legacy on brutality during post-colonial civil wars. This measure is gathered from the date of colonial independence—indep variable (Hensel 2009)—subtracted from the start date of the incidence—syear variable (Valentino et al. 2004). I utilize these dichotomous variables in order to control for the time since independence to provide measures with which I can easily examine the effect of time, if any, on colonial legacy. I then created three dummy variables, indepyr, fiftyyear, and hundredyear, and coded for each instance.
Another factor for control is that of multiple colonizers. Areas that saw more than one colonial regime—such as Somalia which saw both British and Italian colonizers—will have the last colonizer as the basis for “colonial legacy.” This measure should more accurately represent the type of regime which blossoms post-independence and should logically have the most influence on violence and post-colonial regime. While this measure does not account for the length of time a colonizer may have been in control of the colony, I follow Hensel’s (2009) dataset which currently accounts for the last colonizer. This variable will control for potential biases in the data that may skew results, which may result from the length of time the original colonizer had control.
I will use logit analysis because I am working with a binary dependent variable. Previous scholars (Hensel 2001; Valentino et al. 2004) have noted that a binary dependent variable is best measured by a logistic or standard probit regression.
Results and Discussion
Three stages of logit analysis were performed to test my hypotheses. In the first stage, I examined the relationship between instances of mass killings and colonial history in all 166 observations in the data. In the second, I tested in 68 observations the hypothesis that genocide is more prevalent in former colonies which saw extractive colonial policies. In the third stage, I tested in more than 146 observations the hypothesis that politicide is more prevalent in former colonies which saw assimilative colonial policies. Due to perfect collinearity in tests 2 and 3, observations were dropped.
Results for the three models are listed in Table 1. The coefficient and robust standard errors are listed for each variable, and Z score is marked for significance with asterisk. Table 2 and Table 3 report findings of relative risk probability that certain former colonies will see mass killings, genocide or politicide, using means for Log Population and Conflict Duration and other variables binary.
The results strongly support my hypothesis that instances of mass killing are greater in former colonies than in those with no colonial legacy. Of particular interest is the high statistical significance of British and Belgian colonial holdings to instances of mass killings, with Spanish, French and Portuguese holdings showing moderate significance, and Other colonies not statistically significant. These instances of mass killing occur within these former colonies due to their respective colonial legacies and through colonial policies, which affect the way these states behave post-independence.
Interestingly, former Spanish colonies were negatively and nominally significantly correlated (at p<0.1) with mass killings. This could be because Spain was able to set up better systems of government with better hand-overs to the locals prior to independence, or that they generally assimilated the population as well as themselves into the new colony. Anecdotal evidence could be taken from some of the former Spanish colonies in the Philippines.
My tests for both genocide and politicide reduced my total number of observations. Starting with 166 total cases, there was a decrease to 68 in genocide test, and 146 in politicide test. Instances of genocide and politicide are indeed rare, which limited my sample size. Despite the rarity, however, I was still able to draw significant conclusions from the results of these analyses.
The results strongly support Hypothesis 2a that genocide would occur more frequently in former colonies which saw extractive policies. Both Britain and Belgium see statistical significance, as does France. Spanish, Portuguese and Other former colonial holdings were omitted due to perfect collinearity, which reduced the total number of observations, and perhaps skewed results. Despite the dropped observations, results still indicate that colonies with highly extractive policies hold strong significance to instances of genocide.
Results for Hypothesis 2b are also strongly supported, with France, Portugal, and British observations seeing high statistical significance, and Belgian/Other observations seeing moderate significance. With both France and Portugal being assimilative colonizers, I expected their significance to be high. Unexpectedly, however, was the strong significance of p<.01 of Britain and p<.05 for Belgium. Perhaps, then, there is room to suggest that, while hypotheses 2a and 2b are statistically significant, they may not be entirely accurate. Their respective significance could be related to the few observations I have in the data for geno-/politicide. It may also indicate that Britain in particular used both assimilative and extractive policies, depending on the colony. This is an interesting finding and may prove to be of value in later research discussing why some former colonies see violence that others do not.
Unexpectedly, former British colonies show extremely high statistical significance in each test, as do Belgian colonies. While I expected perhaps some overlap between extractive and assimilative colonial regimes, I had not anticipated such strong results. This could be emblematic of colonial rule by major powers that were, for much of the colonial rule, at war at home rather than a byproduct specifically of regime type. Granted, there are many other unobserved actions being taken within the colony that may simply not be addressed in this research.
All three tests show strong significance for High civilian support. This makes sense, as an insurgency requires a great number of supportive civilians in order to be effective at overthrowing a regime which, in turn, provides more perceived need for the government to crack down with violent force. This provides the conduit through which extreme action may be taken by the government in order to maintain legitimacy, and we should see higher instances of mass killings and geno-/politicides. The data does seem to support this idea, pointing to the ever growing importance of governments’ ability to, if not garner more civilian support for itself, keep rebel groups from receiving civilian support.
Temporal controls were also found to be a statistically significant factor in the results. Interestingly, the 0-49 years post-independence period from the former colonizer is unlikely to see mass killings, politicide or genocide, while the 50-99 year period is negatively significant in instances of mass killings. Observations for geno-/politicide were dropped due to collinearity. While statistically significant, I did not set out to determine what effect the length of time following colonial independence has on instances of mass killings or geno-/politicide, nor do I have a hypothesis as to why. These variables were chosen as controls to account for temporal domains in which I could easily measure the effect, if any, on the colonial legacy.
Following the three logit regressions, I ran marginal effects to determine probabilities of mass killings, genocide and politicide during civil wars occurring in these former colonies. Results were truly interesting. Former British colonies, based on tests 1-3, were most significantly correlated with instances of mass killings, genocide, and politicide, and also showed very high probabilities of mass killings when other factors are present. For example, in Table 2, I show the marginal effect data for mass killings to occur in a former British colony where there is little/no civilian support, mid-level polity, a guerrilla threat to the regime, and current guerrilla war between 50-99 years post-independence(1). In the second test I changed only civilian support and saw dramatic changes in estimation of the likelihood for a mass killing.
This is a novel finding, in that we can now extrapolate with some degree of confidence that a mass killing may occur statistically, given the presence of a number of other factors. For example, in Table 3 are the results of testing the current state of affairs in Nigeria between the government and Boko Haram: Nigeria is a former British colony. It is 50-99 years post-independence, has mid-polity, is facing a guerrilla threat, identity, and guerrilla war, but there is no appreciable civilian support for Boko Haram. Log pop and duration are held at means, and all other variables are at 0. Results show that currently, Nigeria has an 8.26 percent likelihood of seeing a mass killing in its current conflict. If, however, the government loses civilian support and those on the fence move to support Boko Haram, results show a 95.3 percent likelihood of seeing a mass killing in the conflict. Results show that what the government does during times of rebellion greatly affects the likelihood of mass killings based on the colonial past. That such a modest change (in civilian support) significantly increases the likelihood of mass killing can be tied to colonial legacy.
Again, due perhaps to the rare nature of genocide and politicide, I was unable to produce consistent results running marginal effects across former colonizers for genocide and politicide. It is likely that with better data on genocide and politicide I would be able to produce more reliable results that would give a better sense of likelihood of geno-/politicide, which is an area for future research. Despite the inability to run marginal effects models, the data still shows promise for further discussion.
The main focus of my research revolves around the effect of colonial legacy on brutality during civil wars. Specifically, I examined its effect on instances of mass killings, genocide, and politicide. I conducted three tests around the hypotheses: (1) Former colonies are more likely to see higher rates of mass killings, (2) former colonies that saw extractive regimes are more likely to see higher incidences of genocide, and (3) former colonies that saw assimilative regimes are more likely to see higher instances of politicide, finding support for each. This research was a novel addition to the literature on both colonial legacy and brutality.
While much of the research showed to be statistically significant, there are problems in the data itself. The loss of observations due to collinearity is not ideal, and likely skews some of the results for genocide and politicide. This was particularly prevalent in my testing of genocide, which dropped 98 total observations. Again, this is likely due to the relatively uncommon nature of genocide and politicide, but could be better controlled for utilizing different data. For preliminary results, however, this shows promise as an area which could benefit from further research.
That the temporal controls seem to be a statistically significant factor is also intriguing. There are a multitude of factors that may help to explain why this is so, though I did not set out to explain the significance here. This, too, then, should be an area for further study and thought—why are the first 99 years post-colonial independence statistically significant in relation to mass killings, genocides, and politicides?
With better data for observations of genocide and politicide, a more robust study could be conducted to better understand the relationship between colonial legacy and brutality. New data would also better account for the few observations of geno-/politicide, giving a much better sense to the effect colonial legacy has on these instances of brutality. This is important for many reasons—as policy and even into analysis and advisement to local, regional, and international governments to better account for these violent acts.
That so little has been done to address the effect of colonialism on brutality in civil wars is striking. With conflict almost continually ongoing throughout the world in hotspots like the DRC, Syria, Somalia, Egypt, Burma, and Palestine/Israel, we should look to the past for clues to the present and future in order to mitigate the effects of colonialism. With further study and deliberate movement by governments, it may not be entirely impossible to outlive the past and see a more peaceful future.
- Blanton, Robert, T. David Mason, and Brian Athow (2013). “Colonial Style and Post-Colonial Ethnic Conflict in Africa.”Journal of Peace Research 38 (4): 473–491.
- Cowell, Alan. (2013). "War Deaths in Syria Said to Top 100,000." New York Times June 26, http://www.nytimes.com/2013/06/27/world/middleeast/syria.html?_r=2&.
- Clapham, Cristopher (1985). Third World Politics: An Introduction. Madison, WI: University of Wisconsin Press.
- Demeritt, Jacqueline H. R.. “Delegating Death : Military Intervention and Government Killing.” http://www.psci.unt.edu/~demeritt/research_files/dd.pdf
- Eck, K., and L. Hultman (2007). “One-Sided Violence Against Civilians in War: Insights from New Fatality Data.” Journal of Peace Research 44 (2) March 1: 233–246.
- Gurr, Ted Robert, Barbara Harff, Roy Licklider, Monty Marshall, Frederic Pearson, Jeffrey Pickering, Patrick Regan, and William Stanley (2005). “International Intervention and the Severity of Genocides and Politicides.” International Studies Quarterly 49: 363–387.
- Harff, Barbara (2003). “No Lessons Learned from the Holocaust? Assessing Risks of Genocide and Political Mass Murder Since 1955.” American Political Science Review 97 (01) March 12: 57. doi:10.1017/S0003055403000522.
- Heger, Lindsay, and Idean Salehyan (2007). “Ruthless Rulers: Coalition Size and the Severity of Civil Conflict.” International Studies Quarterly (51): 385–403.
- Henderson, Errol A, and J David Singer (2000). “Civil War in the Post-Colonial World, 1946-92.” Journal of Peace Research 37 (3): 275–299.
- Hensel, Paul R. (2001). "Contentious Issues and World Politics: The Management of Territorial Claims in the Americas, 1816-1992." International Studies Quarterly 45 (1): 81-109
- Hensel, Paul R. (2009). "ICOW Colonial History Data Set, version 0.4." http://www.icow.org/colhist.html.
- Kalyvas, Stathis N. (2006). The Logic of Violence in Civil War. Cambridge: Cambridge University Press.
- Keesing's Record of World Events. 1976. "June 1976 - Internal Developments, September 1975 to May 1976 - State of Virtual Civil War - Syrian Intervention - Election of Successor to President Frangie." 22 June: 27765.
- Keesing's Record of World Events (2003)."Democratic Republic of Congo." 49 April: 45331.
- Lacina, Bethany (2006). "Explaining the Severity of Civil Wars." Journal of Conflict Resolution 50 (2): 276-89.
- Lange, Matthew, and Andrew Dawson (2009). “Dividing and Ruling the World ? A Statistical Test of the Effects of Colonialism on Postcolonial Civil Violence.” Social Forces 88 (2) December: 785–817.
- Los Angeles Times (1991)."Casualties of Mideast Wars." (March 08) http://articles.latimes.com/1991-03-08/news/mn-2592_1_civil-war
- Mason, T. David, and Dale A Krane (1989). “The Political Economy of Death Squads: Toward a Theory of the Impact of Terror.” International Studies Quarterly 33 (2): 175–198.
- Melander, E., M. Oberg, and J. Hall (2009). “Are ‘New Wars’ More Atrocious? Battle Severity, Civilians Killed and Forced Migration Before and After the End of the Cold War.” European Journal of International Relations 15 (3) August 20: 505–536.
- New York Times. 1909. "Congo Atrocities Revealed." New York Times October 25. http://query.nytimes.com/mem/archive-free/pdf?res=F60D10FF3D5412738DDDAC0A94D8415B898CF1D3
- New York Times. 1909. "Tells of Atrocities in Belgian Congo." New York Times November 24. http://query.nytimes.com/mem/archive-free/pdf?res=F10C14F83D5412738DDDAD0A94D9415B898CF1D3
- Valentino, Benjamin, Paul Huth, and Dylan Balch-Lindsay (2004). “‘Draining the Sea’: Mass Killing and Guerrilla Warfare.” International Organization 58 (02) May 19: 375-407.
|Polity IV >6||-3.682***
|Polity Score -6—6||-0.149
|High civilian support||4.767***
|0-49 years post-colonial||-3.182***
|50-99 years post-colonial||-6.296***
|100+ years post-colonial||1.561
Note: Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
|Probability of Event||Standard Error|
(1) Mid-Polity & No Civ. Support
(2) Mid-Polity & High Civ. Support
Probability of Event