Cultural differences between peacekeepers and local populations: A catalyst for violence?

Cultural Differences between Peacekeepers and Local

Populations: A Catalyst for Violence?

Master’s Thesis: Armed Conflict Christopher Vollhardt S2290634 Thesis Supervisor: Dr. Roos van der Haer 9999 Words

Second Reader: Dr. Corinna Jentzsch 31/07/2019

1

A

BSTRACT

United Nations peacekeeping operations often face severe violence. As of yet, reasons for these attacks against peacekeepers, who are supposed to be impartial mediators in a (post-)conflict setting, are not thoroughly understood. Recently, the ethnocultural composition of peacekeeping missions has caught the attention of scholars, who found that it can significantly influence the effectiveness of a mission. However, it remains unclear how exactly it shapes the relationship with the host population and whether it increases or decreases local support. This study fills this gap, by focusing on the ethnocultural composition of a peacekeeping mission and its effects on violence against peacekeepers. The cultural distance between peacekeepers, who often come from a range of different countries, and the local population could lead to the peacekeepers misunderstanding certain cultural dynamics in the country of operations. This could create resentment of the local population against the peacekeeping mission and lead to violence between the two parties. I use two different indices, measuring linguistic and religious distance between peacekeepers and local inhabitants and test their influence on the monthly number of violent incidents against peacekeeping missions in Sub-Saharan Africa from 1990 – 2009. After controlling for conflict-specific characteristics through the means of a fixed-effects model, a significant relationship between the cultural distance and the number of attacks cannot be identified.

2

I

NTRODUCTION

In the 1990s, a range of humanitarian failures fundamentally changed the face of the United Nations’ (UN) most ubiquitous instrument – peacekeeping. The end of the Cold War allowed the UN to significantly increase the scope of peacekeeping by mandating it with a plethora of new, complex tasks, such as peace enforcement and humanitarian aid (Bellamy et al. 2004). Initially, peacekeepers struggled with these new mandates, which not only led to humanitarian catastrophes in Bosnia, Somalia, and Rwanda, but also to a steep increase in violence against the blue helmets themselves (Ibid.). The number of peacekeepers killed during operations grew considerably compared to more traditional missions, in which the monitoring of a ceasefire was the only objective (Fjelde et al. 2016).

At the same time, Western (developed) countries drastically reduced their troop contributions to UN operations (DPKO 2019). In 1990, the countries with the largest peacekeeping contingents were Canada, Austria, Finland and Norway (Ibid.). Today, they mostly entrust developing countries with the provision of troops, which are able to benefit financially and politically from their participation in peacekeeping missions (Beswick 2010; Victor 2010). Ethiopia, Rwanda and Bangladesh are now the top 3 contributors to peacekeeping, while developed countries mostly limit their support to training and financial backing (Ibid.). Many scholars (Seet/Burnham 2000; Victor 2010; Bove/Elia 2011; Gelot et al. 2012; Van der Lijn/Dundon 2014; Fjelde et al. 2016) argue that the increased risk of suffering casualties is a crucial reason for the reluctance of Western leaders to contribute more soldiers to peacekeeping. In Rwanda, for example, the death of Belgian peacekeepers led to the Belgian military completely withdrawing their troops from the mission, as well as no longer providing military peacekeepers to further UN missions (Dallaire 2003; DPKO 2019). Victor (2010) explains how the US government had to refrain from peacekeeping, because it was not able to justify its participation to the public anymore after the deaths of many US soldiers in Somalia during the

3 battle of Mogadishu in 1993. The unprecedented violence that the UN Operation in Somalia II (UNOSOM II) had to face, compared with previous missions, made many Western states less willing to provide peacekeepers (Gelot et al. 2012).

In Africa, where the majority of peacekeeping missions take place, Western governments now often rely on African states to lead peacekeeping initiatives (Beswick 2010; Victor 2010). Many Western leaders have supported the African Union’s (AU) principle to find “African solutions to African problems” (Beswick 2010; Van der Lijn/Dundon 2014). This rubric implies that African forces are generally better equipped to deal with conflicts on their own continent, as they would be perceived to have a more legitimate mandate for intervention from the local population (Gelot et al. 2012). In general, it suggests that a mission comprised of peacekeepers, who share an ethnic or cultural background with the people in the deployed countries, should enjoy greater support from the local population. As of now, this thesis has not been academically tested and therefore lacks an empirical basis.

Although a number of anthropological studies (e.g. Duffey 2000; Rubinstein 2005, 2008; Tomforde 2010) examine the effects of cultural misperceptions within peacekeeping missions, it remains unclear how the cultural backgrounds of peacekeepers themselves influences their interactions with the local population. There seems to be a consensus that a comprehensive understanding of the country’s cultural and societal dynamics, as well as a good relationship with the local population are important for successful peacekeeping (Tomforde 2010; Newby 2018). However, none of these studies focus on the influence of the ethnocultural composition of the peacekeepers.

Bove and Ruggeri (2019) recently addressed this gap by analyzing whether the cultural distance between peacekeepers and local inhabitants, measured in terms of religious and linguistic distance, has an effect on the mission’s ability to reduce battlefield deaths. Echoing “African solutions to African problems”, the authors argue that shared cultural characteristics strengthen

4 the trust between peacekeepers and local citizens, which ensures local support for the mission and in turn their ability to work effectively (Ibid.). Focusing on peacekeeping missions between 1990 and 2015, they prove that a low cultural distance significantly reduces the number of battlefield deaths (Ibid.). However, it remains unexplained how exactly this improved relationship manifests on the ground and what could lead to this increased effectiveness. One of the consequences of greater local support could be that the peacekeeping mission do not get attacked as often. If the concept of “African solutions to African problems” is true, in the sense that culturally similar peacekeepers employ a more trustful and cooperative relationship with the local population, this should mean that violent incidents against peacekeepers should be greatly reduced. This study attempts to explore this hypothesis. It gives new insights into whether the UN should support this notion and request culturally similar troops for peacekeeping missions.

In the following paper, I examine the correlation between ethnocultural factors and the safety of peacekeeping forces. I conduct a large-N study that researches the effect of cultural distance between peacekeepers and local populations on the number of attacks against the UN forces. I ask: Does the cultural distance between the peacekeeping force and the local population affect the likelihood of peacekeepers becoming victims of violence? I begin by reviewing the contemporary literature on peacekeeping in general, as well as current findings on the effects of the composition of the peacekeeping force, and violence against them. Next, I look at theoretical explanations of violence against peacekeepers and how cultural distance could influence the risk of violence. If not otherwise stated, I use the terms “peacekeepers”, “peacekeeping force” or “peacekeeping personnel” interchangeably to describe all individuals participating in a UN peacekeeping operation, i.e. soldiers, police and civilians. I hypothesize that a higher cultural distance could severely increase the room for cultural misunderstandings and could lead to conflicts and a higher risk of violence against the peacekeeping force.

5 Afterwards, I describe the chosen method of this study. In the last section, I present and discuss the empirical results.

L

ITERATURE

R

EVIEW

One of the main strands of peacekeeping literature has evolved around the question of whether and under which conditions might peacekeeping be effective in keeping the peace. Early studies (e.g. Burns/Heathcote 1963; Mackinlay 1989) evaluate the effectiveness of peacekeeping mainly on the basis of the goals set within their official Security Council mandates. After the end of the Cold War and the transformation of peacekeeping towards multidimensional operations, the evaluation criteria used by researchers became more sophisticated. Many studies (Doyle/Sambanis 2000; Fortna 2004; Gilligan/Sergenti 2008) define the effectiveness of peacekeeping by looking at the likelihood that a conflict recurs after a peacekeeping mission took place. They find, among other things, that the type of conflict, its severity, and the type of peacekeeping mission (traditional, peace enforcement or multidimensional) play an important role in the probability of success (Doyle/Sambanis 2000).

Recently, Hultman, Kathman, and Shannon (2013, 2014) have redefined the effectiveness of peacekeeping by focusing on the number of battle-related and civilian deaths. They argue that the rise of peacekeeping deployments to active conflicts constituted a shift away from the traditional objective of ceasefire monitoring, towards a greater emphasis on reduction of violence within a still ongoing conflict (Hultman et al. 2014). They find that the number of deployed peacekeepers has a significant influence on the effectiveness of a mission to prevent one-sided violence against civilians and battlefield deaths (Hultman et al. 2013, 2014). Apart from that, Hultman, Kathman, and Shannon (2013, 2014) examine the specific composition of the peacekeeping force. They point out that military peacekeepers are significantly increasing

6 the effectiveness of peacekeeping operations, while a higher number of UN observers can, on a short-term, even decrease its effectiveness. (Hultman et al. 2013, 2014).

Other scholars (Bove/Ruggeri 2015, 2019; Haass/Ansorg 2018) further research the effects of the internal composition of a peacekeeping mission. Bove and Ruggeri (2015), for example, find that the level of diversity within the peacekeeping force significantly reduces the number of civilian casualties during the deployment of the mission. They argue that within one mission, peacekeeping contingents from different countries are able to complement each other through different skills and areas of expertise (Ibid.). Haass and Ansorg (2018), on the other hand, focus on the differences in training and technical quality of peacekeeping contingents. They identify a significant positive relationship between the number of well-trained and well-equipped peacekeepers and the mission’s effectiveness in protecting civilians (Haass/Ansorg 2018).

Bove and Ruggeri (2019), instead, shift their focus on the ethnocultural composition of a peacekeeping mission in relation to the local population. They examine the influence of cultural and social distance of peacekeepers towards the local population on their effectiveness in reducing battle-related and civilian deaths (Bove/Ruggeri 2019). They argue on the one hand, that the more peacekeepers within a mission come from culturally proximate countries, the more effective they are, since it improves the relationship with the local population and their knowledge of local conflict dynamics (Ibid.). This should lead to better communication with the local inhabitants and higher support for the peacekeeping mission (Ibid.). On the contrary, a higher social distance, in terms of economic and institutional backgrounds, leads to the peacekeepers being perceived as impartial by locals, which can also enhance the effectiveness of a mission (Ibid.) Bove and Ruggeri (2019) argue that peacekeepers from socially distant countries, with a different economic and institutional structure, should more likely be perceived as unbiased towards the conflict parties and therefore enjoy more support from the local population. Bove and Ruggeri (2019) find significant evidence for both these hypotheses.

7 While all the studies on the composition of peacekeeping forces focus on the effectiveness to reduce battle-related and civilian deaths, none analyze the effects on the relationship with the local population itself. Bove and Ruggeri (2019) are able to prove an increased effectiveness of the peacekeeping mission through a low cultural distance of peacekeepers towards the local population, however they are not able to identify the causes that lead to this relation. While they claim that the increased effectiveness stems from the “shared norms, practices and languages” of culturally proximate peacekeepers, their study gives no insight into how exactly the quality of the relationship between the peacekeepers and the local citizens is improved by an increased number of shared cultural characteristics (Bove/Ruggeri 2019: 1651). In this study, I address this academic gap by focusing on a different outcome variable, which has not yet been related to the ethnocultural composition of the peacekeeping operation. A lower number of attacks against the peacekeepers by local perpetrators should be a sign of a stronger trust and greater support by the population. Therefore, I examine whether the low cultural distance between peacekeepers and the local population leads to decreased violence against the peacekeeping force.

Due to the greater availability of data, the study on violence against peacekeeping forces has started to receive a considerable amount of attention in the last years (Fjelde et al. 2016). Some scholars (Seet/Burnham 2000; Henke 2016; Lindberg Bromley 2017) consider the assessment of the danger that peacekeepers face as decisive in their ability to fulfill their designated task. Seet and Burnham (2000) analyze fatality data and identify a rise in the number of killed peacekeepers since the end of the Cold War. They argue that this is a result of the increased number and scale of peacekeeping missions, particularly towards the development of more peace enforcement (Seet/Burnham 2000). Nevertheless, in regard to the last decades, Henke (2016) finds that the ratio of peacekeepers dying in the field has not significantly risen. Both studies, however, are limited by the use of crude data and do not provide any explanation for the occurrence of violence against peacekeepers.

8 Other scholars (Salverda 2013; Fjelde et al. 2016) have addressed this gap, by focusing on strategic considerations of rebel groups to use violence against the personnel of a peacekeeping operation. Salverda (2013), for instance, argues that powerful rebel groups have an incentive to attack peacekeepers. A rebel force, which is significantly stronger than the government army, can realistically expect a military victory and is therefore interested in a continuation of the conflict (Salverda 2013). A peacekeeping mission however, that is tasked with stabilizing the status quo, separating the conflict parties, and often with finding political solutions to the conflict represents a barrier to such a rebel group (Ibid.). Despite finding statistically significant support for this hypothesis, her study is limited by the exclusion of peace enforcement missions authorized under Chapter VII of the UN Charter, who typically face the highest rates of violence (Seet/Burnham 2000; Salverda 2013). Moreover, her unit of analysis “mission-year” cannot account for short-term variation in the number of attacks against peacekeepers within a given year (Salverda 2013; Fjelde et al. 2016).

Fjelde, Hultman, and Lindberg Bromley (2016) use a much more disaggregated and precise set of data; the Uppsala Conflict Data Program’s (UCDP) Peacekeepers at Risk (PAR) dataset. It provides information on specific incidents of violence against peacekeepers, such as the exact date, location, and perpetrator (Fjelde et al. 2016). They argue that rebel groups that suffer from high military setbacks in a given month are more likely to attack the peacekeeping force because they see the peacekeepers as stabilizing an unfavorable status quo (Ibid.). Furthermore, the authors hypothesize that their military setbacks lead to these groups radicalizing and therefore becoming more dangerous for the peacekeepers (Ibid.). This significantly heightens the risk of peacekeepers becoming victims of attacks (Ibid.). Apart from that, they find that rebels being killed in battles with the UN forces intensifies the antipathy of rebels towards the peacekeepers, which also increases the probability of peacekeepers getting attacked (Ibid.).

9 With only two systematic studies on the causes of violence against peacekeepers, this field of study is still very new and not thoroughly researched. Considering the large number of different rebel groups and other actors in many civil wars, incentives to attack UN peacekeepers might be more multifaceted than currently known. In the PAR dataset, for example, over half of the perpetrators of violence against peacekeepers are labeled as “unknown” or simply “civilians” or “mobs” (Lindberg Bromley 2018). In the case of Somalia, Sahnoun (1994: 29) speaks of “marauding armed groups, loyal to no particular warlord, but only to themselves”, as a severe threat to the safety of the peacekeeping personnel. This study therefore, contributes to the current body of research, by looking at another factor that may be correlated with violence against peacekeepers – the ethnocultural composition of the peacekeeping force.

T

HEORY

Within the field of anthropology, culture is not only understood as a set of customs and traditions, but, first and foremost, as shared common values, that constitute what a person judges as appropriate behavior (Rubinstein 2005; Hofstede et al. 2010). Hofstede, Hofstede, and Minkov (2010) call this the “software of the mind”, which is heavily influenced by the social environment in which a person grows up. People from the same cultural background share the same patterns of feeling, thinking and acting, which visibly manifests in different cultural rituals and every-day behavior (Hofstede et al. 2010). Intercultural encounters can, however, lead to unintended conflicts. Foreigners in an unfamiliar environment are often not able to recognize the underlying values or belief systems of a local society, which can lead to misunderstandings (Ibid.). Between different populations, cultural traits can vastly differ and “introduce barriers to interaction and communication between societies” (Spolaore/Wacziarg 2016: 174).

10 A peacekeeping mission can be seen as an example of an intercultural encounter. According to Rubinstein (2005: 528), peacekeeping missions bring together a range of actors from different parts of the world and are consequently “fraught with potential for miscommunication and conflict”. Since they come from different cultural contexts, peacekeepers and locals cannot always rely on the same understanding of appropriate behavior, which can lead to problems (Tomforde 2010). Without a sophisticated knowledge of local cultural patterns of behavior, the risk of peacekeepers reacting inappropriately is vastly increased (Rubinstein 2008). Hence, cultural awareness plays an important role in the common understanding between peacekeepers and locals (Rubinstein 2005; Tomforde 2010). According to Haddad (2010), it is central for a peacekeeping mission to understand the local cultural dynamics. In his study on French peacekeepers in the UN Interim Force in Lebanon (UNIFIL), he finds that while cultural awareness training greatly improved the knowledge of peacekeepers of the local population’s cultural traditions and boundaries, contact with local people was still often disagreeable or even hostile (Haddad 2010). He proclaims that these trainings, used as preparation for the cultural preconditions in the country of operations, can improve the understanding of peacekeepers, but certain other cultural subtleties simply cannot be taught (Ibid.).

Apart from personal interactions between peacekeepers and local citizens, cultural differences can also play a major role on a broader, more conceptual level. Misperceptions of local societal and cultural dynamics and boundaries by peacekeeping officials can influence the overall quality of the mission negatively (Duffey 2000). Talentino (2007) points out that it is pivotal for international peacekeeping and peacebuilding missions to be able to understand and adequately react to the local ethnic and religious structure. If for example, peacekeepers overlook the social status of one particular group, and do not properly include it in the peace process, this can lead to a sense of imposition and spark resentment against the international effort (Talentino 2007). This was a major problem in Somalia, where UNOSOM did not

11 understand the decentralized clan system of the country, thus alienating some clan elders by leaving them out of the peace process (Duffey 2000; Rubinstein 2008).

Autesserre (2014a) draws attention to the issue of different conceptions of conflict-management between peacekeepers and local actors. To prevent resentment aimed at the international mission, peacekeepers must understand and properly account for local conflict resolution mechanisms (Rubinstein 2003). In the case of peacekeeping in East Timor, Rubinstein (2008) illustrates how the creation of Western liberal institutions led to confusion within the population, where understandings of political authority were completely different. Furthermore, in Somalia, traditional Somali peacemaking means have been neglected by UNOSOM, in favor of hierarchical political structures and large peace conferences (Duffey 2000). In these cases, peacekeepers that were more familiar with the local preconditions in these countries could have been more successful in establishing a constructive relationship with the local population. A good, cooperative relationship with the local population is decisive for a peacekeeping operation’s success and effectiveness (Rubinstein 2008; Newby 2018; Bove/Ruggeri 2019). A poor relationship, however, stemming from cultural misunderstandings and misperceptions by peacekeepers, can decrease local support and ultimately disrupt the mission (Rubinstein 2008; Newby 2018). A lack of local consent and support for the mission can even increase the number of attacks targeted at the peacekeeping mission; a fact that has also been acknowledged by the UN (DPKO 1999; Newby 2018). This can make it easier for local rebel leaders or warlords, who oppose the peacekeeping operation, to recruit civilians for their cause. UNOSOM’s failure to integrate a number of smaller clans in the peace process allowed local warlord Mohamed Farah Aideed, who was responsible for most of the attacks on the peacekeeping mission, to convince the Somalis to join his fight (Hirsch/Oakley 1995). On the basis of these theoretical considerations, a peacekeeping force, with a lower cultural distance towards the local population, should be able to better account for local cultural dynamics. It should be more likely

12 to foster a cooperative relationship with local citizens, which in turn could lead to a lesser likelihood of violence against the peacekeepers.

When analyzing cultural distance, two factors are especially important: language and religion. Both have often been used by scholars (Alesina et al. 2003; Fearon 2003) in the past, as categorizations for different ethnic and cultural groups. Fearon (2003: 201) calls them “distinguishing cultural features”, which are often inherent to a large majority of members within one ethnocultural group.

Firstly, language plays a major role in the construction of cultural identities and is often seen as a reflection of cultural differences (Davies/Dubinsky 2018). Apart from language being one of the most important commonly shared characteristics of one ethnocultural group, it is vital for a constructive understanding between peacekeepers and local populations. Their relationship depends heavily on the peacekeepers’ ability to communicate successfully with the local population and the understanding and respect of peacekeepers for their local values and world views (Rubinstein et al. 2008). As a result, speaking the same language should improve the communication with local people and should give both sides the opportunity to express their problems and opinions directly.

Without speaking the local language, a person in a foreign environment remains an outsider and will not be fully capable of comprehending certain cultural subtleties (Hofstede et al. 2010). Even the reliance on interpreters is not safe from failure since Autesserre (2014b) identifies multiple translation mistakes by interpreters during the UN Organisation Mission in the Democratic Republic of Congo (MONUC). Another particularly atrocious example of language barriers leading to massive resentment towards peacekeepers, was a gang rape committed by a local militia in the Democratic Republic of Congo (Autesserre 2014b). According to an internal UN inquiry, the incident was overlooked by a patrol of Indian peacekeepers, who failed to comprehend the situation because of their inability to speak the language and understand

13 warnings by local citizens (Ibid.). Hence, language can be a difficult barrier between peacekeepers and the local population (Bove/Ruggeri 2019). This leads to my first hypothesis: Hypothesis 1: The higher the linguistic distance between peacekeepers and the local population, the higher the number of attacks against the peacekeeping operation.

Apart from language, religion plays a central role in most ethnocultural groups and is a “salient characteristic of human societies” (Spolaore/Wacziarg 2016: 185). Peacefully interacting with people from a foreign society entails accounting for religious preferences of different religious groups. (Rubinstein et al. 2008). Peacekeeping soldiers, who share religious beliefs with local people, can often use their specific cultural intelligence and religious knowledge to foster a more trustful and empathetic connection with the locals (Bosman et al. 2008). The reactions towards peacekeepers from other religious backgrounds, however, can be more hostile. Some Somalians, for example, even perceived UNOSOM as an effort to convert the Muslim population to Christianity (Rubinstein 2008).

In general, peacekeepers’ disregard of religious boundaries of local people can lead to tensions. Bosman, Soeters, and Bari (2008: 695f.) give an example of Dutch soldiers in Kosovo and Afghanistan, who offended local citizens by “sun-bathing half-naked, doing sports in shorts, and casually interacting with the other sex”. In contrast, Turkish soldiers, who mostly shared a Muslim background with the local population, were much more careful and empathetic, allowing them to foster a much closer relationship with the local population (Bosman et al. 2008). During the UN Emergency Force (UNEF), peacekeepers even started firing at a minaret, since they misunderstood the call to prayer as a call for civil disorder (Rubinstein 2008). Such episodes of cultural insensitivity can lead to strong animosity amongst the local population towards the peacekeeping mission. Peacekeeping forces with a similar religious background might thus possess more cultural and religious knowledge, which should improve the

14 relationship with the local population drastically. This should further reduce the number of attacks on the peacekeeping force. Therefore, my second hypothesis is as follows:

Hypothesis 2: The higher the religious distance between peacekeepers and the local population, the higher the number of attacks against the peacekeeping operation.

M

ETHOD

I will test these hypotheses, by using a count variable of violent incidents against peacekeeping forces in UN peacekeeping missions between 1990 and 2009 in Sub-Saharan Africa, taken from the UCDP PAR dataset1. My unit of analysis will be mission-month. The PAR dataset counts observations in which one or more peacekeepers were involved in violence “at a specific location and for a specific temporal duration” (Lindberg Bromley/Greek 2016: 5). The regional focus on peacekeeping operations in Sub-Saharan Africa stems mainly from the fact that these missions are experiencing a far higher rate of violence against their forces, compared with other UN operations (Lindberg Bromley 2018). Hence, recruiting (particularly Western) troops for missions in this region represents a difficult challenge for the UN and is the basis for the concept of “African solutions to African problems”, which proclaims the greater ability for culturally proximate peacekeepers to deal with conflicts in Africa (Beswick 2010; Van der Lijn/Dundon 2014). The sample however, still includes the vast majority of UN missions, since Sub-Saharan Africa is the region with the most deployed operations (DPKO 2018).

Furthermore, I focus on traditional and multidimensional UN peacekeeping missions. Missions by other international organizations, such as the AU or the European Union, and missions, such as the French-led operations in Rwanda and Ivory Coast or the US-led Unified Task Force in

1 _{Unfortunately, the PAR dataset is not complete. It misses data on notable UN peacekeeping missions: UNMEE}

15 Somalia (UNITAF) are excluded. My analysis is based on the study by Bove and Ruggeri (2019), who show that a lower cultural distance between peacekeepers and local populations within UN missions in Sub-Saharan Africa leads to higher effectiveness. Since I use their indices for cultural distance to measure whether it influences the monthly number of attacks against peacekeepers, I apply the same regional and temporal focus. Nonetheless, the theory should also hold for peacekeeping missions, other than UN-led ones.

16 Dependent variable

The dependent variable, Violence against peacekeepers, records the number of attacks on peacekeeping forces per month. To be recorded as an “attack” at least one peacekeeper needed to be the victim of an act of violence, committed by an individual or a group of individuals from the host country. In 123 of 664 months at least one violent incident happened, with 13 being the highest number of attacks on peacekeepers during one month. A peacekeeper is someone working directly under the mandate of a particular UN mission and can either be a soldier or a civilian. The perpetrator of violence can be a rebel group or government forces, but also an unidentified group of civilians. The observed act of violence did not have to result in the death of a peacekeeper, but could also have led to injuries, kidnappings or forced detainments. Through its monthly scope, the PAR dataset can account for short-term changes in the troop composition of a peacekeeping mission. This is particularly important when looking at the cultural distance of the force towards the local population, which can change from month to month, with every major new contingent of forces leaving or being deployed to the mission (Hultman et al. 2014).

Independent Variables

The independent variables measuring the cultural distance between the peacekeeping forces and the local population are language and religion. I rely on data by Bove and Ruggeri (2019), who constructed a measure of cultural distance, which is weighted on the basis of the number of peacekeeping forces contributed in each month by each country to one particular UN peacekeeping mission. The first independent variable, Linguistic distance, measures the linguistic distance between the peacekeeping force in general and the local population. Bove and Ruggeri (2019) use an index created by Spolaore and Wacziarg (2016) as the basis. It measures the linguistic distance between every country with every other country in the world,

17 using language families, while taking into account the different languages of subpopulations (Spolaore/Wacziarg 2016). Linguistic distance ranges between values 0 and 1, with value 0 representing a peacekeeping force that has the exact same linguistic background as the population of the country hosting the peacekeeping operation and value 1 being the furthest possible linguistic distance between the peacekeepers and the local population (Bove/Ruggeri 2019).

The second independent variable, Religious distance, measures the religious distance between the peacekeeping force in general and the local population. Again, the basis for this measure is an index created by Spolaore and Wacziarg (2016), which measures the religious distance between every country in the world, relying on so-called religious trees. These trees are formed by grouping religions into different levels of categories, for example “Near Eastern Monotheistic Religions”, which includes Christianity, Islam, and Judaism (Spolaore/Wacziarg 2016). “Christianity” itself includes Western Catholicism and Eastern Orthodoxy, which are both again further disaggregated into smaller groups (Ibid.) The more categories one religious group shares with another, the more closely related these two groups are (Ibid.) The calculation of the index also takes religious diversity within a country into account, by including all major religious groups that exist in it (Ibid.). Bove and Ruggeri (2019) have weighed the religious distance between each contributor country to a peacekeeping mission and the host country, again based on the contributed number of personnel. The value of the variable ranges between 0 and 1, with value 0 representing a peacekeeping force, which has the exact same religious background as the population in the country hosting the peacekeeping mission and value 1 being the furthest possible religious distance between the peacekeepers and the local population. (Bove/Ruggeri 2019). Since the effects of the cultural distance towards the local population on the number of attacks should not immediately occur at the peacekeepers’ time of deployment, I apply a one-month lag to both independent variables (Ibid.).

18 Control variables

To account for other factors influencing the likelihood of violence against a peacekeeping force, I employ a range of control variables. First, I control for the size of the peacekeeping mission, because a larger mission increases the number of potential targets for violence, which should raise the possibility for a peacekeeper of getting attacked. To account for this, I use the variable Troop Size, which counts the number of peacekeeping personnel within one mission at a given month in thousands (Kathman 2013). Furthermore, to account for the effects of diversity and heterogeneity within the peacekeeping mission itself, the variable Number of participants is included, counting the number of countries participating in the peacekeeping mission (Bove/Ruggeri 2019). Also, conflict intensity should significantly increase the risk of peacekeeping forces becoming the victims of violent incidents (Fjelde et al. 2016). To control for this, I use Battle-related deaths, which counts the number of battle-related deaths at a given month in thousands (Sundberg/Melander 2013).

The specific mandate of the mission is another aspect, which could affect the likelihood of violence against peacekeepers (Fjelde et al. 2016). The dichotomous variable Force mandate controls for the effect that such a robust mandate has on the risk of violence against the mission’s personnel. Apart from that, Fjelde, Hultman, and Lindberg Bromley (2016) found that the number of rebels killed at the hand of peacekeeping forces increases the probability of violence against peacekeepers. To control for this, I include Rebel losses from peacekeepers, which counts the number of rebels killed by peacekeeping soldiers. Furthermore, I control for the findings by Salverda (2013), by including Rebel strength, which measures the relative strength of the main rebel group, compared with the government forces. Lastly, Fjelde, Hultman, and Lindberg Bromley (2016) uncover a significant relation between the time of the last attack against peacekeepers and the probability of another attack. Therefore, I control for the decay function Time since past attack. Similarly to the independent variables, the

19 peacekeeping troop size, as well as the battle-related deaths within a conflict and the rebel losses should affect the dependent variable only after a certain amount of time (Fjelde et al. 2016). Therefore, I apply a one-month lag to Troop size, Battle-related deaths and Rebel losses from peacekeepers. After calculating the variance inflation factors, I found that there is no issue of multicollinearity between the independent variables.

Table 1. Descriptive statistics

Variable Mean Std. dev. Variance Min Max

Violence against peacekeepers 0.33 1.067 1.139 0 13 Linguistic Distance (t-1) 0.97 0.035 0.001 0.69 1 Religious Distance (t-1) 0.82 0.076 0.006 0.39 0.99 Troop size (t-1) 8.29 7.043 49.602 0.002 29.54 Number of participants 18.14 12.398 153.699 1 55 Battle-related deaths (t-1) 0.87 13.329 177.675 0 301.28 Force mandate 0.89 0.313 0.098 0 1

Rebel losses from peacekeepers (t-1)

0.67 3.677 13.523 0 45

20

R

ESULTS

Since the dependent variable is a count variable of events, I use a count model over a standard linear regression (Hilbe 2014). Furthermore, since its variance is far higher than its mean, which indicates overdispersion, I conduct a negative binominal regression, instead of a Poisson regression (Ibid.). Table 2 reports the results. Model 1 is a base model, without the inclusion of control variables. It shows that contrary to my hypotheses, Linguistic distance and Religious distance are negatively correlated with the dependent variable Violence against peacekeepers.

Model 2 is the full model, including all control variables. The results also clearly contradict my hypotheses. With a coefficient of -3.116, Linguistic distance is negatively correlated with the dependent variable. However, the result is not statistically significant. It suggests that the linguistic distance between the peacekeepers and the local population does not affect the number of attacks against the UN forces. Hence, hypothesis 1 is rejected. The second result is even more surprising. With a coefficient of -3.742, Religious distance is also negatively correlated with the number of violent attacks against the peacekeeping force, but in contrast to Linguistic distance, the result is highly significant. This indicates that a higher religious distance between peacekeepers and local inhabitants might in fact, lead to a lower probability of attacks against the peacekeeping mission. Accordingly, hypothesis 2 is rejected as well.

Troop size is slightly positive and significant, so the number of peacekeepers deployed to the area of operations increases the probability of an attack happening against them. However, the number of Battle-related deaths within a conflict area and the existence of a Force mandate are not significantly correlated with the number of attacks against the peacekeeping force. Both of these results contradict the findings by Fjelde, Hultman, and Lindberg Bromley (2016), who found a significant relation between the variables Battle-related deaths and Force mandate and the number of attacks on a peacekeeping force. This might be due to their different data, which also includes non-UN peacekeeping missions (Fjelde et al. 2016). It seems that the number of

21 battlefield deaths within a conflict and an explicit mandate to use force play a much bigger role in the number of attacks against peacekeepers when the mission is not conducted by the UN, but by another international organization or one single nation.

The Number of participants is also not significantly related to the dependent variable. While Bove and Ruggeri (2019) find that a higher number of countries participating in a mission can significantly increase the effectiveness, the same cannot be said about the probability of an attack happening against the peacekeepers. While the overall organization and conception of the peacekeeping operation might benefit from the skills and knowledge brought by a more diverse force, the security of peacekeepers does not seem to be affected. One explanation for this could be that the provision of safety for the peacekeeping personnel is a much more straightforward task, compared with the overall conception of a complex peacekeeping mission and might therefore not benefit as much from greater diversity. Rebel losses from peacekeepers is, with a coefficient of 0.037, significantly correlated with Violence against peacekeepers. This supports the findings of Fjelde, Hultman and Lindberg Bromley (2016), which indicate that the number of rebel soldiers killed by peacekeepers can increase the probability of an attack. Moreover, Rebel strength does not significantly affect the number of attacks against the peacekeepers, which contradicts the findings of Salverda (2013). This probably stems from the inflexibility of her measure of rebel strength relative to the government, which mostly stays the same within a certain conflict and therefore is not able to account for conflict variations over a shorter amount of time (Salverda 2013; Fjelde et al. 2016). Finally, the decay function Time since past attack proposed by Fjelde, Hultman and Lindberg Bromley (2016) is positive and highly significant, with a coefficient of 1.926. Hence, the more time passes after the last attack, the less likely another attack against the peacekeeping force.

22

Table 2. Peacekeepers’ cultural distance towards local population and number of attacks against peacekeepers

Standard errors in parentheses. *p < 0.1; **p < 0.05; ***p < 0.01 Variables Model 1: No Controls Model 2: Violent attacks on PK Model 3: French missions Model 4: Nonlethal attacks on PK Model 5: Lethal attacks on PK Model 6: Exclusion Somalia Model 7: Fixed Effects Linguistic Distance (t-1) -7.584*** (2.219) -3.116 (2.569) -2.504 (2.421) -3.665 (3.644) -2.74 (2.719) -2.606 (2.806) -4.383 (3.043) Religious Distance (t-1) -6.081*** (0.923) -3.742*** (1.065) -3.961*** (1.003) -4.193** (1.693) -3.839*** (1.119) 0.93 (2.034) -0.757 (2.081) Troop size (t-1) 0.049*** (0.0165) 0.057** (0.023) 0.055*** (0.018) 0.032 (0.021) 0.017 (0.021) Number of participants 0.004 (0.009) 0.014** (0.007) 0.012 (0.013) -0.003 (0.01) 0.013 (0.01) 0.025* (0.014) Battle-related deaths (t-1) -0.001 (0.006) 0.005 (0.006) -0.009 (0.015) 0.000 (0.006) 0.000 (0.006) Force mandate -0.07 (0.407) 0.266 (0.386) -0.432 (0.626) 0.115 (0.47) -0.098 (0.414) -0.321 (0.596)

Rebel losses from peacekeepers (t-1) 0.037** (0.016) 0.046*** (0.014) 0.046** (0.018) 0.036** (0.016) 0.039** (0.018) 0.029* (0.016) Rebel strength 0.196 (0.133) 0.082 (0.121) 0.166 (0.225) 0.211 (0.142) 0.133 (0.143) -0.074 (0.31)

Time since past attack 1.926***

(0.273) 2.109*** (0.25) 2.150*** (0.443) 1.988*** (0.298) 1.731*** (0.301) 1.522*** (0.293) Constant 11.135*** (2.369) 2.865 (2.857) 2.509 (2.685) 2.678 (4.116) 2.236 (2.997) -1.397 (3.346) 0.938 (3.666) Observations 664 664 755 664 664 637 664 Chi² 53.087*** 216.067*** 219.085*** 137.275*** 193.363*** 96.212*** 240.68*** Log Likelihood -469.459 -387.969 -428.682 -200.645 -337.628 -328.558 -375.663 AIC 944.918 795.938 873.557 421.29 695.257 677.116 791.325

23 In order to test the robustness of the results, a number of different regression models are run. Model 3 includes peace missions by the French armed forces in Sub-Saharan Africa between 1990 and 2009, namely Operation Turquoise in Rwanda and Operation Licorne in Ivory Coast. The colonial past of France in Africa, especially in Ivory Coast, could have a significant negative effect on the local perception of a French military mission in the country. This could increase the probability of violence against such a mission. Due to the lack of data however, it is not possible to account for soldiers from the French foreign legion. This will not distort the result too much however, since the foreign legion only constitutes a very small part of the French armed forces (Foreign Legion Info 2018). The results in Model 3 do not notably differ from Model 2. The value of Religious distance stays negative and significant with a slightly higher coefficient of -3.961. Linguistic distance remains insignificant. Hence, the cultural difference between peacekeepers and the local population in French peacekeeping missions does not lead to a different number of attacks against the peacekeepers. Notably, the missions conducted by the French armed forces only constitute a small number of non-UN peacekeeping forces. The effects of cultural distance within other non-UN peacekeeping missions thus need to be explored further.

Nonetheless, Model 3 uncovers a distinct weakness of the linguistic distance index by Spolaore and Wacziarg (2016), in regard to my research design. Since their index is designed to measure the cultural distance between two ethnocultural groups on the basis of linguistic differences, it only takes first languages into account, but does not include the possibility of any second languages people might speak (Spolaore/Wacziarg 2016). In this study, the index obscures the possibility of peacekeepers and local inhabitants communicating in another language. Especially in Sub-Saharan African countries, a majority of the population is able to speak English or French, despite these not being the languages they might use to communicate with their family, neighbors or other compatriots.

24 In the data for Model 3, the value of Linguistic distance between French peacekeepers and the local populations of Rwanda and Ivory Coast is in both cases 1, which is the furthest linguistic distance observable. Since the first language in both African countries is not French, the linguistic distance between people from France and people from Rwanda or Ivory Coast is very high (Eberhard et al. 2019). Still, a high percentage of the population in Rwanda and Ivory Coast speak French as a second language (Ibid.). Furthermore, it is an official language in both countries (Ibid.). Hence, communication issues between the French soldiers and the local inhabitants were unlikely a problem, since they mostly would have been able to communicate in French. The same can be said about many other peacekeeping missions.

Although Autesserre (2014b) identified language issues as a significant problem in peacekeeping missions, it is doubtful that they are relevant when it comes to the mere prevention of misunderstandings and conflicts, where actors can often rely on their second languages to communicate. This could explain the insignificant relation between Linguistic distance and the number of violent attacks against peacekeeping forces. It seems that cultural misunderstandings that could potentially lead to violence can mostly be resolved through communicating in a second language. Nonetheless, it must be concluded that a high linguistic distance between peacekeepers and local inhabitants does not significantly affect the number of violent incidents.

In Models 4 and 5, the dependent variable is further disaggregated. Does the severity of the attack make a difference? Fjelde, Hultman, and Lindberg Bromley (2016) point out that the motivations for attacks on peacekeepers with a lethal outcome could be different from nonlethal ones. They argue that lethal attacks have a far higher “signaling value” and need a higher strategic and coordinated effort, while nonlethal attacks could be the result of simple loot-seeking (Fjelde et al. 2016: 618). With the help of the PAR dataset, two alternative dependent variables are constructed (Lindberg Bromley 2018). Model 4 takes non-lethal outcomes of

25 attacks against the peacekeeping force into account, while Model 5 focuses on lethal attacks. The results of both models are largely in accordance with the results on Violence against peacekeepers in general. The coefficient of Religious distance stays negative and significant, while Linguistic distance remains insignificant. Neither non-lethal attacks, nor lethal attacks are therefore related to the Linguistic distance. However, similarly to attacks in general, non-lethal and non-lethal attacks are significantly reduced by a higher Religious distance between peacekeepers and the local population. Since the coefficients for Religious distance of -4.193 and -3.839 are only marginally higher than the original result (-3.742), it does not seem to make a big difference whether the attacks have a lethal or nonlethal outcome.

The data has an important outlier case: Somalia. According to the PAR dataset, the peace enforcement mission UNOSOM II suffered from a much higher number of violent incidents compared to other peacekeeping missions (Lindberg Bromley 2018). Together, both UNOSOM I and II experienced 75 separate attacks on their forces, often with a high number of harmed or killed peacekeepers. The percentage of months, in which at least one attack happened is 74%, which is extremely high, compared to 16% during other peacekeeping missions. To test the robustness of the results, the case of Somalia is excluded in Model 6. The results show that, similar to the Linguistic distance, Religious distance also loses its significance. The fact that it loses its significance completely, suggests that the religious distance between peacekeepers and locals may have played an important role in Somalia, but does not significantly affect the overall number of attacks on peacekeeping forces.

When looking specifically at the data for UNOSOM I and II, it becomes evident, how this particular case has impacted my initial results. Before excluding the outlier case of Somalia, Religious distance consistently correlated negatively with the dependent variable Violence against peacekeepers, with the result being statistically significant. This result has been unexpected, since it suggests a completely opposite relation, compared to hypothesis 2. The

26 missions in Somalia had a great effect on this result, since the religious distance of the UNOSOM I and II peacekeepers towards the Somali population, was considerably lower compared to the other peacekeeping missions. While the overall mean of the variable Religious distance is 0.82, the average religious distance in the case of Somalia is only 0.64. Pakistan was the main troop contributor to UNOSOM I, while also being one of the major troop-contributing countries to UNOSOM II, alongside other countries with large Muslim populations, such as Egypt, Bangladesh, Malaysia and Morocco (Kathman 2013). These troop contributors were a major factor for the religious distance towards the mostly Muslim population of Somalia being very low. Nonetheless, the number of violent attacks on the peacekeeping forces in Somalia was substantially higher, compared with other peacekeeping missions. This led to the surprising result of Religious distance at first being negatively correlated with the number of attacks against peacekeepers.

The special case of Somalia shows that my research model seems to be biased by certain omitted variables, specific to each conflict and peacekeeping mission. Unobserved conditions that only occur in specific peacekeeping missions influence the number of attacks against the mission far more than the linguistic or religious distance of the peacekeepers towards the local population. This finding is further researched in Model 7, which replicates the full model, but this time with the inclusion of conflict fixed-effects, which account for conflict-specific bias. The results confirm the findings in Model 6. Again, Linguistic distance remains negative and insignificant. Similarly, Religious distance loses its significance, when controlling for fixed-effects. In accordance with Fjelde, Hultman, and Lindberg Bromley (2016), Rebel losses from peacekeepers and Time since past attack remain positive and significant. Even after controlling for fixed effects, the number of rebels killed by peacekeepers and the time since the last attack on the peacekeeping force significantly raise the Violence against peacekeepers. The independent variables, however, do not stay robust. It seems that other conflict-specific effects have a greater influence than the cultural distance between peacekeepers and local inhabitants.

27 Overall, the results of the statistical analysis show that the cultural distance between peacekeeping forces and the local population does not significantly affect the number of attacks against peacekeepers. This study does not produce any empirical evidence for the hypothesis, that a higher cultural proximity leads to greater support from the host population and therefore reduces violent attacks on the peacekeeping force. The data on Somalia especially shows that a low cultural distance of peacekeepers towards the local population does not necessarily prevent misperceptions. Cultural distance on its own, measured by linguistic and religious distance, is not a cause for eventual cultural misperceptions by the peacekeepers. The notion that culturally more similar peacekeepers automatically ensure a greater support by local inhabitants, as is suggested by the concept of “African solutions to African problems”, has to be rejected.

This, however, does not mean that cultural misperceptions and misunderstandings do not play any role for peacekeepers on the ground. A substantial understanding of the local societal and cultural dynamics in the country of operations can be indispensable, in ensuring a secure and effective work environment for UN peacekeeping personnel. The case of UNOSOM illustrates this. Although in terms of Religious distance, the peacekeeping force itself was quite similar to the Somali population, a multitude of scholars (Sahnoun 1994; Hirsch/Oakley 1995; Duffey 2000; Rubinstein 2008) have shown how cultural misperceptions by the UN forces have led to a disastrous outcome not only for the population of Somalia, but also for the blue helmets themselves. The Somali society and specific dynamics of the civil conflict have largely been misunderstood or even ignored by UN officials (Duffey 2000).

Indeed, Sahnoun (1994: 37) describes how his efforts to create peacemaking initiatives on a grassroots level in local communities have been “undermined” by the tendency of the UN to centralize the peacemaking process in Mogadishu. Negotiations have been designed in a top-down manner, incorporating only those leaders that have been considered the most powerful at the time, which illegitimately empowered clan leaders like Mohamed Farah Aideed and Ali

28 Mahdi, and marginalized others (Hirsch/Oakley 1995; Duffey 2000). To make matters worse, Aideed in particular resented the latter peace enforcement mission UNOSOM II, which he believed to be biased against him (Hirsch/Oakley 1995). This made his entire sub-clan hostile towards the peacekeepers and even led him to spur further resentment within the Somali population (Ibid.). These fundamental issues led to the catastrophic outcome of the UNOSOM II operation, not only for the population suffering from the intense fighting, but also for the peacekeepers themselves (Ibid.). Hence, a sophisticated understanding of local culture, society and belief systems is critical for a peacekeeping mission to work effectively and foster a trustful and cooperative relationship with the host population.

C

ONCLUSION

This study has researched the effect of cultural distance between peacekeeping soldiers and the local population on the number of attacks against the UN forces. Cultural distance was defined by two distinct features inherent to the culture of a society – language and religion. While Salverda (2013) and Fjelde, Hultman, and Lindberg Bromley (2016) focus on certain rebel group’s strategic reasons to attack the peacekeeping force, I have theorized that cultural misunderstandings, stemming from a high linguistic and religious distance between the UN forces and the locals, can be another reason for violence against the peacekeepers. However, the results do not support these hypotheses. The linguistic distance between peacekeepers and the local population is not significantly related to the number of attacks on the peacekeeping mission. The linguistic distance index created by Spolaore and Wacziarg (2016) is only measuring the distance between first languages spoken by the population in a particular country, without taking any second languages into account. Consequently, it is not able to account for the possibility of peacekeepers and locals communicating in another language. It seems that this more often than not allows the two parties to overcome their linguistic differences and still

29 foster a cooperative relationship. Initially, a negative and significant correlation between the religious distance of peacekeepers and local inhabitants and the number of attacks against the UN forces was found. After controlling for unobserved conflict-specific variables and excluding the outlier case of Somalia, this relation disappeared.

Therefore, using linguistic and religious distance as a way of measuring the room for cultural misperceptions of peacekeepers is limited. On the one hand, a high cultural distance itself does not lead to less support of the local population for the peacekeeping mission. On the other hand, even a low cultural distance does not prevent peacekeepers from crucially misreading cultural dynamics in the country of operations and creating local resentment against them, as the case of Somalia has shown. Deploying a culturally similar peacekeeping force in terms of a low linguistic or religious distance alone does not guarantee a cultural understanding and sensitivity for local issues. Peacekeepers and UN officials must be very open for the opinions and viewpoints of the local population and must make sure to include them properly in the peacemaking and peacekeeping process. A stronger emphasis on bottom-up approaches of peacemaking, instead of simply imposing foreign peace initiatives, can ensure not only a more successful mission, but also greater support from the local population and better safety for the peacekeeping personnel (Autesserre 2014b).

Further studies should test more specifically for cultural misperceptions, which could possibly affect the relationship between local inhabitants and the peacekeeping mission, independently of their cultural distance. For instance, there are no thorough studies exploring what effect different conceptions by peacekeepers and local decision-makers regarding peacemaking and conflict resolution (top-down or bottom-up) have (Autesserre 2014a). Differing values or understandings of government can also be constituted by culture and could have an influence on the relations between peacekeepers and local citizens (Rubinstein 2005). A range of other unobserved factors could have influenced the relationship between the peacekeeping forces and

30 the local population. Qualitative and quantitative scholars should in the future examine other reasons that could lead to this relationship being good and fruitful.

Most importantly, further studies should focus on identifying other reasons for violence against peacekeeping forces. This is particularly relevant for UN officials and national militaries conducting peacekeeping missions. The collection of data on violence against peacekeepers needs to be expanded as well. The PAR dataset for example, is not considering attacks on the peacekeeping force, that do not result in a harmed or dead peacekeeper (Lindberg Bromley/Greek 2016). These attacks could still negatively affect the overall peacekeeping mission. Since the fatal mistakes in conflicts like Rwanda and Somalia, the UN has taken considerable measures to improve peacekeeping by making it more safe and effective. By further studying these aspects of peacekeeping, academics can significantly contribute to the further positive evolution of this crucial, unique instrument of international conflict resolution.

31

R

EFERENCES

Alesina, Alberto / Devleeschauwer, Arnaud / Easterly, William / Kurlat, Sergio / Wacziarg, Romain. 2003. Fractionalization. Journal of Economic Growth 8(2): 155-194.

Autesserre, Séverine. 2014a. Going Micro: Emerging and Future Peacekeeping Research. International Peacekeeping 21(4): 492-500.

Autesserre, Séverine. 2014b. Peaceland: Conflict Resolution and the Everyday Politics of International Intervention. Cambridge: Cambridge University Press.

Bellamy, Alex J. / Williams, Paul / Griffin, Stuart. 2004. Understanding Peacekeeping. Cambridge: Polity Press.

Beswick, Danielle. 2010. Peacekeeping, Regime Security and “African Solutions to African Problems”: exploring motivations for Rwanda’s involvement in Darfur. Third World

Quarterly 31(5): 739-754.

Bosman, Femke / Soeters, Joseph / Bari, Fatima Ait. 2008. Dutch Muslim Soldiers During Peace Operations in Muslim Societies. International Peacekeeping 15(5): 695-705.

Bove, Vincenzo / Elia, Leandro. 2011. Supplying peace: Participation in and troop contribution to peacekeeping missions. Journal of Peace Research 48(6): 699-714. Bove, Vincenzo / Ruggeri, Andrea. 2015. Kinds of Blue: Diversity in UN Peacekeeping Missions and Civilian Protection. British Journal of Political Science 46(3): 681-700. Bove, Vincenzo / Ruggeri, Andrea. 2019. Peacekeeping Effectiveness and Blue Helmet’s Distance from Locals. Journal of Conflict Resolution 63(7): 1630-1655.

Burns, Arthur Lee / Heathcote, Nina. 1963. Peace-Keeping by U.N. Forces: From Suez to Congo. London: Pall Mall.

32 Dallaire, Roméo. 2003. Shake Hands with the Devil: The Failure of Humanity in Rwanda. London: Arrow.

Davies, William D. / Dubinsky, Stanley. 2018. Language Conflict and Language Rights. Ethnolinguistic Perspectives on Human Conflict. Cambridge: Cambridge University Press.

Doyle, Michael W. / Sambanis, Nicholas. 2000. International Peacebuilding: A Theoretical and Quantitative Analysis. American Political Science Review 94(4): 779-801.

Duffey, Tamara. 2000. Cultural issues in contemporary peacekeeping. International Peacekeeping 7(1): 142-168.

Eberhard, David M. / Simons, Gary F. / Fennig, Charles D. 2019. Ethnologue: Languages of the World. 21nd Edition. Dallas: SIL International.

Fearon, James D. 2003. Ethnic and Cultural Diversity by Country. Journal of Economic Growth 8(2): 195-222.

Fjelde, Hanne / Hultman, Lisa / Lindberg Bromley, Sara. 2016. Offsetting Losses: Bargaining Power and Rebel Attacks on Peacekeepers. International Studies Quarterly 60(4): 611-623 Foreign Legion Info. 2018. French Foreign Legion structure in 2018. Retrieved 3 July 2019, from: http://foreignlegion.info/2018/01/16/french-foreign-legion-structure-in-2018/

Fortna, Virginia Page. 2004. Does Peacekeeping Keep Peace? International Intervention and the Duration of Peace after Civil War. International Studies Quarterly 48(2): 269-292. Gelot, Linnéa / Gelot, Ludwig / de Coning, Cedric. 2012. Supporting African Peace Operations. Uppsala: Nordiska Afrikainstitutet.

Gilligan, Michael J. / Sergenti, Earnest J. 2008. Do UN Interventions Cause Peace? Using Matching to Improve Causal Interference. Quarterly Journal of Political Science 3(2): 89-122.

33 Haddad, Said. 2010. Teaching Diversity and Multicultural Competence to French

Peacekeepers. International Peacekeeping 17(4): 566-577.

Haass, Felix / Ansorg, Nadine. 2018. Better peacekeepers, better protection? Troop quality of United Nations peace operations and violence against civilians. Journal of Peace Research 55(6): 742-758.

Henke, Marina E. 2016. Has UN Peacekeeping Become More Deadly? Analyzing Trends in UN Fatalities. New York: International Peace Institute.

Hilbe, Joseph M. 2014. Modeling Count Data. Cambridge: Cambridge University Press. Hirsch, John L. / Oakley, Robert B. 1995. Somalia and Operation Restore Hope. Reflections on Peacemaking and Peacekeeping. Washington: United States Institute of Peace.

Hofstede, Geert / Hofstede, Gert Jan / Minkov, Michael. 2010. Cultures and Organizations. Software of the Mind. Intercultural Cooperation and Its Importance for Survival. 3rd Edition. New York: McGraw-Hill.

Hultman, Lisa / Kathman, Jacob / Shannon, Megan. 2013. United Nations Peacekeeping and Civilian Protection in Civil War. American Journal of Political Science 57(4): 875-891. Hultman, Lisa / Kathman, Jacob / Shannon, Megan. 2014. Beyond Keeping Peace: United Nations Effectiveness in the Midst of Fighting. American Political Science Review 108(4): 737-753.

Kathman, Jacob. 2013. United Nations Peacekeeping Personnel Commitments, 1990-2011. Conflict Management and Peace Science 30(5): 532-49.

Lindberg Bromley, Sara / Greek, Maria. 2016. UCDP Peacemakers at Risk (PAR) Codebook and User Guide. Version 1.1-2016. Uppsala: Department of Peace and Conflict Research.

34 Lindberg Bromley, Sara. 2017. Keeping Peace while Under Fire. The Causes, Characteristics and Consequences of Violence against Peacekeepers. Uppsala: Department of Peace and Conflict Research.

Lindberg Bromley, Sara. 2018. Introducing the UCDP Peacemakers at Risk dataset, sub-Saharan Africa, 1989-2009. Journal of Peace Research 55(1): 122-131

Mackinlay, John. 1989. The Peacekeepers: An Assessment of Peacekeeping Operations at the Arab-Israel Interface. London: Unwin Hyman.

Newby, Vanessa F. 2018. Power, politics and perception: the impact of foreign policy on civilian-peacekeeper relations. Third World Quarterly 39(4): 661-676.

Rubinstein, Robert A. 2003. Cross-Cultural Considerations in Complex Peace Operations. Negotiation Journal 19(1): 29-49.

Rubinstein, Robert A. 2005. Intervention and Culture: An Anthropological Approach to Peace Operations. Security Dialogue 36(4): 527-544.

Rubinstein, Robert A. 2008. Peacekeeping Under Fire. Culture and Intervention. Boulder: Paradigm.

Rubinstein, Robert A. / Keller, Diana M. / Scherger, Michael E. 2008. Culture and Interoperability in Integrated Missions. International Peacekeeping 15(4): 540-555.

Sahnoun, Mohamed. 1994. Somalia. The Missed Opportunities. Washington: United States Institute of Peace.

Salverda, Nynke. 2013. Blue helmets as targets: A quantitative analysis of rebel violence against peacekeepers, 1989-2003. Journal of Peace Research 50(6): 707-720.

Seet, Benjamin / Burnham, Gilbert M. 2000. Fatality Trends in United Nations Peacekeeping Operations 1948-1998. Journal of the American Medical Association 284(5): 598-603.

35 Spolaore, Enrico / Wacziarg, Romain. 2016. Ancestry, Language and Culture. In: Ginsburgh, Victor / Weber, Shlomo (Ed.): The Palgrave Handbook of Economics and Language. London: Palgrave. 174-211.

Sundberg, Ralph / Melander, Erik. 2013. Introducing the UCDP Georeferenced Event Dataset. Journal of Peace Research 50(4): 523-532.

Talentino, Andrea Kathryn. 2007. Perceptions of Peacebuilding: The Dynamic of Imposer and Imposed Upon. International Studies Perspectives 8(2): 152-171.

UN Department of Peacekeeping Operations (DPKO). 1999. Multidisciplinary Peacekeeping: Lessons From Recent Experience. Retrieved 23 July 2019, from:

https://www.globalpolicy.org/component/content/article/199/40912.html

UN Department of Peacekeeping Operations (DPKO). 2018. List of Peacekeeping Operations 1948 – 2018. Retrieved 3 July 2019, from:

https://peacekeeping.un.org/sites/default/files/180413_unpeacekeeping-operationlist_2.pdf UN Department of Peacekeeping Operations (DPKO). 2019. Troop and Police Contributors. Retrieved 10 July 2019, from: https://peacekeeping.un.org/en/troop-and-police-contributors Van der Lijn, Jaïr / Dundon, Jane. 2014. Peacekeepers at Risk: The Lethality of Peace Operations. Stockholm: Stockholm International Peace Research Institute (SIPRI).

Victor, Jonah. 2010. African peacekeeping in Africa: Warlord politics, defense economics, and state legitimacy. Journal of Peace Research 47(2): 217-229.