characteristics, context and risk: ngo insecurity in

25
Characteristics, context and risk: NGO insecurity in conflict zones Larissa Fast Visiting Assistant Professor, Kroc Institute for International Peace Studies, University of Notre Dame, US This paper reports on research conducted on the insecurity of non-governmental organisations (NGOs) between 1999 and 2002, with the goal of contributing to the debate on the reasons why NGO actors are targets of violence in conflict settings.The research involved the collection of data from three countries—Angola, Ecuador and Sierra Leone—and exploration of the relationship between levels of insecurity, context and the characteristics of NGOs. Four risk factors appear to heighten the degree of insecurity that NGOs face: ( 1) carrying out multiple types of activities and providing material aid; (2) ‘operationality’ (that is, implementing programmes and activities); (3) working with both sides of the conflict; and (4) integrating into the local community.The paper discusses the methodological approach adopted for the research, the differences between ambient and situational insecurity and the findings related to risk factors. It concludes with a summary of the study’s implications. Keywords: development, NGOs, relief, security Introduction The bombing of the headquarters of the United Nations (UN) in the Canal Hotel com- plex and the placing of explosives outside the offices of the International Committee of the Red Cross (ICRC) in Baghdad, Iraq, in 2003 serve to highlight the risks associated with humanitarian action.The kidnapping and murder of Margaret Hassan, long-time director of CARE, in Iraq in 2004, the shooting of a member of the Disaster Assist- ance Response Team (DART), United States Agency for International Development (USAID), in Darfur, Sudan, in March 2005, and the deaths of 17 non-governmental organisation (NGO) workers in Sri Lanka in August 2006 provide further evidence of the sometimes deadly dangers aid workers face. The prevailing perception is that violence against aid workers is increasing globally. The Office of the United Nations Security Coordinator determined that 229 inter- national and national UN staff died in malicious attacks between 1992 and mid-2005 (UN, 2005). A 1998 ICRC report chronicles security incidents from around the world, including places not normally thought of as experiencing high crime or violence (ICRC, 1998). A number of contemporary books document aid workers’ experi- ences, with stories that routinely underline the hazards of their work (Danieli, 2002; Bergman, 2003). Additional studies suggest a broader set of factors that deteriorate the climate in which aid workers operate, such as the widespread availability of small arms and light weapons (SALW) (see Muggah and Berman, 2001; Muggah, 2001; Beasley, Disasters, 2007, 31(2): 130 - 154. © 2007 The Author(s). Journal compilation © Overseas Development Institute, 2007. Published by Blackwell Publishing, 9600 Garsington Road, Oxford, OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA doi:10.1111/j.0361-3666.2007.01001.x

Upload: others

Post on 08-May-2022

3 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Characteristics, context and risk: NGO insecurity in

Characteristics, context and risk: NGO insecurity in conflict zonesLarissa Fast Visiting Assistant Professor, Kroc Institute for International Peace Studies, University of Notre Dame, US

This paper reports on research conducted on the insecurity of non-governmental organisations (NGOs) between 1999 and 2002, with the goal of contributing to the debate on the reasons why NGO actors are targets of violence in conflict settings. The research involved the collection of data from three countries—Angola, Ecuador and Sierra Leone—and exploration of the relationship between levels of insecurity, context and the characteristics of NGOs. Four risk factors appear to heighten the degree of insecurity that NGOs face: (1) carrying out multiple types of activities and providing material aid; (2) ‘operationality’ (that is, implementing programmes and activities); (3) working with both sides of the conflict; and (4) integrating into the local community. The paper discusses the methodological approach adopted for the research, the differences between ambient and situational insecurity and the findings related to risk factors. It concludes with a summary of the study’s implications.

Keywords: development, NGOs, relief, security

IntroductionThe bombing of the headquarters of the United Nations (UN) in the Canal Hotel com-plex and the placing of explosives outside the offices of the International Committee of the Red Cross (ICRC) in Baghdad, Iraq, in 2003 serve to highlight the risks associated with humanitarian action. The kidnapping and murder of Margaret Hassan, long-time director of CARE, in Iraq in 2004, the shooting of a member of the Disaster Assist-ance Response Team (DART), United States Agency for International Development (USAID), in Darfur, Sudan, in March 2005, and the deaths of 17 non-governmental organisation (NGO) workers in Sri Lanka in August 2006 provide further evidence of the sometimes deadly dangers aid workers face. The prevailing perception is that violence against aid workers is increasing globally. The Office of the United Nations Security Coordinator determined that 229 inter-national and national UN staff died in malicious attacks between 1992 and mid-2005 (UN, 2005). A 1998 ICRC report chronicles security incidents from around the world, including places not normally thought of as experiencing high crime or violence (ICRC, 1998). A number of contemporary books document aid workers’ experi-ences, with stories that routinely underline the hazards of their work (Danieli, 2002; Bergman, 2003). Additional studies suggest a broader set of factors that deteriorate the climate in which aid workers operate, such as the widespread availability of small arms and light weapons (SALW) (see Muggah and Berman, 2001; Muggah, 2001; Beasley,

Disasters, 2007, 31(2): 130−154. © 2007 The Author(s). Journal compilation © Overseas Development Institute, 2007. Published by Blackwell Publishing, 9600 Garsington Road, Oxford, OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA

doi:10.1111/j.0361-3666.2007.01001.x

Page 2: Characteristics, context and risk: NGO insecurity in

Characteristics, context and risk: NGO insecurity in conflict zones 131

Buchanan and Muggah, 2003; Buchanan and Muggah, 2005) or the ‘War on Terror’ (Helton and Loescher, 2003; Donini, 2003; Macrae and Harmer, 2003). Understandably, bombings and murder draw the media, but the numbers of aid workers killed in the line of duty are not an accurate reflection of the issue. Security incidents differ in degree, and the least severe of these garner little attention. Dramatic incidents, like those mentioned above, capture headlines, while accounts of rape, armed robbery or individuals caught in the crossfire seldom result in more than a report to head-quarters. The rape of and violence against national staff members are underreported (see, for example, InterAction, 2001; Fast, 2002), and various studies acknowledge the dearth and incompleteness of data (Dworken, 1998; Sheik et al., 2000; Barnett, 2004). Some analysts intimate that the increased numbers of humanitarian workers operating in conflict areas accounts, in part, for the rise in exposure to violence. Because accurate and comparable statistics have been difficult to find and few agencies have conducted systematic incident analyses, it has been impossible to assess if this is true or what might have caused specific incidents (Van Brabant, 1998; Sheik et al., 2000). Recent studies, however, counter the prevailing wisdom, proposing that deaths have stabilised since the mid-1990s (Buchanan and Muggah, 2005). Furthermore, ‘broad brush assumptions about insecurity are unhelpful: each context for humanitarian work needs to be con-tinually assessed for emerging threats and changing threat patterns’ (Barnett, 2004, p. 25). Despite advances in security management approaches (see, for example, Rogers and Sytsma, 1998; Roberts, 1999; Van Brabant, 2000; Bickley, 2003; Mayhew, 2004) and legal protection through UN Security Council Resolution 1502, the 1999 Convention on the Safety of United Nations and Associated Personnel and the 2005 Optional Pro-tocol, the available statistics point to the substantial risks to humanitarians. Research projects have attempted to analyse trends, especially in fatalities, or compile data on the type and number of incidents involving humanitarian personnel. Sheik et al. (2000) identify intentional violence as the leading cause of death (253 of 375 fatalities between 1985 and 1998), find that many deaths occur early in an assignment, and conclude that the total number of deaths of NGO workers is increasing, whereas deaths of UN programme employees and peacekeepers began to decrease after 1995. The authors suggest that these trends reflect the violence related to refugee movements and dis-placement in general (Sheik et al., 2000). In a related study documenting the numbers of deaths of UN peacekeepers, Seet and Burnham (2000) conclude that the rise in deaths is due to the growing number and scale of missions, and that the relative risk to peacekeepers escalates in missions that include humanitarian assistance. The analysis by King (2002) reveals that there are more reports of deaths due to acts of violence than to accidents, and highlights the elevated risks associated with ambushes on vehicles and convoys. His later study names 2003 as the ‘year of living dangerously’, following the highest number of humanitarian worker fatalities (King, 2004). Another data set of 729 fatalities among humanitarians between 1990 and 2000 based on news sources affirms acute violence as the cause of 82 per cent of all deaths, with NGO agencies experiencing the highest number, and names Africa as the most dangerous continent, with 66 per cent of reported fatalities (Abbott, 2005). No Relief confirms

Page 3: Characteristics, context and risk: NGO insecurity in

Larissa Fast132

the injurious impact of firearms in terms of the deadly and more subtle threats that pervade the daily lives of humanitarian and development workers (Buchanan and Muggah, 2005). A number of ongoing initiatives are focusing on the morbidity and mortality of humanitarian workers or examining the operational, policy and philosoph-ical implications of security concerns for humanitarian action. Despite these developments, few collaborative initiatives to document and analyse security incidents exist, for various reasons, such as confidentiality, liability and the sensitive nature of the subject. Furthermore, many of the authors that do address the issue of the insecurity of aid personnel aim to establish trends or concentrate on the contextual reasons for violence against them, rather than seeking to explain why cer-tain NGOs are targeted and not others, even among those operating in the same country or area. If global trends are to blame, what accounts for the different levels of insecurity particular NGOs face in the same country? Exactly which NGOs are targeted and what differentiates them from their sister organisations that are somehow more insulated from violence? This paper addresses these questions by summarising research conducted between 1999 and 2002 on the characteristics of NGOs that heighten their levels of insecurity. Specifically, the contribution of the research is threefold:

• first, it articulates a difference between situational and ambient insecurity; • second, it suggests four potential risk factors (multiple activities combined with

material aid; ‘operationality’ (that is, implementing programmes and activities); work-ing with both sides of the conflict; and more integration into local communities) for NGO actors in conflict zones; and

• third, it highlights some implications arising from the research.

The paper begins with a description of the methodological approach and its general parameters, challenges and limitations before exploring the difference between ambient and situational insecurity. Next, it outlines the study findings, and finally, it points up some implications of the research.

Methodological approach, challenges and limitationsTo examine the question of why some NGOs appear to experience higher levels of insecurity in the same environment, the research explored the influence of NGO characteristics on levels of insecurity in similar and different contexts to determine if and how the profiles of more and less insecure NGOs differed across them. To do this, the study employed two types of comparative strategies:

• a small ‘n’ study of three countries—Angola, Ecuador and Sierra Leone—that enabled a structured comparison at the contextual, macro level; and

• a large ‘N’ study that compared micro-level variables across NGOs.

Page 4: Characteristics, context and risk: NGO insecurity in

Characteristics, context and risk: NGO insecurity in conflict zones 133

The research was multi method, employing expert interviews, a literature review and a survey of NGOs to contrast levels of insecurity in the three country cases. It involved qualitative and quantitative analysis of country-level contextual variables and selected micro-level variables (that is, characteristics of NGOs). Specifically, it com-pared reported incidents of violence and indicators of NGO characteristics to assess the impact of these features on levels of NGO insecurity in similar (situations of ongoing violence) and different (relatively stable) contexts. A central premise of the research was that contextual factors could explain overall levels of insecurity in a country, but not why any one NGO in particular found itself embroiled in security incidents. To test this assertion, the research compared three cases—two similar (Angola and Sierra Leone) and one different (Ecuador)—based on a series of macro-level indicators. The author’s choice of methodology for studying the three country cases drew on a hybrid of the focused comparison models of Andrew Faure (1994) and Alexander George (1979).1 This method served the goal of deter-mining whether it was possible to generate a hypothetical profile of an ‘insecure’ organisation that held up across contexts. The purpose was not to isolate one or more contextual variables (or even to conduct an in-depth analysis of countries), but to compare similar and different cases in order to discount context as a key factor in ex-plaining the difference in levels of NGO insecurity. In essence, context became the variable on which the third country case (Ecuador) differed from the other two (Angola and Sierra Leone). The literature review and interviews generated three categories of NGO charac-teristics that appeared to affect an organisation’s level of insecurity: organisational identity; programmes and operations; and the philosophy and practice of interven-tion.2 These categories constituted the micro-level variables referred to above (see Figure 1). Organisational identity refers to those characteristics that define who the NGO is, or the qualities it presents to the world. These characteristics are relatively fixed, and include the organisation’s primary mandate, nationality (location of its headquarters), and affiliation (faith-based or secular body). This study examined only an organisation’s primary mandate in relation to insecurity. Whereas an organisation’s identity extends beyond its activities in any one country, its specific programmes and operations vary across countries and respond to needs, context and/or cultural differences. An organi-sation’s programmes and activities included:

• the type of activities in which an NGO engaged (such as public health, conflict reso-lution, education or human rights);

• the location of its activities (used to assess the impact of working in a region contain-ing conflict commodities); and

• whether the NGO was operational (meaning it implements its own programmes) or non-operational (provides funding or support to other organisations to permit them to conduct activities).

Additionally, programmes and operations included:

Page 5: Characteristics, context and risk: NGO insecurity in

Larissa Fast134

• how long an organisation was involved in a particular country; • its programme ‘target groups’; and • its partner organisation(s).

This study investigated the impact of the diversity of activities, location of its activi-ties and its method of operation (‘operationality’). Furthermore, NGOs frequently have similar tasks, but typically they differ in the way in which they execute them (for instance, they involve different groups of people in programmes, or identify or emphasise different objectives that affect their relationships with partner groups). These less tangible and verifiable elements refer to an NGO’s philosophy and practice of intervention. These elements pertaining to how the organi-sation intervenes, or its style of operating, are often difficult to measure and the study investigated only three:

• ‘impartiality’ (defined as working with both sides3); • ‘engagement’ (defined as an organisation’s level of integration into the local com-

munity); and • ‘participation’ (defined as the extent to which the individuals or communities an

organisation worked with were involved in programme decision-making).

Together, these three categories of characteristics comprised the conceptual frame-work, and the seven variables became the focus of the survey phase of the research (see Figure 1). The survey targeted organisations working in Angola (January 1998–December 1999), Ecuador (January 1999–December 2000) or Sierra Leone (May 1997–July 1999) to

ORGANISATIONAL IDENTITY

PROGRAMMES AND OPERATIONS

PHILOSOPHY AND PRACTICE OF

INTERVENTION

- Diversity of activities- Location of activities- Method of operation

- ‘Impartiality’/working with both sides

- Engagement/integration- Participation

- Primary mandate

Figure 1 Micro-level factors influencing levels of NGO insecurity

Page 6: Characteristics, context and risk: NGO insecurity in

Characteristics, context and risk: NGO insecurity in conflict zones 135

obtain background information, as well as information on organisational policies and procedures, practices and principles, security incidents and security measures. The survey itself included 20 closed questions and 21 open-ended questions. Three were in the form of matrices and asked a series of questions about organisational policies, acts of violence the organisation experienced, and the types of security measures an organi-sation had in place. The results provided the data for assessing the micro-level factors related to NGO insecurity. The sampling frame for the research included all NGOs with headquarters outside of the case-study country—either with a permanent staff presence or those who sent staff to monitor and/or assess projects or provide techni-cal assistance on a short-term basis—that worked in one or more of the case-study countries during the period under investigation.4 Because one survey was incomplete, the final N for this study was 26. A total of 20 NGOs participated, with a number of these submitting surveys for more than one country.5 This represented a 37.5 per cent survey response rate. The 20 participating organisations varied greatly in size, from a handful to several hundred staff members. In terms of nationality, the actual sample was weighted towards the US (65 per cent), with the remainder headquartered in Canada or Europe. Most of the organisations had a development mandate (46 per cent), followed by multi-mandate institutions (39 per cent) and relief entities (12 per cent). The sample did include those with conservation or conflict resolution mandates (most of which appeared under the multi-mandate grouping), but it did not include any human rights or advo-cacy bodies. It incorporated faith-based and secular organisations, and operational (65 per cent) and non-operational organisations (12 per cent), as well as those that combined the two (17 per cent) or conducted periodic field visits (six per cent). Due to the relatively low N for the study and the various biases of the sample population, it is not possible to generalise the results of this research beyond the sample population. Despite this limitation, the results, as discussed below, point to the need for a more complex analysis of security issues—one that includes contextual factors, the primary focus and emphasis of most analysts, and the more understudied elements of an NGO’s mission, mandate and operations. Finally, to determine an organisation’s level of insecurity, the survey asked respond-ents about threats or acts of violence the NGO experienced in a particular country during a specific period, including an open-ended question that invited respondents to elaborate on the possible reasons for any one reported targeted incident. Although some respondents provided particulars, the criterion variable (called insscore, short for insecurity score) combined only the frequency of incidents with a weighted score for severity. To obtain this variable, the research categorised incidents according to ‘low’, ‘medium’ or ‘high’ severity (see Table 1). Low severity incidents received a weight of 0.2, medium incidents a weight of 0.6, and high incidents a weight of 1.0. Thus, if an organisation experienced three low incidents (3 x 0.2 = 0.6), one medium incident (1 x 0.6 = 0.6) and two high incidents (2 x 2 = 2) during the specified period, its criterion variable insscore would be 3.2. The research classified organisations with an insscore value of less than two as ‘less insecure’, and organisations with an insscore value

Page 7: Characteristics, context and risk: NGO insecurity in

Larissa Fast136

of more than two as ‘more insecure’.6 These less and more insecure groupings became the basis for the levl_ins (level of insecurity) variable. To analyse the data, the research used a combination of quantitative, statistical analysis and qualitative, frequency analysis to look for patterns in the responses. Research findings are discussed in more detail below. The difficulties of conducting research on NGO security issues are numerous and they have confounded other data collection and research attempts. This study was no exception. The lack of data was a particular handicap, making it impossible to gener-alise the results to a larger population. Because the sample population was small, the research used qualitative and quantitative data and analysis and attempted to verify or confirm the results through both mechanisms. Additionally, it was not possible to con-trol for confounding variables using statistical techniques because of the type of data (non-parametric) and the limited sample size. The research results suggest correlation rather than causation, and hence refer to factors that appear to heighten levels of risk. Two additional challenges, underreporting and targeting, both discussed further below, also limit the value of the findings. A number of issues emerged concerning the underreporting of incidents: who reports, when, and what they report. First, national and expatriate staff based in the

Table 1 Incident severity classifications

LOW (weight=0.2) Examples of incidents

Ambient insecurity—present in the situation, by virtue of being in the wrong place at the wrong time

• Aggressive crowd • Generalised violence due to war• Evacuation due to threats• Verbal threat of violence• Stolen vehicle• Landmine (no harm to individuals)• Physical harassment

MEDIUM (weight=0.6) Examples of incidents

Acts of violence against property, ambient violence with an increase in level of destruction/damage

• Theft at office compound • Theft at warehouse/storage site• Theft at residence• Break-in at residence• Break-in at office compound• Carjacking

HIGH (weight=1) Examples of incidents

Acts of violence against individuals, targeted acts against property or individual

• Artillery fire • Aerial bombardment• Fatality• Armed robbery• Kidnapping/hostage taking• Rape or sexual violence• Stabbing• Bombing• Arson• Physical attack• Landmine (harm to individual)

Page 8: Characteristics, context and risk: NGO insecurity in

Characteristics, context and risk: NGO insecurity in conflict zones 137

field and at headquarters responded to the survey. Headquarters staff members tend to be more removed from the realities of fieldwork and have a different sense of the im-portance, number or type of security incidents. In contrast, field personnel have a better sense of the overall picture of such incidents. This first-hand experience, how-ever, has implications for incident reporting. ‘Danger habituation’, or the phenomenon of acclimatising oneself to higher levels of violence and eventually considering these higher levels to be normal, inevitably affects reporting. Individuals who have lived in violent environments for extended periods may not report low, medium, or even high severity incidents precisely because they no longer consider them to be security inci-dents or believe them to be anomalies or unimportant events. Instead, they become a normal part of daily life. Muggah (2001, pp. 16–17) broaches the issue of the impact of ‘thresholds of personal insecurity’ on incident reporting in relation to danger habitua-tion, citing an instance in which survey respondents reported never having experienced a ‘serious incident involving firearms’, yet also responded affirmatively to having been fired upon. The second issue emphasises the importance of real-time reporting mechanisms. A limitation of this study is that the period under review was, at a minimum, 18 months in the past, raising the possibility that survey respondents did not accurately recall the frequency or severity of incidents. Third, standardised reporting needs to include definitions of what constitutes an act of violence. For example, one respondent did not report any acts of violence against the NGO in the survey. Yet this same individual indicated in written comments at the end that the organisation’s facilities had been robbed within the past few years but that it did not experience any ‘violent incidents against a person’. Muggah also indicates discrepancies in reporting within the same country office. He found that ‘there were many instances where [organisation staff were] unaware of people being directly or indirectly targeted by firearms, grenades, mortars, artillery, etc.’ (Muggah, 2001, p. 16). If individuals within the same country office are unaware of incidents, then those outside the field office are even less likely to know about them. Bollettino (2003) documents similar issues for the United Nations Children’s Fund (UNICEF) and calls for routine reporting of incidents to enable accurate threat analyses, thereby helping to reduce inci-dents and respond to the field environment. Furthermore, the idea of targeting is particularly difficult to study. The most direct and reliable method of tackling this question is to put it to those who are doing the targeting, a thorny task for a variety of reasons. Instead, the survey enquired whether respondents thought an incident was aimed at their NGO and a follow-up question asked them to describe why. Only 13 of 46 incidents highlighted by respondents who com-pleted this part of the matrix (reported incident N=66) were aimed against the NGO. Respondents may have been reluctant to report targeted incidents for several reasons. The incident may have been random, or the target may not have been clearly identi-fied. Alternatively, if an attack was directed against an organisation, it is conceivable that it directly or indirectly did something to provoke or evoke the event. This raises questions of liability as well as potential issues about fault or responsibility, whether deserved or not.

Page 9: Characteristics, context and risk: NGO insecurity in

Larissa Fast138

Those who did report targeted incidents provided an assortment of explanations. The most prevalent response related to the targeting of the organisation’s assets. Two responses suggested that the incident was the result of an administrative action (for example, the hiring or firing of a particular individual). Another respondent indicated that the level of risk increased at certain times of the year (such as Christmas), while another said that NGOs were a threat to a rebel group’s agenda. At least one survey response indicated that a particular individual, who happened to be employed by the organisation, was targeted. Thus, it was not the organisation that was targeted but the individual. This last response points up an intriguing dimension of targeting. Although the wording of the question (‘Do you think the incident was aimed specifically at your NGO’) is not ambiguous, a ‘no’ response was unclear, as it could have meant the inci-dent was random (a result of ambient levels of insecurity) or it could have been a protest against foreigners or NGOs in general. In other words, the survey did not gather information about targeting of a specific organisation as opposed to whether incidents were directed against (aid) organisations in general. Numerous sources refer to increases in targeted incidents involving aid workers, NGOs, UN personnel or agencies, but no (readily accessible) data currently exists on exactly who or what is being targeted. Are directed acts or threats of violence a pro-test against an individual actor (such as a particular person or organisation) or against a collective presence (for instance, foreigners or the UN)? Events in Afghanistan and Iraq seem to indicate that NGOs and/or humanitarian actors are targeted because they are ‘soft targets’ and therefore more vulnerable. Others, however, link the reasons for the 2003 UN compound bombing in Baghdad to the decade of sanctions that detri-mentally affected the Iraqi population (Independent Panel, 2003, p. 19) or to connec-tions between the United Nations Assistance Mission for Iraq (UNAMI) and the US as an occupying force (Reynolds, 2003).

Situational and ambient securityIn general, security analysts distinguish security threats (resulting from acts of violence) from threats to safety. These acts of violence include but are not limited to robbery, aggressive crowds, landmines, aerial bombardments, artillery fire, ambush, crossfire, and hostage taking. In contrast, threats to the safety of aid workers arise from accidents, frequently driving accidents, or health issues (malaria, human immunodeficiency virus (HIV), disease), some of which derive from unprotected sex (Martin, 1999). Analysts debate whether security or safety threats result in the most casualties among aid work-ers. Several studies (King, 2002; Rowley, 2005; Abbott, 2005) show that the majority of deaths are from intentional violence, while others (Martin, 1999; Peytremann et al., 2001) claim safety incidents are more significant.7 This distinction between security and safety is useful in narrowing the scope of acts or consequences categorised within NGO insecurity. A second distinction related to security provides added clarity. Lee (1995), in eluci-dating the types of danger researchers face in doing fieldwork, differentiates ambient

Page 10: Characteristics, context and risk: NGO insecurity in

Characteristics, context and risk: NGO insecurity in conflict zones 139

from situational danger. Specifically, ‘[a]mbient danger arises when the researcher is exposed to otherwise avoidable dangers simply from having to be in a dangerous setting for the research to be carried out. . . . Situational danger arises when the researcher’s presence or actions evoke aggression, hostility, or violence from those within the setting’ (Lee, 1995, pp. 3–4). A similar distinction emerges from this study, in which ambient insecurity refers to a generalised condition of violence in a country or region and situa-tional insecurity denotes those acts of violence arising out of a particular situation involv-ing a certain NGO. In this way, ambient insecurity is a characteristic of a particular country context. The security incidents NGO personnel face if they are ‘caught in the wrong place at the wrong time’ fall into this category, because it is equally likely that others might experience the same security incident. Situational insecurity, however, covers violence directed at a single NGO or group of NGOs. In some cases, it is obvious why a particular NGO was targeted (such as to obtain resources), while in others the reason is unclear.8 Although to some extent ambient and situational insecurity inci-dents diverge on the basis of targeting and/or intention, the distinction between the two is not this straightforward. Ambient insecurity affects everyone—NGOs, civil-ians, other international actors—whereas situational insecurity involves deliberate and unintended acts of violence against an actor that arise out of a particular situation. The research proposes that contextual factors could explain overall ambient levels of insecurity, but not situational insecurity for NGOs.9 Instead, to explicate situational insecurity—insecurity that affects some NGOs and not others—it is necessary to look to the characteristics of NGOs. An analogy is useful here to illustrate this thesis. If some-one robs a house, it is likely that the thief has one or more reasons for picking it: the neighbourhood in which it is located, poor lighting, a grudge against the owners, or the expected spoils. In much the same way, an NGO that experiences an act of violence (situational insecurity) is likely chosen for a reason. Some characteristic makes the NGO more vulnerable to attack. The simple presence of the NGO may be the cause of the attack, and an action of the NGO that evokes or provokes the violence may be deli-berate or unintended. The point is not to attribute blame, but rather to create analytical categories of risk that allow NGOs to develop strategies and mechanisms to reduce these dangers. Con-trary to the picture of helpless and innocent NGOs caught up in the violence that surrounds them, this research hypothesised that the NGO’s identity, its programmes and operations in a country, and its style of intervening influenced whether or not its staff and/or assets were more likely to be subjects of acts of violence. NGOs have a degree of control over, and therefore ability to change, their mandate and programmes in a country but little influence over the macro-level variables that undermine their levels of security. In other words, they are able to affect situational insecurity but have little ability as individual actors to shape levels of ambient insecurity. Furthermore, not enough consensus exists within the humanitarian or NGO communities about the reasons and corresponding remedies for mounting insecurity; thus, it is impossible to have a collective impact on levels of ambient or situational insecurity. The ‘purists’ argue that increased insecurity is due to the ‘degradation’ of humanitarian principles, while

Page 11: Characteristics, context and risk: NGO insecurity in

Larissa Fast140

the ‘integrationists’ assert that it has more to do with the changing environment in which violence is taking place. Both agree that humanitarian action is changing, but differ on the antidote (Donini, 2003). To some extent, as explained further below, the results validated the above hypothesis.

Risk factors for NGO actors in conflict zonesThe research attempted to identify NGO characteristics that increase their level of insecurity in the context of ongoing violence or relative stability, and the specific hypoth-eses linked the macro-level variables to ambient insecurity and the micro-level variables to situational insecurity. This section reports on the findings, keeping in mind the limitations noted above. The study findings indicate, as expected, that, on a macro level, ambient levels of insecurity were higher in the countries experiencing war than in the relatively stable country. Additionally, the micro-level profiles of more and less insecure NGOs differ across the country context, suggesting that a variety of inter-related fac-tors lead to situational insecurity. Thus, context does matter in determining levels of ambient and situational insecurity. While the findings do not identify hypothetical profiles of less and more insecure NGOs, they do point to several attributes that seem to heighten insecurity in situations of continuing violence. In particular, four characteristics are associated with higher levels of insecurity (see Table 2):

• organisations carrying out two or more types of activities combined with the pro-vision of material aid;

• operational organisations;• those organisations working with both sides of a conflict; and• those organisations that are more integrated into the local community in a country.

For the first two variables, the qualitative frequency analysis established consistent and homogenous responses on the diversity of activities and method of operation variables, highlighting two characteristics that appear to heighten but not result in higher levels of insecurity. In other words, all of the more insecure NGOs carried out two or more types of activities and provided material aid and all of them were operational. Because of the nature of the variables (non-parametric), it is not possible to establish clear quantitative correlations for them. The quantitative analysis, however, supports relationships between the ‘working with both sides’ and ‘integration’ variables and insecurity. Statistical results demonstrate correlations between higher insecurity scores and lower values for the ‘working with both sides’ and ‘integration’ variables, as explained further below. That is, those NGOs working with both sides of a conflict, the third attri-bute, seem to be more insecure than ‘partial’ NGOs, especially in areas with higher levels of ambient (overall) insecurity, a result the qualitative analysis supports as well. As for more integrated NGOs, they seem to be more insecure than non-integrated NGOs

Page 12: Characteristics, context and risk: NGO insecurity in

Characteristics, context and risk: NGO insecurity in conflict zones 141

Table 2 Qualitative and quantitative findings for micro-level variables

Variables Quantitative results Qualitative results (frequency analysis)

Primary mandate Inconclusive* No pattern (that is, heterogeneous responses for both more and less insecure categories)

Diversity of activities (number of types or sectors of activity)

No relationship* No pattern (that is, heterogeneous responses for both more and less insecure categories)

Diversity of activities + material aid No relationship* Consistent and homogenous responses (that is, all ‘more insecure’ organisations carrying out two or more types of activities and providing material aid; variation/heterogeneity across re-sponses for ‘less insecure’ organisations)

Location of activities (present in regions with conflict commodities—diamonds and/or oil)

No relationship* No pattern (that is, heterogeneous responses for both more and less insecure categories)

Method of operation (operational/non-operational/…)

No relationship * Consistent and homogenous responses (that is, all ‘more insecure’ organisations were operational; variation/heterogeneity across responses for ‘less insecure’ organisations)

Working with both/one side of the conflict

Insscore variable, no groupingCorrelation Coefficient=-0.384 (p=0.071, not significant)Insscore variable, grouped (Angola and Sierra Leone)Correlation Coefficient=-0.696 (p=0.002, significant)

Levl_ins variable, no groupingCorrelation Coefficient=-0.292 (p=0.177, not significant)Levl_ins variable, grouped (Angola and Sierra Leone)Correlation Coefficient=-0.482 (p=0.050, significant)

Consistent and homogenous responses (that is, no ‘more insecure’ organisations worked with only one side of the conflict; variation/heterogeneity across responses for ‘less insecure’ organisations). See Table 3

‘Integration’ (extent of integration into local community)

Insscore variable, no groupingCorrelation Coefficient=-0.450 (p=0.021, significant)Insscore variable, grouped (Angola and Sierra Leone)Correlation Coefficient=-0.472 (p=0.041, significant)

Levl_ins variable, no groupingCorrelation Coefficient=-0.290 (p=0.151, not significant)Levl_ins variable, grouped (Angola and Sierra Leone)Correlation Coefficient=-0.299 (p=0.214, not significant)

No pattern (that is, heterogeneous responses for both more and less insecure categories)

‘Participation’ (extent of participation in programming decision-making processes)

No relationship** No pattern (that is, heterogeneous responses for both more and less insecure categories)

Notes* Used Lambda test to assess association between variables.** Used Spearman Rank-Order Correlation Coefficient to determine the absence or presence and direction of a relation-ship between the variables.

Page 13: Characteristics, context and risk: NGO insecurity in

Larissa Fast142

in all contexts. The strength of this relationship decreases when the relatively stable country (Ecuador) is removed from the analysis. Together the results point to the impor-tance of both micro- and macro-level factors and the interactions between them in determining levels of insecurity for NGOs. No relationships between an organisation’s insecurity score or its level of insecurity and the primary mandate, location of activities and participation variables emerged in the qualitative or quantitative analysis. The data do not support any relationship between insecurity and an organisation’s primary mandate, whether it worked in a province containing diamonds (Angola and Sierra Leone) or oil (Ecuador), or whether it involved those individuals or groups it worked with in programming decision-making processes. The data do support, though, relationships between insecurity and the diversity of activities, method of operation, working with both sides, and integration variables for the sample population, each of which is discussed below. First, the more insecure organisations are those carrying out multiple activities com-bined with the provision of material aid, even though not all organisations with this characteristic are classified as more insecure. It is somewhat surprising that the diver-sity of activities variable emerges as related to insecurity whereas the results for the primary mandate variable are inconclusive. Commentators have connected conflict-ing mandates (such as humanitarian relief and advocacy, conflict resolution or peace-building) to the apparent rise in targeted incidents involving NGOs (see, for example, African Rights, 1994). Presumably conflicting mandates lead to greater diversity of activities. The key finding here, however, is not related to the number of activities alone, but rather to the number of activities coupled with the provision of material aid. In the sample population, ‘material aid’ includes many items, ranging from food aid to edu-cational resources (for instance, books and paper). The analysis does not account for differences in types of material aid. The nature of relief and development operations means that they usually require some outside resources in the form of equipment (such as computers and communication devices, vehicles or food aid, all of which become even more valuable in situations of scarcity. Those attacking simply want access to and/or possession of these resources. The strategies vary to manage the risk associ-ated with material aid. In some cases, it is possible to manage the risk by making the commodities less ‘steal-able’. For example, in some contexts agencies discovered that they could lower their vulnerability to attack by sending cooked rice in a food convoy—uncooked rice had a substantial street value whereas cooked rice did not. Second, the more insecure NGOs are operational. Again, some less insecure organi-sations are also operational, but the consistency of the more insecure categorisation identifies this as a characteristic that heightened insecurity. This result is not particularly surprising, given that operational organisations’ field staff members often experience prolonged and constant exposure to ambient (overall) violence in contexts of ongoing violence, as in Angola and Sierra Leone. Additionally, operational organisations employ more staff, which could explain higher numbers of security incidents, and therefore levels of insecurity. Related research suggests some nuances related to risk factors for

Page 14: Characteristics, context and risk: NGO insecurity in

Characteristics, context and risk: NGO insecurity in conflict zones 143

staff members. One study (Sheik et al., 2000) reveals that one-third of all fatalities occurred within the first three months of an assignment, and one in six in the first six months of an assignment, highlighting different risk levels according to length of time in a new assignment. Third, the ‘working with both sides’ variable findings indicate that working with both sides of a conflict is correlated with both insecurity score and level of insecurity for the two countries with higher ambient security (Angola and Sierra Leone). A further intriguing element is that higher insecurity scores are correlated with lower values for this variable; the lower value corresponds to the response option of ‘worked with only one side of the conflict’ (see Table 3). The original hypothesis claimed that organisations working with both sides would be less insecure. However, the data disconfirm the original hypothesis and suggest the opposite, because those NGOs that worked with both sides appear to be more insecure than the ‘partial’ NGOs that worked with only one side (see Table 2).10 The frequency analysis (see Table 3) corrobo-rates these results. It shows that the more insecure organisations worked with both sides of the conflict, but to varying extents (that is, only, frequently, occasionally, in-frequently). None of the more insecure NGOs worked with only one side of the conflict, while seven (of 17) of the less insecure circled this response in the survey. Otherwise, both less and more insecure NGOs exhibit heterogeneity across the response categories. What explains the counter-intuitive finding related to this variable? In a polarised society, those intervening are often perceived as ‘for us or against us’ and defined in terms of in-groups or out-groups (Devine, 1995). If an NGO helps all sides, the danger is that all will see it as ‘against us’ by helping ‘the other’. Consequently, those NGOs working with two or more sides of the conflict may be targeted because they are seen to be helping the other side, despite the fact that relationships likely extend and assist-ance and/or resources likely flow to both sides. Partial NGOs (defined here as those working only with one side of the conflict), while allied with a particular group, may be more secure, paradoxically, precisely because they do not cross conflict lines. In this way, partiality provides protection while working with both sides may make inter-veners universally unpopular. Related to this, one interviewee proposed that giving assistance to some and not others would lead those without assistance to take simply what they needed, thus increasing the risk for all organisations. Van Brabant (2000, p. 13)

Table 3 Frequency analysis for ‘working with both sides’ and level of insecurity (N=23)

Degree of ‘working with both sides’ Less insecure More insecure

Worked only with two or more sides of the conflict 2 2

Worked frequently with two or more sides of the conflict 5 1

Worked occasionally with two or more sides of the conflict 2 2

Worked infrequently with two or more sides of the conflict 1 1

Worked only with one side of the conflict 7 0

Page 15: Characteristics, context and risk: NGO insecurity in

Larissa Fast144

highlights a similar interdependence among agencies in the same context, where the inaction, actions or strategies of one or several organisations affect all. Furthermore, in both Angola and Sierra Leone, NGOs had to acquire permission from the government to operate in the country. NGO activity was therefore concen-trated in areas under government, and not rebel, control. The operational constraints of permission and limited access could lead to lower levels of insecurity for NGOs. In contrast, the frontlines of violence are where an NGO would interact more fre-quently with both sides of the conflict. Yet to work on the frontlines increases the level of danger for NGOs, as the ambient level of violence rises. As a result, those NGOs most likely to interact with both sides of the conflict probably work in areas with more fighting, and hence higher ambient—as opposed to situational—levels of insecurity. In other words, factors related to the location of NGO activities and ambient levels of violence may explain the results more so than the fact that the NGO worked with both sides of the conflict. This too offers a plausible explanation for the unexpected findings. The findings for integration, the fourth variable, were similarly unexpected. The data suggest that NGOs that are more integrated into the local community are more in-secure. The conceptual framework proposes that the more integrated NGOs would have better connections and relationships, and ostensibly more credibility and legitimacy, in the communities in which they live and work. These connections and relationships, in turn, would provide a measure of security and protection, thereby making them more secure rather than more insecure. This idea is closely related to the acceptance model of security management (Van Brabant, 2000; Working Group on NGO Security, 1997). However, the results of this analysis suggest that more integration (or accept-ance) is related to higher levels of insecurity. Moreover, the correlation gets weaker when the data are grouped into similar and dissimilar contexts. It is still significant for the similar cases, but not significant in the dissimilar case. The frequency analysis un-covers no real patterns in the responses. More and less insecure organisations are weighted toward higher degrees of integration, but both vary across response subcategories. The distinction between ambient and situational insecurity is helpful in explaining the findings. Presumably, more integrated NGOs are more susceptible to the same (high) levels of violence that affect the civilian population as a whole, since an organisation’s status as an ‘outsider’ decreases as it becomes more integrated into and therefore engaged with the local community. Thus, integrated NGOs are more insecure but they may be more insecure because of the ambient levels of violence as opposed to situational violence. Alternatively, if they were targets of violence, it might be precisely because their level of identification (and possibly partiality) inhibits them from seeing the bigger picture, or results in targeting by the other side. If this is true, it is important to deter-mine whether the higher levels of insecurity are due to situational or ambient violence, or whether NGOs with a longer history in a country are indeed more integrated. A final element of the study findings relates to the underreporting of security inci-dents. Overall, when counting reported acts of violence from all organisations in the study, the total number of cases increases to 66. In the 26 questionnaires, 11 respondents indicated that their organisation experienced acts of violence, while 14 organisations

Page 16: Characteristics, context and risk: NGO insecurity in

Characteristics, context and risk: NGO insecurity in conflict zones 145

did not experience acts of violence and one respondent did not know. In classifying the severity of incidents (as explained above) and the frequency of incidents, the follow-ing picture emerges (see Table 4): the highest percentage of NGOs experienced low severity incidents (for a total of 39 per cent) and only 12 per cent (three out of 26) of NGOs in the sample experienced six or more incidents over the two-year period of the study. Comparing respondents’ perceptions of the frequency of incidents involving all NGOs in the country (the subject of one question in the survey) and the reported frequency of incidents involving NGOs in the sample unearths some inconsistencies between belief and reality. Forty-four per cent of the respondents indicated that NGOs were involved in several incidents per year while an aggregate of 43 per cent thought more than two to three times per year. Yet only three of the 26 NGOs, a very small subset of the total number of NGOs in a country, account for at least 18 of 66 total incidents over a two-year period across all three countries (see Table 4). Overall, survey respondents reported far fewer threats and/or acts of violence than expected, given media attention and other studies. This raises the likelihood that the frequency and even severity of security incidents are underreported in the survey results. The sample contained six organisations classified as more insecure and 20 organisations classified as less insecure. It is impossible to determine whether this distribution reflects or represents the actual numbers of more and less insecure NGOs in Angola, Ecuador and/or Sierra Leone. One limitation of the data was that participants self-selected, in part, based on their willingness to answer questions about the criterion variable. This could mean that the sample has higher numbers of less insecure organisations, which presumably were more willing to answer questions about insecurity, or lower numbers of less insecure NGOs, which did not see this type of research as relevant to their work. The reasons NGOs gave for not participating in the research reflected both assumptions. The underreporting of gender-based violence is of particular concern and sensitivity. Only one questionnaire specified the gender of the affected individual(s) as female, while four indicated that both genders were affected by an incident.11 Acts of gender-based violence are conspicuously, yet not surprisingly, absent from the survey results.

Table 4 Frequency and severity of reported incidents (N=26)

Frequency\severity

Low severity Medium severity High severity Total incidents (all severities)

None 61% (16) 77% (20) 69% (18) 46% (12)

Infrequent (1–2 incidents)

31% (8) 11.5% (3) 19% (5) 23% (6)

Occasional (3–5 incidents)

8% (2) 11.5% (3) 8% (2) 19% (5)

Frequent (6+ incidents)

0% (0) 0% (0) 4% (1) 12% (3)

100% (26) 100% (26) 100% (26) 100% (26)

Page 17: Characteristics, context and risk: NGO insecurity in

Larissa Fast146

The demographic questions in the survey did not ask for the gender of the respondent, but the majority of identified survey respondents were male. A study on the humani-tarian impact of small arms (Muggah, 2001) found anecdotal evidence that individuals from the same country office were unaware of incidents involving their colleagues. Given the type of incident and the social sensitivity of sexual and/or gender-based violence, it is even more likely that male (or even female) respondents would be unaware of acts of sexual violence, whether in the same office or not, simply because of its nature.

Implications of the studyThe limitations make it impossible to generalise the results of the study to a broader population. Nevertheless, the research highlights the importance of context for secu-rity, alludes to some potential implications for NGO activities and security policies, and identifies crucial areas for further analysis.

Context is crucial in determining levels of ambient and situational insecurityA central question of the research focused on the importance of context in determin-ing the degree of insecurity for individual NGOs on the micro and macro level. The research confirms what practitioners already know intuitively: context matters for both. Although a larger sample would make it possible to confirm or contradict these findings, the importance of context is not likely to change. On the micro level, the research findings show that it is not possible to generalise particular attributes that character-ise insecure NGOs across contexts since traits do not hold across similar and different settings, thus demonstrating the significance of contextual factors. On the macro level, overall ambient insecurity in Angola and Sierra Leone appears to be similar, while NGOs in Ecuador experienced lower levels of ambient insecurity. This is hardly surprising given the intense periods of violence that engulfed Angola and Sierra Leone in 1997, 1998 and 1999. Logically, contexts of ongoing violence will have higher levels of ambient violence. Crime and banditry, however, are important in determining levels of ambient and situational insecurity. More recent research suggests that crime and banditry may be the greater perceived and actual threat to humanitarians (Buchanan and Muggah, 2005). These findings reveal the importance of analysing the context in which NGOs and others work in order to contextualise better responses to insecurity. Together they suggest that a more complex model of NGO insecurity may be lurking in the shadows, one that deserves more attention in future research. For example, the conceptual frame-work mentioned above treats NGO characteristics as individual factors and context as separate from them. These are elements that a more interactive and dynamic conceptual model could address holistically. That is, the way in which context interacts with NGO characteristics, or the presence of particular combinations of NGO characteristics in certain settings, may be more important in determining an organisation’s level of

Page 18: Characteristics, context and risk: NGO insecurity in

Characteristics, context and risk: NGO insecurity in conflict zones 147

insecurity. Presumably, asking new and different questions and investigating the inter-actions between micro- and macro-level factors will highlight patterns and generate more clear and comprehensive responses to the complex issue of NGO insecurity.

Material aid is a risk factor that demands creative responsesResources remain sources of contention in conflict zones, whether they are supporting programming (for example, vehicles) or integral to the programming itself (for instance, food aid, seeds and tools). Obviously material aid covers a wide array of resources of differing value. Simply decreasing the amount of material aid or accompanying equip-ment is not necessarily an appropriate or viable option in an emergency or non-emergency situation. Responding to material aid as a risk factor for organisations active in a variety of sectors requires creativity and imagination. Anderson (1999, p. 41) pro-vides several examples of creative NGO programming to reduce opportunities and incentives to steal, including one NGO that redistributed pink radios to women involved in its public health programme after all of the original radios had been stolen.

‘Operationality’ as a risk factor needs to take account of the distinctive risks and implications for national and expatriate staffAt first glance, one obvious way to address ‘operationality’ as a risk factor is to become non-operational by withdrawing or minimising the number of expatriate staff when the violence escalates. At a deeper level, however, this option is less attractive as a single or dominant response. The idea of ‘programming by remote control’ (Hansen, 2004) raises operational issues about the continuity and quality of programmes, as well as ethical considerations about ‘different value for different life’ (Anderson, 1999) and leaving national staff to continue to bear the burdens of programming and risk. The latter ethical considerations feed into and reinforce critiques of the neo-colonial nature of the humanitarian and development enterprise. National staff members, by virtue of being from the country in which they work, face unique risks that cancel out the protection they may gain through intimate knowledge of the context. In this study, respondents indicated the nationality of the affected individual(s) for 26 of the total incidents (N=66); of these 26, 18 incidents (69 per cent) were directed against a national(s) of the country in question. This statistic highlights the pervasive risk for national staff. A prevalent but changing assumption is that security issues are related to expatriate and not national staff. Even though the original solicitation letter for this study did not specify expatriate staff as the focus of the study (and indeed they were not), the replies of potential and actual respondents indicated that many believed this was the case. An InterAction report on national staff security identifies a similar bias (Inter-Action, 2001), and one study points up that national staff in many organisations were half as likely as international personnel to receive security training (Beasley, Buchanan and Muggah, 2003). This bias is changing, as NGOs begin to pay more attention to the particular security issues confronting national staff. The InterAction report describes

Page 19: Characteristics, context and risk: NGO insecurity in

Larissa Fast148

best practices in security management for national staff (InterAction, 2001) and a November 2004 conference concentrated on how best to support the well-being of national staff working for Western humanitarian organisations.12

Working with one or both sides of a conflict in situations of ongoing violence deserves special analysis and attentionThe counter-intuitive finding that those NGOs working with both sides of the conflict tend to be more insecure alludes to the complexity of operational decisions and their relationship to NGO insecurity. The discussion above offered two possible reasons for the findings related to working with both sides, one having to do with the perceptions of working with both sides (and therefore implicitly helping ‘the other’) and the second related to operating on the frontlines of violence with higher levels of ambient inse-curity. While the first point relates directly to working with both sides, the second reason highlights the possibility that confounding factors, such as proximity to violence and other operational constraints or limitations, generated the association between the variables. In other words, it is plausible that additional factors that the study could not assess or control for are responsible for the correlation between higher insecurity and working with both sides of a conflict. Each of these potential explanations points to the need for further study on this issue and for the creation of a more complex and dynamic model of NGO insecurity. It is not clear from the data whether partiality provided protection for NGOs or whether it isolated them from much of the fighting. The findings do not imply that partiality is an effective way of reducing insecurity, but they do suggest that partial organisations may be less vulnerable to violent attacks. While this partiality may be forced (such as through lack of access), its impact on insecurity and its effects on the role an organisation plays in a particular location are both important considerations for any security management strategy. The security implications of working with only one side suggest possible roles for partial interveners in settings with high levels of ambient violence. That is, if partiality provides some measure of protection in contexts with high levels of ongoing violence, minimising the threats to and the vulnerability of NGOs might involve working only with one side of the conflict, at least until the violence subsides. Obviously, human suffering is never confined to only one side, and such a decision involves difficult moral choices regarding who receives assistance and operational decisions and chal-lenges concerning how to intervene. In some situations, for example, working on the frontlines may not be as dangerous as actually getting there, since the home and work environments tend to be more secure than the routes between them.13 To address these risks, security management or other coordinating bodies could help identify which organisations work on only one side so that all belligerents receive assistance but from different bodies, thus minimising the number of organisations operating across conflict lines. Such decisions should not be taken lightly, as a determi-nation to be partial could compromise or limit the ability of the organisation to work with the other side(s) at a later stage. In some cases, however, the security ramifications

Page 20: Characteristics, context and risk: NGO insecurity in

Characteristics, context and risk: NGO insecurity in conflict zones 149

may warrant such a choice. Security coordinators could track and analyse incidents and tailor security management strategies to tackle working with both sides of a con-flict as a risk factor. The glaring and controversial caveat of this finding is that it provides a disincentive to abide by the traditional humanitarian principles of neutrality and impartiality, con-sequently challenging them and even according further fuel to the argument that these values are outdated in a contemporary context. Some organisations will and should choose to continue to work with any and all sides based solely on need, for reasons related to their mission and guiding principles. The results from this research do not and should not repudiate the value of working with both sides, but instead suggest that these organisations should devote extra attention to security management if and when they opt to cross conflict lines. Furthermore, the finding related to working with both sides appears to contradict conventional wisdom that it is the erosion of impartiality and neutrality that has sub-jected humanitarian actors to increased levels of insecurity. These results, though, are neither ‘thick’ (Geertz, 1983) nor robust enough to suggest causality and disprove conventional wisdom; they imply a correlation between insecurity and working with both sides for this sample population. More research could specify why and how they are related. On a macro level, the principles of impartiality and neutrality are key to operationalising access and establishing the legitimacy of humanitarian action (Plattner, 1996), but myriad factors affect perceptions of impartiality on a micro level as well. For example, the behaviour of individual staff members influences how conflict parties perceive impartiality and neutrality, which impacts on levels of insecurity facing the organisation. Practically, this is where a context analysis that examines how character-istics and perceptions shape the ambient and situational security environment(s) and vice versa could help to mould and tailor security strategies and responses.

More integration into local communities and social networks should be undertaken cautiously in contexts with high levels of ongoing violenceThe findings on integration seem to indicate that incorporation into local commu-nities and social networks may be less effective in lowering levels of insecurity in situations of ongoing war or violence. In terms of security management, integrating into local communities and networks fits within an acceptance strategy, where agen-cies endeavour to establish and maintain ‘political and social “consent” for [their] presence and [their] work in a particular context’ (Van Brabant, 2000, p. 58). Accept-ance strategies are about integration as well as attaining legitimacy and credibility for an organisation’s programmes and activities. In situations of continuing violence, social and political consent will be tied to the broader, global context and not just to one organisation’s activities. Nevertheless, these findings do not imply that the acceptance model is an irrelevant security strategy or that integrated organisations are always more insecure. The purpose of humanitarian NGOs, for instance, is not to integrate into local communities but to provide life-sustaining assistance. Assuming that insecurity is dependent on context,

Page 21: Characteristics, context and risk: NGO insecurity in

Larissa Fast150

security management strategies need to be too. Indeed, some organisations may decide that acceptance and integration are inherently related to their ability to implement an effective programme, or are integral parts of their institutional mission. They may determine that it is worth taking the security risks given the positive effects on their programmes and operations. In these cases, organisations pursuing strategies of accept-ance or integration in situations of ongoing war or violence should think about possible additional and contextualised policies and mechanisms to provide extra security for their assets and staff members.

Standardised reporting mechanisms and more research on this topic would assist in identifying security incident patterns and trendsThis is not a new observation, as others have voiced similar recommendations. Re-searchers and practitioners alike have recognised the need for further research, and this study and those mentioned earlier have begun to fill the gap. Although previous efforts to establish security incident reporting mechanisms have failed (such as the Humani-tarian Security and Protection Network), the UN has issued a Request for Proposals for a Security Incident Reporting System to report and track security incidents. Infor-mation sharing and security coordination at a country level are important for increas-ing the security of all NGOs and international agencies working in a particular country. Standardised reporting, broader information sharing and collaborative research will contribute to identifying patterns and trends across and within countries that will assist in setting and adapting security policies to respond better to the changing realities of working in conflict zones.

AcknowledgementsI wish to thank the following individuals for their suggestions and advice regarding this article and research: George Lopez, Larry Minear, Helen Young and two anonymous Disasters reviewers, Christopher Mitchell, Daniel Druckman, Frances Harbour, and Janice Gross Stein.

CorrespondenceLarissa Fast, Visiting Assistant Professor, Kroc Institute for International Peace Studies, University of Notre Dame, Notre Dame, IN 46556, US. Telephone: +1 574 631 7096; e-mail: [email protected].

Endnotes1 This article is a condensed version of a much longer study. For a more complete description and

discussion of the methodology (including the macro-level contextual variables), limitations and find-ings, see Fast, 2002.

Page 22: Characteristics, context and risk: NGO insecurity in

Characteristics, context and risk: NGO insecurity in conflict zones 151

2 Although the conceptual framework identified 12 potential micro-level variables related to insecurity, I chose only seven of these for further analysis (see Figure 1). To maintain a cases:variables ratio of at least 3:1, I selected those variables most likely to affect the organisation’s level of insecurity.

3 Obviously, this definition of impartiality differs from its traditional meaning in humanitarian action. The sample included NGOs with a variety of mandates, not solely humanitarian actors, meaning that no common definition of ‘impartiality’ existed within the sample population. Consequently, the research defined impartiality in terms of ‘working with both sides of a conflict’, a denotation more akin to the humanitarian principle of neutrality. Nevertheless, working with both sides of a conflict is a conceivable outcome of the principle of impartiality, which specifies distribution of resources exclusively on the basis of need (Slim, 1997). The survey itself made no mention of impartiality, referring instead to ‘work-ing with both sides of a conflict’. Fast (2002) referred to the variable as ‘impartiality’. For clarity, this article discusses the results for this variable with reference to ‘working with both sides of a conflict’. See the section on the study’s findings for further discussion of this issue.

4 Based upon this sample population, I sent solicitation letters to approximately 180 organisations, of which 45 were not working in any of the countries during the period designated for a particular country, or never had programmes in any of these countries. Those organisations that agreed to take part or expressed further interest received a participation packet including consent forms and the survey for the appropriate country/countries. In total, I sent participation packets to 47 organisations (for a total of 72 surveys), of which 20 (42.5 per cent) completed the survey(s) (for a total of 27 surveys).

5 I wish to thank the following organisations for their involvement: Africare (US); Care International (Canada and US); The Carter Center (US); Catholic Relief Services (US); Christian Children’s Fund (US); Christian Extension Services/CRWRC (Sierra Leone and US); Christian Reformed World Relief Committee (US); Church World Service (US); Compassion International (US); Concern World-wide (Ireland); Danish Refugee Council (Denmark); Development Workshop (Canada); FORUT (Norway and Sierra Leone); GOAL (Ireland); International Medical Corps (US); International Rescue Committee (US); The Nature Conservancy (USA); OIC International (US); Planning Assistance (US); and SOS-Kinderdorf International (Austria). Ten other individuals provided valuable information and insights in the form of interviews during the first phase of the research process (see Fast (2002) for a complete list).

6 I make no claims in this study that an organisation that experienced at least two high incidents (or four medium incidents or 10 low incidents or any combination thereof) is highly insecure. This dis-tinction represents a reasonable decision about an appropriate threshold between more and less insecure organisations in this sample.

7 King (2002) credits Jan Davis, a security expert with Registered Engineers for Disaster Relief (RedR), with recognising a potential reporting bias that would skew results in favour of security over safety incidents. Individuals may be less likely to report safety incidents (such as road accidents) for publicity and other reasons. Barnett (2004, pp. 14–23) highlights definitional issues (for example, a lack of com-mon definitions, especially in terms of road accidents) that complicate an accurate assessment of these claims.

8 The study made no distinction between obviously targeted (intentional) violence and those acts of violence for which motivation or intention was unclear.

9 As explained above, I designed the study to examine the importance of the country context on levels of NGO insecurity. I chose the time frames to maximise the similarities between the three cases. Angola and Sierra Leone experienced more ambient insecurity simply because of the ongoing wars in those countries during the periods of study. In contrast, Ecuador had a lower level of ambient insecurity.

10 See Table 2 for statistical results. For NGOs in Angola and Sierra Leone, this ‘working with both sides’ variable was highly correlated with both insecurity score and level of insecurity, but these same vari-ables were not correlated when NGOs in Ecuador were added. The intriguing aspect of this relationship

Page 23: Characteristics, context and risk: NGO insecurity in

Larissa Fast152

was the negative value for the Spearman Rank Order Correlation Coefficient, meaning the variables were discordant. Thus, higher insecurity scores were correlated with the lower values for ‘working with both sides’, which specified working with only one side of the conflict.

11 The total number of incidents was 66, but only 23 specified the gender of the affected individual(s). Thus, the number of incidents involving females could have been higher in the actual sample.

12 The conference was titled ‘Cross-Cultural Perspectives on Psychosocial Issues of Humanitarian Staff Care’ and was held in Melbourne, Australia, on 16−17 November 2004. For more information, see http://groups.psychology.org.au/Assets/Files/mitigating_stress.pdf.

13 I owe credit for this observation to Antonio Donini, Senior Researcher at the Feinstein International Center, who made a similar remark to me based on his experiences in Afghanistan.

ReferencesAbbott, M. (2005) ‘Random or targeted violence? The problem of humanitarian aid worker security’.

Paper presented at the 46th Annual Convention of the International Studies Association, 2 March, Hilton Hawaiian Village, Honolulu, HI.

African Rights (1994) Humanitarianism unbound? Current dilemmas facing multi-mandate relief operations in political emergencies. African Rights, London.

Anderson, M.B. (1999) Do no harm: How aid can support peace—or war. Lynne Rienner, Boulder, CO.Barnett, K. (2004) Security report for humanitarian organizations: Directorate-General for Humanitarian Aid,

European Commission Humanitarian Aid Office (ECHO), Brussels.Beasley, R., C. Buchanan and R. Muggah (2003) In the line of fire: Surveying the perceptions of humanitarian

and development personnel of the impacts of small arms and light weapons. Centre for Humanitarian Dialogue and Small Arms Survey, Geneva.

Bergman, C. (ed.) (2003) Another day in paradise: International humanitarian workers tell their stories. Orbis Books, Maryknoll, NY.

Bickley, S. (2003) Safety first: A field security handbook for NGOs (second edition). Save the Children Fund, London.

Bollettino, V. (2003) ‘Death in the field: How UNICEF is confronting the toll of violence’. Humanitarian Affairs Review. Spring. pp. 8−11.

Buchanan, C. and R. Muggah (2005) No relief: Surveying the effects of gun violence on humanitarian and develop-ment personnel. Centre for Humanitarian Dialogue and Small Arms Survey, Geneva.

Danieli, Y. (ed.) (2002) Sharing the front line and the back hills: Peacekeepers, humanitarian aid workers and the media in the midst of crisis. Baywood Publishing Company, Amityville, NY.

Devine, P.G. (1995) ‘Prejudice and out-group perception’. In A. Tesser (ed.) Advanced social psychology. McGraw-Hill, New York, NY. pp. 467–512.

Donini, A. (2003). ‘Issues Note’. Paper presented at ‘The Future of Humanitarian Action: Implications of Iraq and Other Recent Crises’. Brainstorming workshop organised by the Feinstein International Famine Center, Friedman School of Nutrition Science and Policy, Tufts University, 9 October, Boston, MA. http://hwproject.tufts.edu/pdf/iraq_issues_20031022.pdf.

Dworken, J. (1998) Where there is no data: Patterns, trends, and unanswered questions concerning relief worker secu-rity problems. Center for Naval Analyses, Alexandria, VA.

Fast, L.A. (2002) Context matters: Identifying micro- and macro-level factors contributing to NGO insecurity. Un-published PhD dissertation, Institute for Conflict Analysis and Resolution, George Mason University, Fairfax, VA.

Faure, A.M. (1994) ‘Some methodological problems in comparative politics’. Journal of Theoretical Politics. 6(3). pp. 307−322.

Geertz, C. (1983) ‘“Thick description”: Toward an interpretive theory of culture’. In R. Emerson (ed.) Contemporary field research, Little, Brown & Co. Inc., Boston, MA. pp. 37–59.

Page 24: Characteristics, context and risk: NGO insecurity in

Characteristics, context and risk: NGO insecurity in conflict zones 153

George, A. (1979) ‘Case studies and theory development: The method of structured, focused comparison’. In P.G. Lauren (ed.) Diplomacy: New approaches in history, theory and policy, The Free Press, New York, NY. pp. 43–68.

Hansen, G. (2004) ‘Humanitarian action in Iraq—Emerging constraints and challenges’. Humanitarianism and War Project, Tufts University, Somerville, MA. http://hwproject.tufts.edu/new/pdf/Hansen_ report_Iraq_final.pdf.

Helton, A.C. and G. Loescher (2003) ‘NGOs and governments in a new humanitarian landscape’. 24 June. http://www.opendemocracy.net/articles/ViewPopUpArticle.jsp?id=2&articleId=1307.

ICRC (International Committee of the Red Cross) (1998) Respect for and protection of the personnel of humani-tarian organizations. ICRC, Geneva.

Independent Panel (2003) Report of the Independent Panel on the Safety and Security of UN Personnel in Iraq. United Nations, New York, NY.

InterAction (2001) The security of national staff: Towards good practices. InterAction, Washington, DC.King, D. (2002) Paying the ultimate price: Analysis of the deaths of humanitarian aid workers (1997−2001). 15 January.

http://reliefweb.int/symposium/PayingUltimatePrice97-01.html.King, D. (2004) The year of living dangerously: Attacks on humanitarian workers in 2003. Humanitarian Informa-

tion Unit, US Department of State, Washington, DC.Lee, R.M. (1995) Dangerous fieldwork. Qualitative Research Methods Series. Vol. 34. Sage Publishers,

Thousand Oaks, CA.Macrae, J. and A. Harmer (2003) Humanitarian action and ‘the global war on terror’: A review of trends and issues.

Humanitarian Practice Group, Overseas Development Institute, London.Martin, R. (1999) ‘NGO field security’. Forced Migration Review. April. pp. 4−7. Mayhew, B. (2004) Generic security guide for humanitarian organizations: Directorate-General for Humanitarian

Aid. ECHO, Brussels. Muggah, R. (2001) Perceptions of small arms availability and use among Oxfam-GB field personnel. Small Arms

Survey and Oxfam-GB, Geneva.Muggah, R. and E. Berman (2001) Humanitarianism under threat: The humanitarian impacts of small arms and

light weapons. UN Interagency Standing Committee, New York, NY.Peytremann, I., M. Baduraux, S. O’Donovan and L. Loutan (2001) ‘Medical evacuations and fatalities of

United Nations High Commissioner for Refugees field employees’. Journal of Travel Medicine. 8(3). pp. 117−121.

Plattner, D. (1996) ‘ICRC neutrality and neutrality in humanitarian assistance’. International Review of the Red Cross. 311. pp. 161−179.

Reynolds, P. (2003) ‘Why the UN is a target’. BBC News. 19 August.Roberts, D.L. (1999) Staying alive: Safety and security guidelines for humanitarian volunteers in conflict areas.

ICRC, Geneva.Rogers, C. and B. Sytsma (1998) A shield about me: Safety awareness for World Vision staff. World Vision,

Geneva and Monrovia, CA.Rowley, E. (2005) ‘Mortality and morbidity among humanitarian workers’. Paper presented at the 46th

Annual Convention of the International Studies Association, 2 March, Hilton Hawaiian Village, Hono-lulu, HI.

Seet, B. and G. Burnham (2000) ‘Fatality trends in United Nations peacekeeping operations, 1948−1998’. JAMA: Journal of the American Medical Association. 284(5). pp. 598−603.

Sheik, M. et al. (2000) ‘Deaths among humanitarian workers’. British Medical Journal. 321(7254). pp. 166−168.

Slim, H. (1997) ‘Relief agencies and moral standing in war: Principles of humanity, neutrality, impartial-ity and solidarity’. Development in Practice. 7 (4 November). pp. 342−352.

UN (United Nations) (2005) ‘Safety and security of humanitarian personnel and protection of United Nations personnel: Report of the Secretary-General’. A/60/223. UN General Assembly, New York, NY. http://daccessdds.un.org/doc/UNDOC/GEN/N05/459/59/PDF/N0545959.pdf?OpenElement.

Page 25: Characteristics, context and risk: NGO insecurity in

Larissa Fast154

Van Brabant, K. (1998) ‘Cool ground for aid providers: Towards better security management in aid agen-cies’. Disasters. 22(2). pp. 109−125.

Van Brabant, K. (2000) Operational security management in violent environments. Humanitarian Practice Network, London.