Appendix 1, Statistical Report
Statistical Appendix to the Report of the Truth and Reconciliation Commission of Sierra Leone
A Report by the Benetech Human Rights Data Analysis Group to the Truth and Reconciliation Commission
5 October 2004
The materials contained herein represent the opinions of the authors and editors and should not be construed to be the view of the Benetech Initiative, any constituent projects, the Benetech Board of Directors or the donors to Benetech.
Copyright (c) 2004 Human Rights Data Analysis Group, Beneficent Technology, Inc., part of The Benetech Initiative.
480 S. California Ave., Suite 201
Palo Alto, CA 94306-1609
Certain rights are granted under the Creative Commons “Attribution-NonCommercial- ShareAlike” license, available on the web at: http://creativecommons.org/licenses/by-nc-sa/1.0/legalcode
The license terms are summarized here:
Attribution: The licensor permits others to copy, distribute, display, and perform the work. In return, licensees must give the original author credit.
Noncommercial: The licensor permits others to copy, distribute, display, and perform the work. In return, licensees may not use the work for commercial purposes – unless they get the licensor’s permission.
Share Alike: The licensor permits others to distribute derivative works only under a license identical to the one that governs the licensor’s work.
The Benetech Initiative
Tel: +1 650-475-5440
Fax: +1 650-475-1066
To fulfill its mandate, the Sierra Leone Truth and Reconciliation Commission (TRC) collected as many statements as possible from the victims, witnesses and perpetrators of human rights violations committed during the 1991–2000 period of conflict in Sierra Leone. The Commission collected 7,706 statements of Sierra Leoneans, living in Sierra Leone and also as refugees in Gambia, Guinea and Nigeria. The statements they gave offer detailed insight into the experience of particular victims or perpetrators, and every statement therefore deserves careful study.
It is also valuable to study what the statements can mean in the aggregate. This means to extract information from the TRC statements about each of the human rights violations they document, enter this information into a database, and develop statistics that describe the nature and extent of the violations experienced and perpetrated by the statement-givers as a whole. The resulting dataset enables an overview of the nature and extent of human rights violations experienced during the conflict.
The analyses presented here reviews the broad dimensions of data available from the the TRC’s database. In a general sense, the analysis is guided by the overall research questions the Commission was charged to investigate, as well as specific questions posed by TRC researchers. However, this section does not offer original interpretation of what the graphs and tables mean — that analysis has already been presented in the main body of the report.
Instead, this appendix simply offers the interested reader additional detail about the statistical findings available in the database. It is more like a statistical abstract than it is like an independent report. In a very real sense, this chapter is an invitation to historians, journalists, social scientists and others to pursue further quantitative inquiry by downloading the TRC’s statistical dataset. The statistical dataset is available on the Internet at http://www.hrdag.org. All of the personal information about victims and statement-givers has been removed from the published dataset, but the dataset offers a rich resource for continuing analysis of statistical patterns of human rights violations in Sierra Leone documented by the TRC.
In the first two sections of this appendix, we describe the background and methodology for the processing, entry, and storage of the information contained in the TRC statements. We also explain the concept and scope of special coding studies that were conducted when the standard information coded from the statements was insufficient to answer certain questions, or when a particular aspect of the conflict merited closer study. This section concludes with notes about the nature of the TRC’s sample and the limits of the statistical interpretation.
The third section presents a descriptive analysis of the statistical patterns in the statements given to the Commission. The section examines the demographic patterns of the statement- givers, patterns of different types of violations over time and space, patterns in the age and sex of the victims, and the relationship of different perpetrator groups to these dimensions.
The fourth section examines the study of redress and reparations.
Background and Methodology
The conflict in Sierra Leone began in March 1991. The number of warring factions proliferated with the emergence of civil militias, employment of international mercenaries, regional and international interventions, military coups at home, and incursions by foreign soldiers and irregulars. While initially confined to the South and East, the conflict eventually engulfed the entire country, culminating in an attack of the capital Freetown by the Armed Forces Revolutionary Council (AFRC), in January 1999. Where previous attempts to broker peace had failed, the 7 July 1999 Lome Peace Agreement succeeded, and included a clause allowing for the creation of the Truth and Reconciliation Commission1. Due to resurgent violence in May 2001, the Commission’s work did not begin until the latter half of 2002.
One of the first phases of the Commission’s work was to gather as many victim and perpetrator statements as was possible given the time and funding constraints. While not every victim or perpetrator was interviewed, the statement-takers tried to be as comprehensive as possible, attempting to reach every chiefdom in Sierra Leone in order to record the experiences of the population, including experiences of specific groups such as women, children and amputees. Because of security and accessibility issues, 9 of the 149 districts in Sierra Leone were not reached for interviews.
From Figure 4.A1.1a, it is clear that there were substantial numbers of statements taken across Sierra Leone and neighboring countries. Women gave approximately one-third of the statements, while men gave approximately two-thirds.
Both male and female deponents gave statements with roughly equal proportions of motivations. Males were slightly more frequently direct victims of violations, while females were similarly slightly more likely to be witnesses to violence against family members.
Male deponents are slightly older than female deponents, as Figure 4.A1.1c shows. A higher proportion of female deponents than male deponents are in each of the age categories up to age 45–49. So, for example, while 8.5% of female deponents were of ages 15-19, 5.8% of male deponents were 15–19 years old. However, 8.0% of male deponents were 60-64, while 5.4% of female deponents were in this category.
The largest ethnic group among the statement-givers were the Mende, with 44.3% of all deponents coming from this group. A smaller but substantial number — 20.5% — of deponents came from the Temne, while smaller numbers of statements were given by members of other groups.
Statement-taking was completed in March 2003 with 7,706 human rights narratives collected. Subsequently the statements were coded, so that the victims, perpetrators and abuses in each statement were identified and listed on forms in accordance with the selected data model, which is described below. When coding was complete, the coded statements were entered into a database designed specifically to capture this information while preserving the relationships between the perpetrators, victims, and abuses given in the statements.
The model adopted by the Commission was based on the concepts in “Who Did What to Whom”.2 This data model is designed to account for the fact that a data source, such as a collection of statements, can include information about one or many victims and/or perpetrators, and each victim can suffer one or many human rights violations. It is a model that has been used to provide statistical results presented by other truth commissions and human rights documentation projects, including the truth commissions of Guatemala, Haiti, South Africa, Perú, and East Timor.
Perpetrators were classified as follows:
RUF Revolutionary United Front
AFRC Armed Forces Revolutionary Council including Westside Boys
SLA Sierra Leone Army
CDF Civil Defense Force
ECOMOG Economic Community of West African States Military Observer
GAF Guinean Armed Forces
ULIMO United Liberation Movement for Democracy
Police Police officers including SSD division
AFRC/SLA Abuses committed in 1997 allegedly committed by soldiers but the date information is insufficient to determine if the abuses should be attributed to the SLA or the AFRC
Miscellaneous Minor perpetrator groups
Rebels Abuses attributed to rebels where the statement-giver was unable to name a specific faction. Typically the term describes RUF fighters and ex-SLA fighters loyal to the AFRC
The TRC statements were coded into fourteen violation types using a controlled vocabulary set in order to apply standard definitions in a consistent manner. The violation types and the abbreviations used for them in tables in this appendix are as follows:
DETN Arbitrary Detention
DEST Destruction of Property
CANN Forced Cannibalism
FODI Forced Displacement
FOLA Forced Labour
FREC Forced Recruitment
TORT Physical Torture
SXAB Sexual Abuse
SXSL Sexual Slavery
After all of the coded statements were recorded in the database, the data underwent a matching procedure. Many statements identified people and events that were also identified in other statements. In order to count each violation only once, we identified which people and violations were reported more than once — the process is called “matching” — and we counted them appropriately. To prepare for matching, analysts looked for discrepancies in the data that may have been a result of coding or data entry errors. Changes made to the database were catalogued to determine if the original data was preserved or not in case the corrections themselves were applied incorrectly.
We matched the corrected data by looking at the victim’s name, age, ethnicity, and sex. Taking into account the potential for spelling variations and data entry errors, matches were considered where fields were the same or relatively similar. The acceptable tolerance for age differences was ±3 years. Where age or name fields were empty, they were considered acceptable to match the record to another record (if the non-missing fields matched). While this practice may have missed some matches because witnesses’ memories of dates was not precise, it avoided overmatching records of individuals with the same name. Location information was also used to make judgments about whether or not records reported the same victim, perpetrator and act. Tolerances for distance were kept to small areas within a district to also prevent overmatching of records.
The final result of these steps — coding, data entry, and matching — is the database from which the Commission’s statistics were calculated. The final table from which the Commission’s statistics are generated contains 40,242 violations.3
Special Coding Exercises
On a number of occasions, TRC researchers asked questions that were beyond the scope of the information quantified via the standard statement coding. Also, the results from the conventional coding occasionally suggested aspects of the conflict that merited further, more detailed research. To deal with these situations, a series of special coding analyses were devised:
• ECOMOG (Economic Community of West Africa Military Observer Group) Abuses Study
3For more detail on the creation of the TRC database, see Volume 1, Methodology and Processes Chapter of the Final Report of the Sierra Leone Truth and Reconciliation Commission.
• RUF–NPFL (National Patriotic Front of Liberia) Study
• Assistance and Redress Study
ECOMOG Abuses Study
The ECOMOG intervention force was distinct in that the abuses attributed to it in the statements had a relatively high proportion of killings. The special coding study considered the nature of these killing violations and why the ECOMOG behaviour was distinct.
It is widely believed that the initial RUF incursion into Sierra Leone in 1991 included forces from the Liberian NPFL.4 The special coding looked at the ethnicity of the perpetrators in statements identifying the RUF in the early years of the conflict. This information was used to determine the years in which Liberian forces were committing violations in Sierra Leone and the proportion of RUF abuses that could more properly be attributed to the NPFL.
Assistance and Redress Study
The TRC statements contain a number of questions designed to elicit information on the current circumstances and attitudes of victims and perpetrators, and the forms of assistance from which they, their families, their community, or society as a whole might benefit. This special coding study considered these questions primarily focusing on reparations and reconciliation.
Each of these studies were done with a subset of the TRC statements. The main database was used to select the study statements according to specific criteria. Where possible, all applicable statements were used. If the number of statements was more than could be coded in the time available, the analysis was limited to a random sample of the collected statements.
These studies were done during various stages of the main data entry task. This means that the analyses are representative of the statements entered into the database at that time. Because the statements were entered into the main database in a random order, the special coding study results can be considered as representative of the TRC statement collection as a whole, within the calculated margin of error.
For all studies, the coding aimed to avoid any possibility of bias or exaggeration. Any assumptions made by the coders tended to the more cautious option.
The specific methodology and results of each study are presented in various sections of this report.
Notes about the nature of the sample
Due to the fact that the TRC database represents neither a complete census of human rights violations nor a random sample of these violations, conclusions drawn from this analysis may only apply to the database and not to the general population. Each statistical argument in the report must therefore be understood as “according to statements presented to the Commission,...”
An analysis of the contents of the database indicates the type, and to some degree, the extent of violations. In some cases, the data on certain violations was not sufficient to analyze the patterns (over time, space, perpetrator, or type of victim) for that violation type; forced drugging and forced cannibalism are the violations for which the data are inadequate.
The TRC statement-takers attempted to complete a census of the human rights violations experienced during the conflict, locating and recording the statements of as many victims as possible. According to clause 6 of the Peace Agreement, the principal function of the Commission is to “create an impartial historical record of the events in question.” As such, they strove to take statements in areas that they knew were the sites of severe or numerous violations. It was the intention of the statement-takers to visit every chiefdom in Sierra Leone. Although this target was not attained, interviews were taken in 141 of the 149 chiefdoms as well as in Gambia, Guinea, and Nigeria where refugees from Sierra Leone were living.
Due to a combination of factors, the district of Port Loko in the Northern Province was under- sampled, with the staff taking relatively few statements in its chiefdoms, compared to other districts. Statement-taking in the Western Region was concentrated in Freetown. Furthermore, sexual violations were almost certainly under-reported, and violations for which no witnesses remain could not have been captured by the TRC data collection process. These problems notwithstanding, the Commission’s sample is so large that it represents the experiences of a substantial pool of people, men and women from all of Sierra Leone’s ethnicities, geographically distributed across Sierra Leone.
We do not expect the proportions derived from the database to be precise measurements of the violations suffered by the people of Sierra Leone. There are several limitations on how these data can be interpreted. First, the Commission’s database is not a random sample. Percentages calculated from the Commission’s database cannot be assumed to represent percentages among the population of Sierra Leone more generally. There is no sampling error associated with these calculations. The imprecision associated with the proportions derived from the database is due first to who chose to respond when Commission interviewers invited them to make statements. Other potential statement-givers chose not to speak with the Commission. Other errors include intentional or unintentional inaccuracies in the testimonies provided by the statement-givers, data recording mistakes, data coding mistakes, and data entry mistakes. Direct measurement of these various errors is not possible and estimation of this error is very difficult. For these reasons, creating a margin of error for these statistics using an assumption of simple or complex sampling error would be misleading. We therefore only include margins of error for statistics created from data collected via the special coding exercises. Our assumption in those cases is that these margins of error represent the accuracy of the statistics as they represent all the statements given to the Commission.
To conclude, the statistical findings in this and the other chapters of the Commissions report should be understood as representing the statements provided to the Commission.
Exploratory Data Analysis
There are several ways to count the number of violations in the TRC database. The highest- level unit is a statement. The statement-giver can describe one or more victims, each of whom may suffer one or more violations. Note that each victim may suffer several violations, including the same violation more than once (except killing). Each victim who suffers a particular violation is counted once in the statistical descriptions that follow.
Figure 4.A1.2 shows for each type of violation, the number and proportion of violations, the proportion of victims for that violation type, and a ratio of violations to victims documented in the TRC’s Database. Forced displacement and abduction are the most common violations in the Commission’s database, at 19.8% (7983/40242) and 14.8% (5968/40242), respectively. Together with the third highest violation type, arbitrary detention at 12% (4835/40242), these three violations make up nearly half of all documented violations. Killing and destruction of property follow at 11.2% (4514/40242) and 8.5% (3404/40242), respectively.
There are 14,995 victims in the TRC database. The proportion of victims who suffer each violation sums to much more than 100% because each victim could suffer more than one kind of violation. They might also suffer the same violation more than once. The ratio column shows that for most violation categories the ratio of violations to victims falls between 1 and 1.12, while the ratio of violations to victims for the forced displacement category is 1.28. This means that victims who suffer forced displacement tend to suffer, on average, a higher number of forced displacements each.
The statements indicated various reasons for forced displacements; some take flight out of fear, anticipating an attack, while others are obliged to leave because an armed faction has destroyed their home.
It is known that the act of abduction was over-coded; more abductions are listed in the database than actually were recorded in the statements received by the Commission. Originally abduction was intended to indicate that the victim was kidnapped and taken to another location under the control of the perpetrators. Misinterpretation by the coders led to abduction being coded for other instances in which the victims are at the mercy of the perpetrators, for example when stopped at a checkpoint or caught in an ambush.
Figure 4.A1.3 is a plot of the total number of documented violations over time. The TRC in its Military and Political History of the Conflict Chapter defines the first phase as March 1991 to November 1993, which covers the initial RUF and NPFL assault, primarily in Kailahun, Pujehun, Kenema, Bo, and later Kono. The second phase, from November 1993 to March 1997, reflects the second major RUF assault in which the faction was active in all districts except the Western Area. The third phase, from March 1997 to the end of the conflict, considers the most complex period of activity. It encompasses the formation of the AFRC military government and the rise of the Kamajor militia in the South. The TRC also considers the year 2000 resurgence and demise of the RUF.
Because of the incompleteness of the date information in the TRC database, the phases in this chapter have been rounded to the nearest year. The first phase is 1991 to 1993 inclusive, the second phase is 1994 to 1996 inclusive, and the third phase is 1997 to 2000.
Figure 4.A1.3 shows that the conflict was episodic; the majority of violations occur around three specific periods or episodes of violence. The level of violations was not constant during the conflict. Note that the peak in the first phase occurs in 1991 at the beginning of the conflict. In the second phase the peak is 1995 during a major RUF assault, and the third phase represents the invasion of Freetown in 1999.
The number of violations, broken down by type and by year is shown in Figure 4.A1.4. The number of violations of every violation type follow the overall peaks in documented violations in 1991, 1995 and 1998 demonstrated in Figure 4.A1.3. The greatest peak occurs in forced displacements in 1998.
Figure 4.A1.4 also shows the overall increase of violations in almost every violation type over time. With each of the episodic increases (in 1991, 1995, and 1998–1999), for most violations, the peaks grow larger. Forced displacement is perhaps most responsive to increases in broad levels of violence, as it both rises and falls at a higher proportional rate than other violation types.
Killing is an exception. Documented killings are at their maximum in 1991 (713), which is much higher than the next highest year for documented killing violations in 1995 at 573. Sexual abuse violations also peak in 1991 with 89 documented violations. This number is closer to the totals documented in other years such as 1998 and 1999 at 72 and 85 respectively.5
Violations over space
The Commission’s database cannot be used directly to analyze patterns of human rights violations with respect to space. We can look at the different proportions of human rights violations in the database, but as explained in the introduction to this appendix, this information will represent the proportions of these violations in the actual districts only as well as the statements given to the Commission represent the experiences of all the people in these districts. If people in some districts felt especially uncomfortable with the Commission, then fewer people from those districts would have come to the Commission relative to other districts where more people trusted the Commission. There are patterns across districts which seem consistent with hypotheses advanced on the basis of qualitative arguments elsewhere in the report. It is for this purpose that the following tables are presented.
Figure 4.A1.5 shows the counts of each violation type by district/region. The district with the greatest number of recorded violations is Bo. Kenema and Pujehun have similarly high levels of abuses relative to the other districts. In all three districts, it is forced displacement and abduction that are the main components of the counts. In Bo and Kenema, there are also a relatively large numbers of killings in comparison to other districts.
Figure 4.A1.6 shows the proportion of each violation type by district, where 100% equals the number of violations documented in Figure 4.A1.7. The proportions of the various violation types are roughly consistent across the various districts.
Violence moves through Sierra Leone in distinct waves. In 1991, violence is concentrated in the East and South. Violations decline in the South but continue in the East through 1994. The mid-decade surge in violations starts in Bo and in the East, but by 1995, the patterns of violence are dominated by districts in the Southern Region. In the later period 1998–1999, violence is at its worst in the West and North. The regional-temporal patterns are interesting for two reasons. First, it is useful to observe that violations were reported in all districts, but information from different districts tended to cover different periods. Second, we note that every district of Sierra Leone was substantially affected by the war’s violence. Freetown, which was largely unaffected until 1999, is the site of the war’s most intense attacks in January 1999.
Patterns of documented violations by victim characteristics
Many of the hypotheses considered by the Commission’s researchers posited whether there were systematic campaigns against women, children, or people of certain ethnic groups. This section examines statistical patterns over these social dimensions.
This analysis presented here includes only victims for whom the age at time of the violation is known. Of the 40,242 total violations reported to the Commission, 22,041 have the exact age of the victims documented. Although the findings presented here might be weakened by the inclusion of all the ages (if they were known), this effect cannot be assessed with the existing data. Using internationally accepted definitions, the Commission considers a person under the age of 18 to be a child. The majority, 82% (18040/22041) of the documented violations where the victim’s age is known to the Commission database are perpetrated against adults. A smaller proportion of violations, 18% (4001/22041) were perpetrated against children age 17 and under. There were 18,201 violations with the age missing.
There are 40,103 documented violations in the Commission’s database for which sex of the victim is known. Of these violations 33% (13038/40103) are committed against females and 67% (27065/40103) are committed against males; 139 violations did not have the victim’s sex recorded. These violations represent the experiences of 14,995 victims; 33% (4931) of these victims are female and 67% (9993) of these victims are male.
There are 3,995 (out of 4001) documented violations against children where the sex of the victim is known. Of these violations, 48% (1923) are against girls and 52% (2072) are against boys, with 6 child victims whose sex is unknown to the witness. In contrast, of the 18,040 documented violations against adults where the sex of the victim is known, 29.2% (5272) are against women and 70.6% (12737) are against men. The total numbers of documented violations against girls and boys are nearly equal, while in contrast, the number of documented violations against women is less than half the number of documented violations against men. In short, adult victims tend to be men, while children victims are approximately equally likely to be boys or girls. This pattern will be considered in more detail in the sections below.
Males and females do not suffer the same kinds of violations. In Figure 4.A1.8, it is clear that many violations follow the general 1/3 female : 2/3 male pattern (forced displacement, abduction, assault). Other violations are suffered exclusively by female victims (rape, sexual slavery), and some violations are overwhelmingly perpetrated against male victims (e.g., forced recruitment, forced labour, killing).
In Figure 4.A1.9, below, we examine the count of documented violations by year and the sex of the victims. On average, there are approximate 2 violations suffered by male victims for each violation suffered by female victims. This pattern is only roughly consistent over time, with some variation, from a high of 2.66 in 2000 to a low of 1.64 in 1992.
The number of reported violations against women follow basically the same pattern as violations against men, peaking in 1991, 1995 and 1998–99. The worst year for women, 1999, is the third worst year for men, trailing far after 1998 and more closely after 1991.
As before, a high ratio in Figure 4.A1.10 indicates that relatively more of the victims are males, while a low ratio indicates that relatively more of the victims are females. There is a similar pattern among districts, where the male to female ratio varies from a low of 1.32 in the Western district to a high of 2.82–2.86 in districts in the Northern region. The Western district (containing Freetown) has the relatively highest proportion of female victims of any district. With the existing data, it is impossible to determine whether the relatively larger number of female victims in the Western district is the result of more women recounting their stories in this part of Sierra Leone, or whether this pattern shows that a truly higher proportion of the victims in Freetown were women.
In Figure 4.A1.11 Shows the number of documented violations by perpetrator and sex, including a ratio of male to female. Especially interesting is the column of ratios and what it tells us about the proclivity of the various perpetrator groups to target abuses against women.
The ratio for the Police of 4.54 means that for every 4.54 documented violations targeted against men by the Police, only one violation is targeted against women. In contrast, for the RUF, for every 1.96 documented committed against male victims, one violation is committed against female victims. In other words, the RUF is more than twice as likely to commit a violation with a female victim than are the Police. The pattern of the AFRC is similar to the RUF, while most of the other perpetrator groups fall in between.
Together, the top four perpetrator groups along with rebels make up over 90% of all of the documented violations against women where the sex of the victim is known. The RUF bears the majority of the responsibility, attributed with 62% (8208/13202) of the total number of violations against females.
Patterns by Victims’ Age
The analysis of age will first present summary statistics by type of violation, time, space, and perpetrator. More finely disaggregated analyses of age and sex by type of violation will then be presented in a series of graphs.
The counts specific violations suffered by adults and children are given in Figure 4.A1.12. Among the victims with ages known to the Commission, the relationships between adult and child victims for some violations are logical. For example, the violations involving property (destruction, extortion, looting) are overwhelmingly adult violations. Other age patterns reflect the particular focus of some violations on children: forced recruitment and sexual slavery are majority child, and rape is nearly equally divided between adult and child victims.
Figure 4.A1.13 offers another way of examining the age distribution for each violation type. Note that “Min” stands for the minimum age, “Q1” stands for the age at which 25% of the cases are that age or younger, “Median” means the age at which 50% of the cases are that age or younger, “Mean” stands for the average age of the victims for that violation type, “Q3” stands for the age at which 75% of the cases are that age or younger, and “Max” stands for the maximum age. The “Missing” column gives the percent of all violations for which the age of the victim is unknown to the Commission.
Looking at individual violation types in Figures 4.A1.12 and 4.A1.13 perpetrated against adults and children we find that for documented amputation, assaults/beating, destruction of property, extortion, forced displacement, killing, looting of goods, physical torture, and sexual abuse violations, the distribution of age of victim is solidly centered on adults.
The results in Figure 4.A1.21 demonstrate that documented victims of forced recruitment, sexual slavery and rape were younger than the other violation types. Specifically, the following conclusions can be drawn:
• 50% of the victims of forced recruitment with age documented were 14 years of age or younger when they were forcibly recruited.
• 25% of rape victims with age documented were 13 years of age or younger.
• 50% of sexual slaves with age documented were children age 15 or under when they were abducted.
• 25% of the victims of forced recruitment with age documented were 11 years of age or younger when they were abducted.
The next analysis considers the patterns of victims’ ages over time.
Figure 4.A1.14 shows the counts of violations against adults and children by year. It is striking in this table that the ratio of adults to children tends to decline over time: the highest ratio (indicating the largest number of adults suffering relative to each child) is in 1991, and the lowest is in 2000. This trend briefly reverses in 1996, a year during which the conflict is relatively moderate. However, after the reversal, the trend returns to relatively more child victims per adult victim.
There are some surprises in the relative numbers of children and adult victims shown in Figure 4.A1.15. The ratio between adults and children varies widely, from 9.58 to 2.85. By substantial margins, Pujehun (9.58) and Kenema (7.73) have relatively fewer violations against children than other districts, while Kono (2.85) has relatively more child victims per adult. The variation is shown in more detail below, in Figure 4.A1.16.
Kono stands out as having the highest proportion of documented violations suffered by female children. The Western Area has relatively more adult females suffering violations, and relatively fewer adult males than other districts.
In In Figure 4.A1.17, it can be seen that relative to other perpetrator groups, the RUF and the AFRC have different victim profiles with respect to age category. While the ratio of adult to child victims is 3.89–4.65 for these two groups, for the SLA and CDF the ratios are more than double at 9.14 and 11.83, respectively. This means, for example, that for every 3.89 violations the RUF allegedly committed against an adult, they committed one against a child. Whereas the SLA committed one violation against a child for every 11.83 violations committed against an adult.
Patterns by Victims’ Age and Sex
This section combines the analysis of the previous two sections. By considering the distribution of victims’ age and sex simultaneously, this analysis can unpack the broad age categories in the previous section to show the specific ages that suffered each violation. At the same time, the analysis shows how each violation affected males or females at different ages.
All of the analysis here could be considered in terms of the population rates of each violation’s occurrence. That is, the counts of each violation for each age and sex category could be divided into the total number of Sierra Leoneans of that age and sex. The resulting figures can then be compared across different age and sex categories, simultaneously considering both the count of the violations and the age and sex distribution of the population. Analysis of this kind was presented in the Children’s Chapter in the discussion of rape, sexual slavery, and forced recruitment. For simplicity, the data are presented here as simple counts.
The first group of graphs considers violations against property and the freedom to live in security.
These six graphs encompass destruction of property, extortion, looting of goods, forced displacement, arbitrary detention, and forced labour. Documented violations of the first four types are primarily committed against adults, and mostly against males. Male victims of arbitrary detention and forced labour also tend to be adults, but the female victims are most frequently younger, in the 10–14 age category.
The violations most often against adolescents 10–14 years old are sexual slavery and rape (against girls) and forced recruitment (against boys). These violations should not be confused with sexual abuse, which was interpreted by the Commission primarily as the forced stripping of adult as a means of humiliation. Sexual abuse was most often perpetrated against adult males, while the other two sexual violations were most frequently committed against girls 10- 14.6
Torture, killing, and amputation are directed principally against adult men. Abduction is more complicated with both adult men and adolescent boys subjected to this violation. However, among female victims, girls 10–14 are considerably more frequently subject to abduction than younger girls or older women. Amputation is also directed most frequently at adult men, but among women and girls, the most common age category is 15–19.
Patterns of documented violations by victims’ ethnicity
This section addresses the question of systematic targeting of particular ethnicities for human rights abuses by the various perpetrator groups. Southern ethnicities are defined as Mende, Sherbro, Krim, Vai, Kissi. Northern ethnicities are defined as Koranko, Limba, Loko, Temne and Yalunka. First, in Figure 4.A1.22, we present the number of violations by type and by ethnicity of the victims.
In Figure 4.A1.23 below, responsibility for the violations against each of the ethnicities is shown across the perpetrator categories.
Other analysis of ethnic patterns is presented in the chapter on the Nature of the Conflict.
Patterns of documented violations by alleged perpetrator
Prior sections have considered perpetrators’ patterns over space, and with respect to victims age, sex, and ethnicity. This section considers the patterns of perpetrators with respect to type of violation and time.
The counts of victims, violations, the number of victims per violation, and the proportions of violations attributable to each perpetrator type are given in Figure 4.A1.24. Of the 40,242 violations in the TRC’s database, the RUF has by far the most violations 23,823 (59.2%) and the most victims 61.5%, attributed to them. The RUF also has the highest number of documented violations per victim 2.58; followed by the AFRC with 2.57 violations per victim.
There may be a negative bias against the RUF because the database measures the statement- givers’ perception of who was committing the abuses that they suffered or witnessed. Given the relatively high proportion of violations attributed to rebels, it is clear that there was some confusion in identifying the factions definitively. In terms of dress and behaviour, the RUF and AFRC fighters were virtually indistinguishable; both had ready access to SLA uniforms but commonly combined military fatigues with civilian clothing. In addition, identifiers such as headbands and sticking plasters were shared among factions. During the second phase, the civilian population developed the expression “sobels” to characterize perpetrators whom they believed to be “soldiers by day, rebels by night”. It is possible that many of the violations attributed to the rebels may be more accurately attributed to the RUF, AFRC or even the
SLA, but we were not able to clearly quantify this phenomenon in the data. However, it is discussed in detail in the Military and Political History Chapter.
Perpetrator Responsibility for Particular Violations
In terms of volume, the RUF committed the greatest number of violations for every violation type.
The RUF, rebels, AFRC, and SLA, follow roughly similar patterns of proportions of particular types of violations. Documented forced displacement and abduction violations constitute the highest proportion of all of the documented violations attributed to each of these four perpetrators. They also share nearly equal proportions of documented detention violations from 10.6% (421/3987) for the rebels to 11.8% (465/3950) for the AFRC, 12% for the RUF (2924/24353), and 12.3% for the SLA (327/2724).
The CDF follows a different pattern of violation types. The highest proportion, 16.6% (402/2419), of CDF documented violations is abduction, not forced displacement as is the case for the perpetrator groups discussed in the paragraph above. The proportion of documented CDF violations is higher than the other perpetrator groups for several violations types including assault/beating, torture, detention, extortion, and sexual abuse. However, the CDF committed proportionally fewer property destruction violations.
The RUF accounts for 67.1% (420/626) of documented rape violations.
Out of the documented abuses attributed to the AFRC, amputations constitute a proportionally higher (2.7%, 105/3950) number of their violations compared with the other perpetrator groups. However, the proportion of killing violations is lower for the AFRC (7.4%, 292/3950) than for the RUF (10.8%, 2618/24353) or the SLA (12.3%, 335/2724).
Perpetrator Responsibility for Violations over Time and Space
The RUF’s dominance over all violation types is not true in every period. In the graph series, Figures 4.A1.26a–o, below, the episodic nature of the conflict is clear for nearly every perpetrator, violation type, and year combination. That is, the violation counts start high in
1991 at the beginning of the war, drop in the early 1990s and then rise to the 1995 peak, after which the intensity drops. Violence increases during the expulsion of the AFRC from Freetown, their tour of the Northern districts and their eventual return to attack the capital in January 1999.
For the following violations, the reported counts for the RUF are higher than any other perpetrator category during every year: sexual slavery, rape, looting, killing, forced recruitment, forced displacement, abduction, forced labour, assault, destruction of property, and arbitrary detention. The exceptions to the RUF’s predominance are rare enough that they are noted here. For extortion and torture, the CDF shows peaks in 1997 which exceed the RUF counts of reported violations in that year. The AFRC count of reported acts of sexual abuse exceed the RUF in 1998, and the AFRC count of acts of amputation is greater than for the RUF in 1998.
There are clear differences between the perpetrators in terms of the timing of violations. The RUF has the most documented violations attributed to them in all years of the war, though the number of violations in 1998 and 1999 attributed to the AFRC are substantial. Whilst the SLA is involved in the conflict from the start, the AFRC coup in 1997 changes the nature and allegiance of the army. As a result, the AFRC is treated as a separate perpetrator group, active in the third phase. The SLA is responsible for significant numbers of documented violations during the second phase of the war, and the CDF is responsible for a significant number of violations in the third phase.
The RUF, CDF, and SLA play constant and distinct roles throughout the conflict, while the roles of ULIMO, the AFRC, ECOMOG, and GAF are confined to specific phases of the conflict. Prior to 1996, local militia groups were not coordinated under regional or national structures, but were active in the districts touched by the war. When the Sierra Leone Peoples Party (SLPP) government formed the CDF in 1996, it became common practice to refer to all such militias as CDF groups. The majority of CDF members were so-called Kamajors.7 The Kamajor force mobilized on a grand scale in the third phase of the war, from 1997 onwards. Seventy-four percent (1505/2031) of the recorded violations, with year documented that are attributed to the Kamajors, occur in 1997 or later.
The relatively minor perpetrator groups are those whose participation in the conflict is limited to specific years and geographical areas. Ninety-five percent (260/275) of the documented violations in the Commission’s database (where year is known) attributed to the ECOMOG intervention force, occur between 1997 and 2000. ECOMOG was not deployed by the Economic Community of West African States (ECOWAS) until 1997. The TRC recorded 201 violations attributed to the GAF, of which 155 had known year; of those with known year, 90% (140/155) occurred in 1999 and 2000. 91.8% (89/97) of the violations attributed to ULIMO, where the year is known, occur in 1991. 96% (105/109) of ULIMO violations, where district is known, occur in Bo, Kailahun, Kenema, or Pujehun.
In Figures 4.A1.27–30, we explore the patterns of violations across districts and time for the four factions that are responsible for the highest number of documented violations: the RUF, the AFRC, the SLA, and the CDF.
The Kamajor CDF force (a subset of the violations listed here as CDF) was largely confined to the South of the country: 62.2% (1089/1752) of the violations attributed to the Kamajor CDF militia, where the district in which the violation is known, occurred in the Southern region8; 23.1% (405/1752) in the Eastern, 9.2% (161/1752) in the Northern, and 5.5% (97/1752) in the Western. During the third phase of the conflict in the Bonthe district, the CDF are alleged to have committed the majority of the documented violations, 58.2% (322/553) in all.9
Patterns of documented violations attributed to the RUF appear similar in the first and second phases of the war. The exceptions are documented cases of sexual slavery and amputations which increase in the second phase when compared to the first phase, and documented cases of sexual abuse (Stripping/Naked Humiliation), which decrease in the second phase compared to the first.
The rise in documented sexual slavery in 1993 and 1994 coincides with the transition in the RUF to guerrilla tactics. The RUF fighters adopted a mode of fighting revolving around camps and bases within the bush where they abducted women and kept them as so-called “bush wives” in remote locations.10
Figure 4.A1.31 shows amputations by Perpetrator by Year. The first substantial rise in documented amputations occurs in 1995 and is attributable to the RUF. “Operation Stop Elections” is widely believed to be the first campaign of amputations by the RUF, occurring in late 1995 and early 1996 in order to coincide with the moves by civil society towards multi-party elections. Although there are a few reported amputations before 1995, in this year the reported count more than triples earlier totals. The rise in 1995 is consistent with the view that the RUF engaged in a limited campaign to warn civilians to “take their hands off the war,” in the wake of a failed NPRC peace initiative.11
It is interesting to note that while the RUF is responsible for the greatest number of violations reported to the Commission for each year of the conflict, in 1998, the database shows that the AFRC is responsible for the largest proportion — 48% (62/129) — of the recorded amputations.
Note: the columns do not sum to the total because responsibility for any violation might be shared among several perpetrators.
Figures 4.A1.32 and 4.A1.33 highlight the counts and percentages of violations in each region that are attributed to particular perpetrators. The RUF is alleged to have committed the majority of documented violations in all districts. It is noteworthy that the RUF is alleged to have committed a larger proportion of documented violations, 77.2% (2355/3050), in Kailahun, the district in which the war started, than in any other district. The AFRC is alleged to have committed its largest proportion of violations, 32.4% (669/2063), in Koinadugu, and the CDF is alleged to have committed 18.5% (462/2501) of the documented violations in Bonthe. ULIMO only has violations attributed to it that occurred in the Eastern or Southern regions.
Correlations Between Perpetrator Groups
This section examines the correlations between different perpetrators; in other words, how their patterns of documented violations were similar or different by violation type.
Figure 4.A1.34 shows the correlations between counts of documented violations for perpetrator type over violation type. To interpret this information, keep in mind that a value of one means perfect correlation, and values near zero mean no correlation. In the context of this table, a positive correlation means that as the first category count of violations goes up, the second category count of violations also goes up.
For example, the high correlation between RUF and AFRC in Figure 4.A1.34 (0.97) means that the proportions of RUF documented violations by violation type are highly correlated with the proportions of AFRC documented violations by violation type (e.g., the ratio of amputations to forced recruitments is similar for the two groups). In other words, in terms of the types and relative frequency of the documented violations, the behaviour of RUF and AFRC is broadly similar. In contrast, ECOMOG and GAF show much less correlation (0.65) over violation type.
The patterns of correlations in Figure 4.A1.34 suggest that, within the context of the Commission’s database, the AFRC, Sierra Leone Army (SLA), and RUF constitute a group of perpetrators whose documented abuses for most of the violation types, follow roughly similar patterns, although the volume of violations is different. Furthermore, the rebels behave similarly to this cluster of perpetrators. These patterns, however, do not inform us as to whether the violations are correlated by perpetrator group over time or not. The number of documented forced recruitments, acts of cannibalism, incidents of sexual slavery, and druggings in the TRC database are not large enough for correlation analysis. Perpetrator responsibility for particular violations types is discussed further on violations types more frequently reported in the Commission’s database.
Patterns of documented violations attributed to Liberian perpetrators
To examine the statements for Liberian responsibility at the beginning of the conflict in documented violations, a special coding study was conducted. The special coding was prepared when 6,740 of the TRC statements had been entered into the database.
The criteria was based on a section of the form used by the TRC for statement-taking that gathered demographic information of the perpetrator group, namely their ethnic origin, place of origin, and the languages they spoke. Some statements contained several incidents involving different groups of perpetrators; therefore it was not possible to determine to which group the perpetrator description applied. Inclusion in the study was limited to statements involving one incident, in which the alleged perpetrator is the RUF, with the events occurring between 1991 and 1994. A total of 1,073 of these statements met the required criteria.
A random sample of these statements was taken and stratified according to the year of the abuse. In total, 357 statements — approximately one-third of those available — were coded. For many statements, there was insufficient information to determine the origin of the perpetrators; these statements were not included in the study. The results of the study can be considered as representative of all statements containing one incident attributed to the RUF in the selected period, within the TRC database.
From each statement, the following fields were used to compile the statistics: Year (the year of the incident in which the RUF violations are alleged); Sierra Leoneans Included, (coded true if the statement indicated that the perpetrator group included persons of Sierra Leonean origin); and Liberians Included, (coded true if the statement indicated that the perpetrator group included persons of Liberian origin).12
For the purposes of the study, a perpetrator group consisting exclusively of Liberian fighters was assumed to belong to the NPFL. Similarly, a group consisting exclusively of Sierra Leonean fighters was considered to be part of the RUF/SL i.e. Revolutionary United Front of Sierra Leone. Additionally, many groups were mixed, containing both Sierra Leoneans and Liberians.
The majority of RUF incidents, 52%, were attributed to the NPFL, with 29% to the RUF/SL and 19% to mixed groups.13 Incidents involving both Liberian and Sierra Leonean perpetrators are relatively less common. The statistics are consistent with the view that in the first phase of the war the RUF consisted generally of two factions: the RUF/SL and NPFL.
RUF incidents in which Liberians were documented in the early years of the war showed a declining involvement, from 78% in 1991, to 69% in 1992, to 21% and 13% in 1993 and 1994. This information is consistent with the theory that a substantial proportion of the Liberians had departed from Sierra Leone by 1993.14
In summary, these results are consistent with the theory that there were campaigns of human rights violations by Liberians during the first phase of the war, but that the Liberian involvement in the war tapered out after this phase.
ECOMOG Abuses Study
The ECOMOG abuses study was the first special coding analysis, and it began on 7 November 2003. At that time, a total of 72 TRC statements describing killings by the ECOMOG force had been inputted into the database. A sample of 55 statements was studied; 17 other statements were in use by TRC researchers and could not be coded.
The study identified two types of killing: Indiscriminate Killing, defined as deaths due to bombing, shelling or cases where the victims were caught in crossfire; and Summary Executions, defined as deliberate killing of victims, typically by shooting and often accompanied by allegations that the victim was working in collaboration with “rebel” forces.
To make this distinction, the study considered the method of killing, allegations of collaboration against the victims, the origin of any collaboration accusation, the district where the killing occurred, and the circumstances in which the victim died. Accusations of collaboration may have been made by the perpetrators themselves or could come from civilian sources.
Fifty-six percent (50/89) of the documented and sampled killings attributed to ECOMOG were summary executions. Of the 50 summary executions identified in the statements, 76% (38/50) involved some accusation that the victim was involved with the AFRC or RUF factions. Where such an allegation was made, 70% (28/38) of the victims were accused of perpetrators spoke Liberian English, or were from an ethnic group common to both Liberia and Sierra Leone (Kissa, Vai), and there was no indication in the statement that any of the perpetrators were from Sierra Leone. being “rebels”. The remainder were accused of being either rebel collaborators (6/38), or members of a family containing a rebel (4/38). These results are consistent with the claim that elements within the ECOMOG force targeted and summarily executed suspected rebels and collaborators. ECOMOG is responsible for 0.8% (309/40242) of the total violations reported to the Commission.
Redress and Reparations
This section will also address the results of abuses, the current situation of victims, and, the attitudes of perpetrators and victims. The statistics compiled via the Assistance and Redress Study form the basis of the discussion in this section.
The assistance and redress study was unique in that the results were based on four separate samples. All of the samples were selected after the completion of the data entry of all the statements in the TRC database. Taking into account the margins of error (reported in footnotes), the percentages reported here can be interpreted as applying to all the TRC statements.
The first sample was stratified by country where the statement was taken — Sierra Leone, Guinea, Nigeria, or Gambia. A proportional sample of approximately 5% of the statements was taken, resulting in 296 statements being coded. This sample was used to explore the consequences of the abuse(s) the statement-giver experienced or witnessed, and whether or not the victim received medical attention or counseling following the abuse(s). It also examined how he/she currently supports him or herself.
The second sample of statements was comprised of all statements where a perpetrator was the statement-giver.15 The study examined answers to Section 6, questions 3.4 and 3.5 of the TRC statement form. These questions addressed the willingness of the perpetrator to meet with his/her victim, pay reparations to his/her victim, and what form those reparations would take.
The third sample examined whether or not the statement-giver would be willing to meet the alleged perpetrator of the acts the statement-giver experienced or witnessed.
The final study considered the types of assistance or redress sought by the statement-givers for this sample, and whether the request was intended to benefit themselves, their family, their community, or society as a whole. Some examples of the assistance categories are as follows:
• Homes/Shelter: Provision of homes/shelter; provision of building materials.
• Schools/Education/Training: Building of schools; improvement of schools; access to affordable education and/or skills and vocational training; provision of scholarships, affordable university fees.
• Hospitals/Medical: Building of hospitals or clinics; improvement of hospitals; access to affordable health care; treatment for physical or mental injuries resulting from the conflict.
Results of Abuses
Statement-givers were asked to describe the results of the abuses they experienced or witnessed as part of their statement to the TRC. Responses to this question were included in the first special coding sample for the Assistance and Redress Study.
Fifty-seven percent16 (102/178) of the statement-givers who gave a response about the result of the abuse they experienced or witnessed reported a loss of property. Additionally, 31% of statement-givers reported damage to either their mental (10/178) and/or physical health (45/178) as a result of the violations that they experienced or witnessed.17 Seventeen percent reported being permanently disabled (20/178) and/or unable to work (10/178) as a result of violations.18
The special coding study with this sample also investigated how many victims received medical attention or counseling following the abuses they suffered. As of the time the statement was given, a significant majority, 67% (137/204)19 of statement-givers, had not received medical attention or counseling following the abuses.
Current Situation of Victims
The first sample of statements included in the Assistance and Redress special coding study were also coded to examine the current status of the victim’s health.
Responses by the statement-givers that answered this question are nearly equally split between no longer being effected by the abuses they suffered to being effected on a daily basis.20 Of the statements included in the sample, 50% of the statement-givers reported “fair” (86/196) or “poor” (12/196) health at the time when the statement was given.21
The special coding study explored how statement-givers are currently able to support themselves. Of the statement-givers who responded to this question, over half the responses was divided nearly equally between statement-givers who reported supporting themselves by farming/gardening (44%, 90/205),22 Thirty-one percent (63/205) reported relying on relatives, friends, or children. It is interesting to note that very few statement-givers report supporting themselves through a job/salary (6%, 12/205).23
Attitudes of Victims and Perpetrators
The second sample of the Assistance and Redress special coding study comprised statements where a perpetrator was the statement-giver.24 The study examined answers to Section 6, questions 3.4 and 3.5 of the TRC statement report. These questions addressed the willingness of the perpetrator to meet with his/her victim, pay reparations to his/her victim, and what form those reparations would take.
Eighty-six percent (242/282)25 of the statement-givers included in this sample responded that they would be willing to meet with the victim of the human rights violation they committed.
Perpetrator statements were also coded to examine what he or she would be willing to do to make it up to his or her victim. In the TRC statement, statement-givers were asked to choose among four options in response to this question:
• Accept responsibility and offer apology
• Pay reparations
• Participate in rebuilding
Thirty-five percent (94/268)26 of the statement-givers responded that they would be willing to both accept responsibility and offer apology and participate in rebuilding.
The third sample of the Assistance and Redress special coding study explored whether or not the victim would be willing to meet with the perpetrator of the violations they suffered. An overwhelming 88% (219/250)27 of the statement-givers responded positively to the idea of meeting the perpetrator of the abuses committed against them if the meeting were facilitated by the TRC.
Needs Cited by Statement-Givers
Of all the requests for assistance or redress in the fourth special coding sample, 32% are to benefit the individual, 18% are for the statement-giver’s family, 26% are for the community and 23% concern changes or benefits for society as a whole.28 Typically the statement-giver would request several types of help. For example one statement-giver asked for treatment of his war injuries, education for his children, and the building of roads in the village. Given the approximately equal weight of self and community assistance, it is apparent that all of the following are sought:
• Assistance on an individual or family basis according to need
• Community projects to assist a town or village as a whole.
• Broad changes and reforms for society at large.
The vast majority of statement-givers indicate that the assistance should be provided by the government rather than a third party such as a nongovernmental organization or international donor.
Housing (49%), education (41%), and health care (27%) are the most frequently cited concerns. Housing, education and health are priorities at all scales of delivery — the statement-givers see it as important for the individual, family, community and society as a whole.
For the other forms of assistance there is some variation of the perception of how the assistance should be delivered:
• Unsurprisingly, infrastructure is seen as something that should be primarily delivered at the community level.
• Religious rites are a requirement for the community or society as a whole, rather than for specific individuals or families.
• Institutional and economic reforms are broad benefits required for society as a whole.
• The provision of cash, materials and credit is supported as a benefit for individuals, families and communities.
There were some differences in the weight given to the different types of assistance depending on whether the statement-giver was male or female. Men placed a slightly greater emphasis on assistance to themselves or the community, while women more often cited the need for assistance for the family unit.
The Sierra Leone Truth and Reconciliation Commission collected nearly 8,000 statements from Sierra Leoneans regarding their experiences over a decade of conflict. The purpose of this appendix has been to outline and interpret the descriptive statistics regarding the nature and extent of violations, behaviour of perpetrators, and characteristics of victims that can be gleaned from these statements. To obtain this information TRC staff and consultants undertook coding, data entry, matching, and statistical analysis. While valuable in its own right, the resulting quantitative information is even more powerful combined with the contextual information compiled by the TRC researchers, investigators, and commissioners. Therefore this information is incorporated in greater depth and detail in each of the chapters of the Final Report.
About the Authors
Richard Conibere, B.S., is the Data Processing Manager for the Truth and Reconciliation Commission of Sierra Leone, the Data Entry Supervisor for the American Bar Association Sierra Leone War Crimes Documentation Project, and a field consultant for HRDAG. Until 2001, he was a computer contractor working in database and Internet development. Richard also worked on the development of Analyzer, the data collection, storage and matching application that was used in this project as well as on others around the world. His other projects include ongoing assistance to the Ghanaian National Reconciliation Commission.
Jana Asher, M.S. is the Documentation Coordinator for the American Bar Association Sierra Leone War Crimes Documentation Project, Statistical Consultant for the Truth and Reconciliation Commission of Sierra Leone, and Statistical Consultant for the Human Rights Data Analysis Group of The Benetech Initiative (HRDAG). Previously, Ms. Asher designed the stratification and modeling for the data analysis presented in the HRDAG report to Perú’s Truth and Reconciliation Commission (CVR), and she and Dr. Patrick Ball of HRDAG developed the statistical methods for estimating the death counts outlined in Killings and Refugee Flow in Kosovo March–June 1999. Ms. Asher has served as a statistical consultant on multiple projects for Physicians for Human Rights (PHR), and she was a coauthor on the PHR article “Human Rights Abuses and Concerns About Women’s Health and Human Rights in Southern Iraq” published in the Journal of the American Medical Association in March of 2004. She is currently completing her Ph.D. in Statistics at Carnegie Mellon University under the guidance of Professor Stephen E. Fienberg, as well as designing and implementing the methodology for the ABA/CEELI household survey of human rights violations in Sierra Leone.
Kristen Cibelli, B.A., is the Associate Documentation Coordinator and HRDAG Liaison for the Sierra Leone War Crimes Documentation Project. She is the Analyzer Product Manager and the Project Manager for HRDAG projects in Chad, East Timor, Ghana, and two confidential projects. She was a co-author on the HRDAG report “Preliminary Statistical Analysis of AVCRP & DDS Documents: A report to Human Rights Watch about Chad under the government of Hisseine Habre”. Ms. Cibelli worked with Dr. Patrick Ball of HRDAG on an independent research project investigating local perceptions of the International Criminal Tribunal for the Former Yugoslavia in Bosnia and Herzegovina from the perspective of local NGOs working on post-conflict reconstruction and reintegration. She received her B.A. in International Relations with a certificate in Peace and Justice Studies from Tufts University.
Jana Dudukovich is the Record Linkage Officer for the Sierra Leone War Crimes Documentation Project, HRDAG Data Management Coordinator and Project Manager for HRDAG projects Sri Lanka, Sierra Leone, and Colombia . Her roles include advising the data collection, coding processes, and overseeing data cleaning and matching for HRDAG projects. Ms. Dudukovich has co-authored the following HRDAG reports: “Killings and Refugee Flow in Kosovo March–June 1999”, and “Preliminary Statistical Analysis of AVCRP & DDS Documents: A report to Human Rights Watch about Chad under the government of Hisseine Habre”. She is currently working on data from Chad, East Timor, Sri Lanka, and Sierra Leone.
Rafe Kaplan, B.S., is a Senior Software Engineer for HRDAG. He is the senior developer for Analyzer, the next generation of human rights violation database software to perform quantitative analysis for large-scale human rights data projects. Additionally, he has provided field support to the Afghan Independent Human Rights Commission in Afghanistan, the Human Rights Documentation Coalition of Sri Lanka, and a confidential HRDAG project. Mr. Kaplan holds a B.Sc. degree in Computer Systems Science from the University of Manchester UK and has worked professionally for 6 years writing software in New York City. He is currently working on HRDAG projects in Sri Lanka, Sierra Leone, East Timor and a confidential project.
Patrick Ball, Ph.D., is the Director of Human Rights Programs at the Benetech Initiative. Since 1991, he has designed information management systems and conducted quantitative analysis for large-scale human rights data projects for truth commissions, non-governmental organizations, tribunals and United Nations missions in El Salvador, Ethiopia, Guatemala, Haiti, South Africa, Kosovo, and Perú. Dr. Ball has published several reports on statistics and human rights, and in March 2002, he appeared as an expert witness in the trial of Slobodan Milosevic at the International Criminal Tribunal for Former Yugoslavia (ICTY). His most recent work is an estimate of the total deaths in Peru, 1980–2000, conducted on behalf of the Perúvian Truth and Reconciliation Commission. Dr. Ball has received several awards, including in June 2004, the Eugene L. Lawler Award for Humanitarian Uses of Computing from the Association for Computing Machinery. He is currently involved in HRDAG projects in Sierra Leone, Sri Lanka, East Timor, and Columbia.
1. For further information, see the “Military and Political History of the Conflict” Chapter of the Final Report of the Sierra Leone Truth and Reconciliation Commission.
2. Who Did What to Whom? Planning and Implementing a Large-Scale Human Rights Data Project, Patrick Ball (1996), AAAS: Washington, DC, USA.
3. For more detail on the creation of the TRC database, see Volume 1, Methodology and Processes Chapter of the Final Report of the Sierra Leone Truth and Reconciliation Commission.
4. For further information please see the Military Chapter section on Context, Build-up and Dynamics on Bomaru.
5. Sexual abuse was found by the Commission to be a policy of some insurgent factions that deliberately singled out men in the communities they entered to be stripped naked and otherwise humiliated in front of their communities. This policy was found by the TRC to be an element of the insurgents’ efforts to take control of “target” towns and villages in the first phase of the conflict.
6. For a discussion of “targeting” of girls and boys in these violations, see the Children's Chapter.
7. For information on the formation of the CDF, refer to Phase II of the “Military and Political History of the Conflict” Chapter of the Final Report of the Sierra Leone Truth Commission.
8. Note that geographically, the Eastern region is in the Southern half of the country.
9. See Figure 4.A1.7 for the figures for Bonthe.
10. For more information on the switch to guerrilla warfare, associated objectives and strategies, see Phase II of the “Military and Political History of the Conflict” Chapter of the Final Report of the Sierra Leone Truth and Reconciliation Commission.
11. See Phase II of the “Military and Political History of the Conflict” Chapter of the Final Report of the Sierra Leone Truth and Reconciliation Commission.
12. Statements meeting any of the following criterion were attributed to the NPFL; The statement indicates that the perpetrators were Liberian or Burkinabey, or from a Liberian ethnic group (Mano, Ngio or Pelle), or the Statements meeting any of the following criteria were attributed to the RUF: The statement indicates that the perpetrators were from an exclusively Sierra Leonean ethnic group, the perpetrators spoke Sierra Leonean languages; or the statement specifically states that the perpetrators were from Sierra Leone or a district within Sierra Leone.
13. The margins of error are ± 9%, 8%, and 7%, respectively.
14. By year, the margins of error are ± 9%, 18%, 22%, and 9%, respectively.
15. Although a conscientious attempt to locate all such statements was made, only 300 of the statements that are given by a perpetrator were part of this special coding. While not all of the perpetrators’ statements were included, the results from this analysis can be considered representative of all of the perpetrators who gave statements to the TRC because the number missing is such a small proportion of the whole.
16. The margin of error for this statistic ± 7%.
17. 2%–9% of victims reported damage to their mental health, and 19%–32% reported damage to their physical health.
18. The confidence intervals are as follows: disabled 7%–16% , unable to work %2–9%.
19. The confidence interval is 61%–74%.
20. Victims’ responses to this question were coded according to the following definitions: Excellent: No health problems, Good: Minor illness that doesn’t affect daily life, Fair: Major illness/Disability that somewhat affects daily life, Poor: Daily life greatly affected (can’t work, can’t care for family).
21. For the other categories, 44% reported “fair” health with a confidence interval of 37%–51%, and 6% reported “poor” health with a confidence interval of of 3%–9%
22. The confidence interval for farming/gardening is 37%–51% and the assistance of relatives/friends/children confidence interval is 24%–37%.
23. The confidence interval on supporting oneself by a job/salary is 3%–9%.
24. 300 perpetrator statements were part of this special coding.
25. The confidence interval is 82%–90%
26. The confidence interval is 29%–41%
27. The confidence interval is 84%–92%
28. Note that the figures do not total 100 percent because many statement-givers requested several types of assistance. All of these statistics are significantly different from zero at p=0.05.