The analysis in this chapter is derived from a database compiled by Human Rights Watch of 367 RAB killings.143 It is based on three sources, in order of frequency: Bangladeshi media reports, Bangladeshi human rights organizations, and Human Rights Watch interviews.144 All records included in the database are of killings that have either wholly or partially been attributed to the RAB and that occurred between the start of RAB operations in June 2004 and the end of September 2006.145 For the sake of clarity, all percentages have been rounded to the nearest whole number.
Since its creation, RAB has consistently announced its killings, often in the form of press statements. Nevertheless, as these data are compiled primarily from newspaper and human rights reports, it is possible that they do not capture all RAB killings between June 2004 and September 2006. These data sources may be limited in their ability to report on all killings because:
In addition, the dates of reported killings vary slightly, with some media and human rights groups reporting the day of arrest and others the day of death. In the vast majority of cases the difference is less than one day, so the one-month intervals presented in this chapter mostly capture the period in question.
Many killings were reported by more than one source, and sometimes giving slightly different victim names. But Human Rights Watch took special care to eliminate all potential double-counts, erring on the side of caution when names were close but not the same.
Lastly, Human Rights Watch has not investigated all 367 reported killings. The human rights and press reports on which the database was primarily built strongly suggest that most of the deaths resulted from torture or execution, and Human Rights Watchs own research confirms this trend. But some of the killings may have resulted from a legitimate use of force. Killings that, based on the available information, resulted from what appeared to be a legitimate use of force were excluded.146
Magnitude and temporal pattern of reported RAB killings
Of the 367 RAB killings reported, 77 percent (284/367) were reported by the media, human rights groups, or RAB as crossfire killings (in which the victim was allegedly a bystander in a shootout between the police and an armed group), and 11 percent (42/367) were described as killings during shootouts (in which the victim allegedly took part in a shootout with the police).
Figure 1: Reported Killings by RAB, June 2004 September 2006
As shown in Figure 2, reported RAB killings per month in 2004 and 2005 averaged a similar rate: 11.7 per month in 2004 and 10.3 per month in 2005. For the first nine months of 2006, the average monthly rate of documented killings jumped to 17.9, a dramatic increase over the previous two years. This increase may suggest a greater increase in RAB killings, increased reporting of cases, or both. It may also result from the growth of RABs force from seven battalions with about 5,000 members in 2004 to twelve battalions with about 8,500 members in 2006.
Figure 2: Monthly Reported Killings by RAB,
The monthly pattern of documented RAB killings does vary a little over time, as shown in Figure 3, and is centered around a median of 13 killings per month. The high point was in mid-2006: RAB killed 25 people in May and 37 people in June, the highest for any one month.
Figure 3: Reported Killings by RAB by Month,
There are 12 RAB battalions, each with a specific area of responsibility.147 However, reported RAB killings are not uniformly distributed throughout all six geographic divisions of Bangladesh.148 As seen in Figures 4 and 5, nearly 32 percent (117/367) of reported RAB killings took place in Dhaka division, followed by Khulna division with 29 percent (107/367).This is not surprising because these divisions are by far the most populous in Bangladesh.149 More telling is the number of RAB killings compared to population, as shown in Figures 6 and 7. This reveals that the division with by far the most reported RAB killings compared to population is Khulna, with one reported killing for every 135,223 people, followed by Dhaka division, with one reported killing for every 330,580 people. Khulna is the principal area of responsibility of RAB-6.
Figure 4: Reported RAB Killings by Geographic Division,
Figure 5: Reported RAB Killings by Geographic Division,
Figure 6: Reported RAB Killings by Geographic Division,
Figure 7: Reported RAB Killings by Geographic Division,
In some reported incidents, the media or human rights source cited the specific RAB battalion(s) responsible for the killing.150 In the majority of cases, however, the sources did not report the battalion involved. In these cases Human Rights Watch deduced the battalion based on the respective areas of responsibility (AOR) of the 12 battalions and where the incident took place.151 The results may not be fully accurate because, in some cases, battalions may have operated outside their specific AOR. In addition, RAB grew from seven to twelve battalions.
As Figures 8 and 9 reveal, more than one-quarter of all RAB killings25.6 percentoccurred in the current AOR of RAB-6, Khulna.
Figure 8: Reported Killings by RAB Battalions,
Figure 9: Reported Killings by RAB Battalions,
Victims of all reported RAB killings through September 2006 were male. As shown in Figure 10, these victims were overwhelmingly male adults71 percent (259/367) were between the ages of 20 and 39. The youngest reported victim was 14, Ashiqul Islam Raju, killed in September 2006.152 The oldest was 65, a deed writer named Mohamad Ali, killed in July 2004.153
Figure 10: Reported RAB Killings by Victims Age and Sex,
143 The statistical analysis was conducted and presented by Romesh Silva, statistician with the Benetech Human Rights Data Analysis Group. Patrick Ball, director of Benetechs Human Rights Data Analysis Group reviewed the work.
Benetechs Human Rights Data Analysis Group (HRDAG) develops computer software, data collection strategies, and statistical techniques to measure the patterns and magnitude of acts of violence and human rights violations. HRDAGs technology and quantitative analysis is used by truth commissions, international criminal tribunals, and nongovernmental human rights organizations around the world to identify the trends and patterns that may be evidence of crimes of policy. See www.benetech.org/human_rights/.
144 The killings data collated by Human Rights Watch came from newspaper sources, the Bangladeshi human rights organizations Ain o Salish Kendra, Odhikar, and Hotline Human Rights Bangladesh, and Human Rights Watch field interviews and research. The newspaper sources were the main English-language papers in Bangladesh: Daily Star, Bangladesh Observer, The Independent, New Age, and Weekly Holiday. Information from Bangladeshi human rights groups came from their statements and reports, as well as databases the groups had compiled based on media reports (the Bangla-language newspapers Prothom Alo, Inquilab, Jugantor, Bhorer Kagoj, Jonokhonto, Ittefaq, Janakantha, Dinkaal, and Sangbad).
One data source with the victims name, place of death, and incident date was enough for entry into the database; 76 percent (279/367) of reported killings were documented in one data source, while 19 percent (68/367) of reported killings were documented in two data sources. The number of entries documented in more than one source is undoubtedly higher; Human Rights Watch generally stopped searching after an initial source was found.
A total of 367 individual cases of killings were identified out of a total of 441 reported killings that are documented in the database. Hence, 74 duplicate records of killings were identified in the database. These duplicate records were identified by examining the victims names and locations of incidents in the database and comparing records to each other. Records which contained the same victim name, age (within a margin of five years), the same geographic division of incident, and the same date (within a margin of three days) were deemed likely duplicates and controlled for when carrying out the statistical analysis. When duplicates were identified and more precise data existed in one of the records than the other matching record(s), the more precise data values were used in the statistical analysis.
145 Partial attribution to RAB refers to the 10 killings of suspected Indian insurgents on June 12, 2006, reportedly committed by RAB-7 and the army. See Bodies of Indian Insurgents Handed Over to Baghaichhari Thana Police, The Independent, June 13, 2006, http://independent-bangladesh.com/news/jun/14/14062006cr.htm (accessed November 7, 2006), and Army, RAB Raid Kills 10 Indian Insurgents, Daily Star, June 13, 2006, http://www.thedailystar.net/2006/06/13/d60613013120.htm (accessed November 7, 2006).
146 For example, in September 2004 RAB in Dhaka reportedly opened fire on a suspected criminal named Abdus Sobhan. Apparently by accident they killed a five-year-old girl, Mayesha Rahman. (Crossfire Between RAB and Goons Kills 5-year-old, Daily Star, September 10, 2004, http://www.thedailystar.net/magazine/2004/09/02/news.htm (accessed November 30, 2006.) According to the press account, RAB fire also struck Sobhan and, as he fell, a RAB member continued to fire repeatedly at point-blank range. Both Sobhan and Rahman were excluded from the database.
147 See above, Chapter II, Structure of RAB.
148 The six geographic divisions of Bangladesh are Dhaka, Khulna, Rajshahi, Chittagong, Sylhet, and Barisal. Within these divisions are 64 districts.
149 According to the 2001 census, the population of Dhaka division was 38,677,876, Rajshahi division 29,992,955, Chittagong division 23,999,345, Khulna division 14,468,819, Barisal division 8,112,435, and Sylhet division 7,899,816. See http://www.statoids.com/ubd.html (accessed November 21, 2006).
150 In six cases, two RAB battalions were reportedly involved. In one incident, RAB was reportedly responsible together with the army.
151 See Chapter II, Structure of RAB, for the 12 battalions and their areas of responsibility.
152 RAB Accused of Killing 14-year-old Boy, Daily Star, September 21, 2006, http://www.thedailystar.net/2006/09/21/d60921060370.htm (accessed November 5, 2006).
153 Pichchi Hannan Killed, Weekly Holiday, August 13, 2004, http://www.weeklyholiday.net/2004/130804/mis.html (accessed November 7, 2006).