II. Statistical Methodology
Data Sources
The three data sources used in these analyses were compiled by different research organizations: Physicians for Human Rights, Human Rights Watch and ABA/CEELI-Center. Slightly different data collection techniques were employed, and it must be understood that the data were gathered by researchers working within often chaotic conditions to interview Albanian witnesses to killings or the evidence of these killings. Nonetheless, through these witness accounts, it is possible to estimate the total number of Kosovar Albanians who were killed during the violence between March 20 and June 12, 1999.
Physicians for Human Rights34
Between April 19 and May 3, 1999, Physicians for Human Rights and the Program on Forced Migration and Health of Columbia University’s Joseph L. Mailman School of Public Health interviewed representatives from 1,180 households of Kosovar Albanian refugees. The interviews were conducted at 31 refugee camps or collective centers35 in Albania and Macedonia. The research team used a modified random systematic sampling method to identify the households that would be selected for interview.
With a target sample of 1,000 interviews, the research team selected 31 camps or centers in Macedonia and Albania. In Macedonia, six camps were sampled, while in Albania, 25 camps or centers with refugee populations greater than 1,000 were sampled. Within each camp, the number of households to be sampled was computed by comparing the number of refugees in the camp to the estimates of the total number of Kosovar refugees. All households were identified and the total number of households in the camp was divided by the number of households to be sampled. This number, “i,” was the sampling interval. For example, if there were 200 interviews to be conducted in a camp with 1,000 refugee households, every 1,000/200, or 5th household would be interviewed. The research team selected a random starting point and sampled every ith household. In total, 1,209 households were sampled for participation in the study. 1,180 of the 1,209 sampled households participated in the study for a response rate of 98 percent.
In these structured interviews, respondents were asked to provide demographic information on themselves and the other household members, including the respondent’s age, gender and profession and the village of residence of the household. Respondents also provided information on any attacks on their homes or the homes of others or the after effects of these attacks that they witnessed. They provided information on killings that they had witnessed or bodies that they had observed. Respondents were also asked whether they or members of their families had been victims of any of a range of violent attacks. The perpetrators (Serbian police, Serbian soldiers and/or paramilitary forces or UÇK forces) of these abuses were documented whenever possible. In total, there were 59 reports of killings of members of the sampled household unit. Twenty-six of these killings were reported to have occurred between March 20 and June 12, 1999.
While the research team attempted to follow standard systematic sampling methods, adaptations were made. For example, those refugees who did not reside within camps or collective centers were not sampled for interview. These unsampled refugees consist primarily of those who resided in private households in Albania or those who never left Kosova/Kosovo. The experiences of the members of these households may differ from those refugees who resided in the camps or centers. In addition, households in Albanian camps in with refugee populations under 1,000 were not interviewed, due to the expense involved in conducting a relatively small number of interviews in a large number of camps spread out over a wide geographical area.
The Human Rights Watch dataset was derived from 591 interviews with Kosovar Albanian refugees between March and October of 1999. HRW researchers interviewed refugees as they crossed the border into Albania, Macedonia and Montenegro. Researchers also interviewed refugees in Albanian refugee camps. After June 12, 1999, researchers traveled into Kosova/Kosovo to interview additional witnesses within their home villages. The geographic regions within Kosova/Kosovo for interviewing were selected based on two criteria. First, areas were selected based on refugee reports of mass human rights violations. Second, reports of mass human rights violations were reported by other sources than refugees,36 and these were also used to identify areas to conduct interviews.
By interviewing witnesses as they crossed the border out of Kosova/Kosovo, and those who returned home to Kosova/Kosovo, HRW researchers included reports from refugees who did not necessarily reside in the refugee camps. Some of the refugees in the HRW study may have resided in private homes in Albania while others may never have left Kosova/Kosovo. Thus, this study was able to interview a broader (albeit smaller) cross-section of the Kosovar Albanian population than the PHR study.
Witnesses were asked open-ended, unstructured questions about their experiences between March 24 and June 12, 1999. Information that was volunteered about incidents that occurred outside of these dates was also documented. Accounts were translated into English and recorded verbatim by researchers. These narratives were then coded for among other things, killings that the respondents witnessed or bodies that they observed. Of the 591 interviews, 376 contained reports of killings. Of the 376 reports with killings, there were 293 that included killings that occurred between March 20 and June 12, 1999. These were divided up as follows: 9 were from interviews at the Macedonian border or within Macedonian camps, 93 were from interviews at the Albanian border or within Albanian camps and 173 were reported during interviews conducted within Kosova/Kosovo. Within these reports there were 5,705 documented killings.
A third source of data is from the ABA/CEELI War Crimes Documentation Project (WCDP) and the Center for Peace Through Justice. Researchers interviewed refugees at camps in Albania and Macedonia as well as refugees who traveled to Poland and the United States. Using semi-structured interviews, researchers asked witnesses about the experiences surrounding their departure from Kosova/Kosovo. In addition, they were asked about incidents of violence and intimidation that they experienced or witnessed.
As of October 1999, ABA/CEELI - Center had compiled a database of 1,582 interviews of witness testimonies, in which 1,622 incidents of one or more killings were recorded. The database records were not coded for specific killing-related information, so the original interviews were identified and recoded. During the recoding process, an additional 81 incidents of killings were identified. A random sample of thirty interviews that were not originally identified as involving reports of killings were sampled and coded to determine whether they contained reports of killings. Only two (6.7 percent) of these also indicated any reports of killings. Thus, AAAS decided that it was not necessary to return to all of the records in the database to identify any additional reports of killings.
In total, from the combined data sources, there were 2,422 incidents of killings, involving at least 6,374 victims killed between March 20 and June 12, 1999.37
34 This discussion is drawn primarily from Physicians for Human Rights, War Crimes in Kosovo (1999).
35 Collective centers included farms, schools and large buildings, such as factories, etc.
36 For example, mass human rights violations were reported by international observers (e.g. the OSCE-KVM) and media outlets reporting on the crisis.
37 This estimate is based on the very conservative assumption that those incidents that contain an unknown number of killings involve only two killings.
 |
|
|
|
 |
 |
 |
|