CHANGES IN THE ETHNIC COMPOSITION IN THE MUNICIPALITY OF VI[EGRAD BETWEEN 1991 AND

CHANGES IN THE ETHNIC COMPOSITION IN THE MUNICIPALITY OF VI[EGRAD BETWEEN 1991 AND 1997 1 EXPERT REPORT FOR THE MILAN LUKI] AND SREDOJE LUKI] CASE (IT-98-32/1) EWA TABEAU DEMOGRAPHIC UNIT, OFFICE OF THE PROSECUTOR, ICTY SEPTEMBER 1, 2008 Summary of Results This report summarises changes from 1991 to 1997 in the ethnic composition of the Vi{egrad municipality in the middle-east of Bosnia and Herzegovina. Its goal is to provide reliable demographic statistics that allow for an assessment of the type and scale of the changes. Secondly, I discuss basic demographic profiles of persons who went missing in the Vi{egrad municipality during the war. The study of missing persons aims at showing who were those who went missing and finding, whether there were any particular periods in which people disappeared. In this report I analyse three data sources: the 1991 Population Census, the 1997 Voters Register, and the ICRC list of missing persons (the 2005 edition), and use standard statistical and demographic methods. The major findings are the following: - In 1991 Vi{egrad had two main ethnic groups: Muslims (63.5%) and Serbs (31.8%). There were also Others - mainly Yugoslavs (4.5%), and Croats (0.2%), that lived there. - The 1991 population was however unequally distributed within Vi{egrad: out of the total of 159 settlements within the municipality, 93 had a clear Muslim majority, 60 a clear Serb majority and only six could be considered as mixed, including the Vi{egrad town. - The post-war ethnic structure of Vi{egrad was dominated by Serbs (95.9%). During the war Muslims entirely disappeared from Vi{egrad. This is reflected also at the level of settlements. - Scale of the process of changes in the ethnic composition in Vi{egrad was substantially larger than in other municipalities in the surroundings (excluding Srebrenica and Bratunac). - The vast majority of persons missing in Vi{egrad were Muslim men, mainly aged 15 to 44 years. - The process of going missing was most intense in May and June 1992; over a half of the missing persons disappeared in Vi{egrad town. - Number of persons missing in Vi{egrad in 1992 was larger than in any other municipality in the surroundings. Timing of disappearances in 1992 in Vi{egrad is similar to the timing in the surrounding municipalities. 1 This report is an updated version of the expert report: E. Tabeau and J. Bijak, 2001, Changes in the Ethnic Composition in the Municipality of Vi{egrad, 1991 and 1997; prepared for the VASILJEVI] case (IT-98-32); ERN: 0213-8385-0213-8433 (English). The 2001 report was presented to the Trial Chamber during the expert witness testimony of Ewa Tabeau on 19 th September 2002. 1 September 2008 1

The Reference Map of the Pre-war Municipality of Vi{egrad and its Surroundings @ivinice Kalesija Kladanj [ekovi}i Zvornik Bosnia and Herzegovina Olovo Vlasenica Bratunac Area shown on the main map Sokolac Han Pijesak Srebrenica SERBIA Pale Rogatica Vi{egrad Legend: State borders Municipal borders Gora`de ^ajni~e Rudo Olovo Municipality names Dayton Line Major towns Fo~a MONTENEGRO Approximate scale: 0 10 20 km 1 September 2008 2

1. Introduction This report summarises changes in the ethnic composition of the Vi{egrad municipality in the middleeast of Bosnia and Herzegovina between 1991 and 1997. Its goal is to provide reliable demographic statistics that allow for an assessment of the type and scale of the changes. In this report I analyse three data sources: the 1991 Population Census, the 1997 Voters Register, and the ICRC list of missing persons (the mid-2005 edition). These sources are reliable and relevant to the objectives of this report. Changes in the ethnic composition are shown by comparing the 1991 Censusbased statistics with statistics based on the 1997 Voters Register. The 1991 Census is the latest complete population survey conducted directly before the war and gives a very accurate perspective on the population and its ethnic composition in this period. As to the 1997 Voters Register, I realise that ideally a period closer to the early phase of the conflict should be taken for the comparison with the 1991 Census. Such sources are, however, generally unavailable and therefore cannot be analysed here. Even though the year 1997 is quite distant from the period in which the most population movements occurred, the 1997 Voters Register can be seen as a valuable source to discuss the war-related changes in the ethnic composition. Firstly, because in 1997, (as I will prove in this report), many displaced persons still lived in areas different from their place of residence in 1991. Secondly, because the returns of internally displaced persons that took place in 1996-97 were not considerable which suggests that many displaced persons resided in 1997 in temporary locations acquired during the war. The report consists of the Summary of Results (included at the beginning of this report), four main sections, and three Annexes. Section 1 is a general Introduction. In Section 2 I discuss details of Data Sources and Methods, and in Section 3 I summarise Changes in the Ethnic Composition 1991-1997. Section 4 is devoted to Basic Demographic Distributions of the Missing Persons in Vi{egrad. Finally, the Annexes provide the reader with some additional information. Annex A contains the list of persons who went missing in the Vi{egrad municipality, being an excerpt from the ICRC list of missing persons. Annex B provides information about the ethnic composition of all settlements in Vi{egrad municipality in 1991 and of a selection of them in 1998. Annex C contains the description of statistical methodology (chi-squared goodness-of-fit test) used in the report. One practical remark regarding the use of the Voters Register is that when the total number of the 1997 voters is broken down by place of registration, the resulting samples can be very small. In order to increase the sample size, I use the 1997 Voters Register in combination with the Register from the 1998 election, which was also supervised by the OSCE. The type of information available from the two Registers is the same. The 1998 Register is only used for those voters who registered first time in 1998. Approximately 95% of the voters registered first time in 1997 and only about 5% in 1998. Therefore, my results should be seen (and will be further referred to) as the 1997 figures. 1 September 2008 3

2. Data Sources and Methods 2.1 Population Census for Bosnia and Herzegovina 1991 My source of information on the pre-war population of Vi{egrad is the 1991 Census for Bosnia and Herzegovina. The Census was taken in April 1991 (officially per March 31, 1991), just before the outbreak of hostilities in the former Yugoslavia. In statistical practice, the population census is the largest and most complete source of information about the population in a country. The 1991 Population Census covered the entire population of Bosnia and Herzegovina as of 31 March 1991. During the Census, information was collected about a total of 4,4 million individuals. The information about individuals was obtained in face-to-face interviews based on a Census questionnaire designed in a uniform way for the whole country, i.e. the former Yugoslavia. Preparations for the 1991 Census started about a decade earlier and included among other things: preparing and adopting the Census Law at the country and republican levels, appointing the authority responsible for the conduct of the Census (in Bosnia it was the Republican Statistical Office in Sarajevo; the municipal Statistical Offices were the local agencies responsible for the actual Census taking), developing the Census questionnaire, issuing manuals and training of the Census inspectors and interviewers, developing procedures for duplicate elimination, error checking and corrections, designing data processing schemes, preparing and conducting a Pilot Census, and after the actual 1991 Census - a Control Census, plan of publication etc. The electronic Census files contain one record for each enumerated person. These records include information on a large number of variables, such as the municipality and settlement of residence, name and surname, father s name, household sequential number, personal ID number, date and place of birth, sex, occupation, ethnicity, mother tongue, religion, educational attainment, the number of children born (for women only), and many more. The overall data quality is good, except for frequent errors in the persons names. These errors are mostly consequences of poor optical scanning of the original forms (for example misreading V for U, as in MVSIĆ) and no subsequent checking and editing. To correct the scanning errors the Demographic Unit - OTP employed several strategies. First, computer software was developed and applied to detect combinations of letters that are impossible in the B/C/S language. The software used the B/C/S syntax in order to assess the viability of combinations. The impossible combinations were corrected by eliminating miss-shaped (illogical) characters and inserting their most likely equivalents. Secondly, we developed correction tables to eliminate scanning mistakes from the names. The tables contained the actual names and their correct versions which both were used by a computer programme to produce suggestions regarding the corrections needed. Then, these suggestions were controlled manually to discard any wrong corrections produced by the software. The accepted corrections were then applied to the data. Native speakers of the B/C/S language, who in addition were familiar with naming traditions in Bosnia and Herzegovina, undertook all these tasks. Furthermore, we also developed and applied computer software that utilised household information to correct surnames within households. The software checked the correctness and consistency of family names within the same households. Household members, whose family name was different from the (correct) name of others in this particular household, received the correct name. For instance, if MUSI] was the correct surname in a household, the person enumerated as part of this household under the name MVSI] would become MUSI]. 1 September 2008 4

A second data quality problem is that for a number of records the unique 13-digit personal ID number (jedinstveni mati~ni broj, JMB), introduced in the former Yugoslavia in 1981, is only partly available. The JMB consists of date of birth (DOB, 7 digits), region of birth (2 digits), a sex-specific sequential number (3 digits), and a check digit (1 digit). For our needs the date of birth is essential, other components of the JMB being of less value. The date of birth is missing only for a few per cent of the 1991 population; i.e. for only 4.3% of the 1991 population of Vi{egrad. In my opinion, data-related problems do not discredit the 1991 Census as a powerful source of information about the pre-conflict population in Bosnia and Herzegovina. The Census includes a variable on the ethnicity of the enumerated individuals. This allows us to study the population in the context of the 1991 ethnicity for all those individuals whose records have been linked between two data sources (e.g. the 1991 Census and ICRC list). The question on ethnicity in the Census questionnaire was open-ended meaning that individuals could declare themselves as belonging to any ethnicity. The majority of the 1991 Census population declared themselves as belonging to one of the three major ethnic groups in Bosnia and Herzegovina: Serbs, Muslims, or Croats. Other ethnic declarations in the 1991 Census included Yugoslavs (relatively frequently), combinations of ethnicities, such as Serb-Croat or Muslim-Serb (infrequently), and other national (e.g. Vlach or Gypsies) or foreign (e.g. Hungarians) ethnicities (less frequently). Those who called themselves Yugoslavs, or by names combining two ethnicities, were often children from mixed marriages. The Yugoslavs did not feel they belonged to any particular ethnic group and frequently disliked ethnic categorisation. For this report, four ethnic groups were distinguished on the basis of ethnicity declarations in the 1991 Census: Serbs, Muslims, Croats, and Others. The last group, Others, is a residual category and covers persons who declared themselves as Yugoslavs, combinations of ethnic groups, and other national or foreign ethnic groups. 2.2 OSCE Voters Registers 1997 and 1998 The Voters Registers discussed in this section were established under the auspices of the OSCE; i.e. the Organization for Security and Co-operation in Europe; they are therefore often referred to as the OSCE Voters Registers. The basis for establishing these Registers was the 1991 Population Census that after the conflict was the latest available complete source of information about the population of Bosnia and Herzegovina, and in particular about the eligible voters. To register to vote, people had to be included in the 1991 Census, be 18 years old at the time of elections, or present other evidence that they were eligible to vote. Registration stations were established in all municipalities of BH and in many foreign countries. It was possible to register in the municipality of current residence, (i.e. residence in 1997 or 1998), that was different from the one where the person lived in before the war. At the same time, the voters could give their votes for the municipalities of their pre-war residence. The municipality of registration to vote was thus a good proxy for the actual place of living of the voters in 1997 and 1998. The registration to vote was voluntary, which implies that the registers exclude those who did not register because they were not interested, ill, too old or too young. Still, since countrywide at least 1 September 2008 5

75% of the eligible population registered for the elections, the results on IDPs and refugees presented in this report cannot be greatly weakened. The reason for this is simply that since such a large proportion of the population registered themselves, the errors which may be caused by people who did not register, are not large enough to seriously bias these results. Because of the voluntary character of the registration, Voters Registers cannot be used as a source on the overall population size in 1997 and 1998. In these years the population of Bosnia was certainly larger than the approximate 2,7 million voters covered in the Registers (probably around 3.5 or more million). The Registers can be, however, safely seen as a large sample of the population that survived the 1992-95 conflict in Bosnia and Herzegovina, including IDPs and refugees. Noteworthy, this population was aged 18 years or older at the time of elections; children below 18 years of age, who are not eligible to vote, are not represented in the Registers. The 1997-98 Voters Register is a large sample of the 1997-98 population of eligible voters of Bosnia and Herzegovina. All voters who registered to vote in 1997 and in 1998 are covered in this source. The Demographic Unit - OTP merged the two Voters Registers (1997 and 1998) in one (1997-98). The overlap of these two lists is large. Only about 150,000 records are new in 1998 (1 st registration in 1998). All other records reported in the 1998 Register are also covered in the 1997 Register. While merging the Registers, we included all records from 1997 (1 st registration in 1997) and additionally the new records from 1998 (150,000 records from the 1 st registration in 1998). In most cases, the 1998 records appeared to cover municipalities where the registration was less complete in 1997. The total size of the merged 1997-98 Voters Register is 2,674,506 records and it mainly covers the year 1997. Besides the variables such as the surname, first name, sex, date of birth, and mati~ni broj (JMB), four location items were registered: municipality of residence in 1991, municipality (and centre) of registration to vote in 1997 (or 1998), and municipality for which the person wanted to vote in 1997 (or 1998). In the 1998 Register, also the settlement of registration is available. The Voters Register has some of the same data quality problems as the 1991 Census, although not as many, especially misspelled names and missing or incomplete date of birth or mati~ni broj. The deficiencies are mostly due to optical scanning of the registration forms. The mati~ni broj was checked and found to be complete and valid for 83.7% of the 1997 voters from Vi{egrad. The names were checked and corrected with the computer programme and other procedures mentioned above in the section on the 1991 Census. There have been allegations that some people registered fraudulently to vote in the 1997 elections, by using false names (i.e. names of dead people). Brunborg and Urdal (2000) investigated this thoroughly for Srebrenica and found no evidence of massive fraud in the registration of voters in 1997 and 1998. For Srebrenica only 12 persons (out of about 7,661 missing) were found both in the lists of missing persons and in the 1997 and 1998 Voters Registers. 2 2 Helge Brunborg, Ewa Tabeau and Arve Hetland, 2005: Missing and Dead from Srebrenica: The 2005 Report and List. Expert report for the case of VUJADIN POPOVI] et al. (IT-05-88), 16 November 2005. ERN 0501-6180-0501-6209, Exhibit No. P02413. These 12 records were excluded from the 2005 Srebrenica list of missing. 1 September 2008 6

2.3 The ICRC List of Missing Persons for Bosnia and Herzegovina The ICRC started the registration of missing persons from the territory of Bosnia and Herzegovina (BH) already during the 1992-1995 conflict, primarily to register persons believed to be in detention. 3 The registration of the missing persons from Bosnia was one of the largest ICRC operations after the WWII. The National BH Red Cross/ Red Crescent Societies were involved in collecting tracing requests as well as and in public campaigns run in Bosnia and Herzegovina in order to increase the awareness of the society of the ICRC activities. The registration continued after the war until the present time, although recently at a much lower pace. In the year 2007, 40 additional tracing requests were registered. The work of ICRC in Bosnia and Herzegovina has so far resulted in the publication of eight editions of their list of missing persons (the 8 th edition published in 2007), as well as an addendum containing about 1,000 entries (published in 2000). The latest editions of the ICRC books (starting with the 4 th ) contained records of still missing persons as well as known deaths. In addition to publishing of these books, ICRC maintains a website where the names of still missing persons from Bosnia and Herzegovina are presented. The website, available at http://www.familylinks.icrc.org/mis_bos.nsf/bottin, is regularly up-dated. ICRC used a standard questionnaire for tracing requests. Thus, the same data items were collected for all missing: names (first, father s, family), date and place of birth, date and place of disappearance, who reported etc. Each missing person received a unique identification number (the so-called BAZ number). As respondents only close relatives or eye witnesses were accepted. Data was computerized with IT specialists and is available in a database format. Data was systematically provided from Sarajevo to Geneva for further processing, cross-referencing with other sources, and including on the web. Importantly, as a result of cross-referencing, ICRC systematically excluded closed cases both dead and alive, as well as administrative exclusions. The 2005 up-date of the ICRC list of missing persons for all of Bosnia and Herzegovina used for this report was provided directly by the Geneva Office of the ICRC on 17 August 2005 (ERN: D000-1714- D000-1714). The list sent to the OTP in August 2005 is broader than the web-based list of still missing only and includes some information about the body for those still missing and about persons who are not missing any more. The 2005 ICRC list provided to the OTP has five components: - still missing with information about the body not yet available (14,105 records); - still missing with information about the body already available (1,528); - ICRC closed cases, i.e. confirmed deaths (6,093); - alive persons, i.e. cases no more valid as part of the missing persons list (434); - administrative exclusions (52). Altogether these lists contain 22,212 records, of which 21,726 are related to still missing or dead persons and 486 are no more relevant. 3 Sources for this section include: Special Report by ICRC on The issue of missing persons from Bosnia and Herzegovina, Croatia and the Federal Republic of Yugoslavia, Feb 1998, ERN 0349-2128-0349-2143; ICRC Annual Report 2007- downloaded from the ICRC website. 1 September 2008 7

The 2005 ICRC list, as all previous editions of the list, includes data on surname, first name, father s name, sex, date and place of birth, and date and place of disappearance (reported as the place municipality ). It is noteworthy that even though ICRC obviously has improved their records throughout the years since the publication of their first list in 1996, empty or incomplete fields are still seen on the 2005 ICRC list. The most frequently incomplete items are date of birth (28.8 % incomplete; 6,403 incomplete DoB out of 22,212 records; but only 12 without year of birth) and date of disappearance (11.8 % incomplete; 2,624 incomplete out of all 22,212, but only one record without year of death). The other variables are recorded for almost everybody but that does not necessarily mean that they are always correct. Errors are seen in the spelling of names of persons and places. Moreover, from comparing several lists we know that there are errors, although mostly small, in variables such as date of birth. Such errors are common all over the world in data collected through questionnaires in surveys, censuses and elsewhere. It is, therefore, not surprising that there are errors in variables concerning tragic events collected in a chaotic and traumatic situation. In connection with a study of the number of missing persons from Srebrenica, the Demographic Unit OTP did a thorough analysis of the quality of the 2005 ICRC list. We found that the quality of the data is generally good and reliable, although there are some deficiencies in the data, as those mentioned above. The ICRC list is primarily a list of missing and not dead people. It is generally assumed, however, that most (if not all) of these people are dead. The ICRC itself also expressed this opinion 4. 2.4 Methods Our approach has been to match information about individuals from the 1991 Census with individual records from the lists of missing/dead persons and the OSCE Voters Register for the 1997 elections. When comparing various lists with data on individuals our approach has been to use the Access database program to search for records on one list that match records on the other list. If key variables are identical in two lists the matched records are assumed to represent the same person, otherwise not. This would have been a fast and easy procedure if all individuals on each list were uniquely determined by one or more variables, such as an ID number. However, this is not the case with all lists available to us. Although a unique ID number was introduced in Yugoslavia in 1981, it is not used by e.g. ICRC in their database. Moreover, when it is used, such as in the 1991 Census and the OSCE Voters Register, it is sometimes missing or inconsistent. The matching of two lists always began by searching for records with identical names and date of birth. It is very unusual that two different persons have identical names and are born on exactly the same date, especially if we are only considering the population of a limited area, such as a single municipality. Quite often, however, names are spelled differently or the date of birth is recorded slightly differently or missing altogether in one or both lists. Consequently, for persons not matched in the first round we made the search criteria gradually broader for one or more variables, for example by including only the year (and not the full date) of birth, or only the initial of the first name, in 4 See the ICRC s special report on The issue of missing persons from Bosnia and Herzegovina, Croatia and the Federal Republic of Yugoslavia, February 1998. ERN 0349-2128-0349-2143. 1 September 2008 8

addition to the surname. The results of such matches have, however, to be inspected visually to decide if the matches are likely to be of the same person or not, by looking at other available information, such as municipality and place of birth or residence. For example, the place of birth may be given as a municipality on one list and a small hamlet, located in the municipality, on the other list. It would be very complicated, if possible at all, to automate such checks. For difficult cases we checked the 1991 Census for additional information, e.g. information about family members of the person in question. To record the quality and basis for a match, a parameter (a quality indicator) was assigned to each matched person depending on the criteria used for the match. This parameter was used to study the number of accepted matches according to the type and quality of the match. We believe that the accuracy of this method is very good and that it yields reliable results. As a result of the matching process the DU-OTP was able to identify 10,522 survivors (i.e. registered voters from 1997 and 1998) 5 out of 21,198 individuals reported as living in Vi{egrad in the 1991 Population Census, i.e. 49.63%. Some changes are due to natural population movements before, during and after the war, in particular natural deaths and out-migrations. The largest group of people that we do not have any information about, are the eligible voters who did not register to vote. We also do not posses any post-war information about survivors from age 0 to 17 years. Thus, all estimates of survivors provided in this report are minimum numbers. The true figures are substantially higher. Many analyses presented in this report are made by ethnicity, obtained from exactly the same definition for both analyzed years, for 1991, for 1997-98, and also for the moment of disappearance. The definition I applied is the one used in the questionnaire of the 1991 Population Census, where ethnicity was a self-reported response to an open-ended question. In the original census forms, the citizens of Bosnia and Herzegovina mentioned about one hundred of ethnic categories. The Demographic Unit - OTP re-grouped these categories into four major clusters: those who reported themselves as Muslims, Croats or Serbs were regarded as members of these particular groups, all remaining categories, including Yugoslavs, were taken together as Others. With regard to the definition of internally displaced persons, the 1991 and 1997-98 municipality of residence were compared for each person studied. If an individual resided in 1991 in a different municipality than the municipality where he/she registered to vote in 1997-98 elections, than the person was considered internally displaced. Comparisons were made for post-dayton municipalities, which involved creating a new variable, post-dayton municipality, for all individuals reported in the Census. This task was successfully completed for the Visegrad area (including Visegrad and other municipalities studied in this report). It needs to be noted that internal migration in former socialist countries, such as Yugoslavia and, in particular, Bosnia and Herzegovina, was limited in the years until 1991. Our analysis of differences in the place of residence before and after the conflict is therefore fully justified as a method for assessment of population movements during the 1992-95 conflict. Pre-conflict internal migration in Bosnia and Herzegovina was negligible. Moreover the usual causes of internal migration (labour market, housing, education etc.) did not operate during the conflict. Poor housing was one of the 5 The term survivor as used here relates to records matched between the 1991 Census and the 1997-98 Voters Registers. More broadly, sources such as the BH government registration of IDPs and refugees can be used to identify survivors in addition to the voters registration. The resulting number is then 12,658, slightly higher than 10,870. 1 September 2008 9

reasons for low population mobility in Bosnia and Herzegovina before 1991. Also the urbanisation process was relatively slow in Bosnia when compared with dynamic Western countries. The process was controlled by the socialist party. Labour migration did not play much role as unemployment did not exist in the socialist system. Jobs were guaranteed for everyone. Making career was related to factors largely beyond individual ambition and readiness to move for a job. These factors were related to, for example, socialist party membership or employment policies of the leading party. The working age population of the former Yugoslavia, including Bosnia, mainly men, travelled, however, to Western European countries for temporary jobs and better income, but this temporary (external, not internal) migration returned systematically back home. Refugees were persons who in 1991 were reported in the population of Bosnia and Herzegovina (including those temporarily residing abroad) and who in 1997-98 registered to vote in countries different than Bosnia. There were approximately 300,000 out-of-country voters from Bosnia and Herzegovina who satisfied this criterion. Note that our definitions of internally displaced persons and refugees are statistical, not legal. As such the numbers of IDPs and refugees presented in this report should be seen as approximations of the actual true figures. Note also that obtaining the true figures is in our view an impossible task due to limited existing sources of information and fragmentary information contained in these sources. 1 September 2008 10

3. Changes in the Ethnic Composition in the Vi{egrad Region, 1991-1997 3.1 Vi{egrad Municipality, 1991-97 In this section I discuss changes in the ethnic composition in the Vi{egrad municipality by comparing the 1991 structure with the structure for 1997. In addition, a comparison between ethnic structures of selected settlements (i.e. administrative sub-units within the municipality) in the years 1991 and 1998 is provided. At the end of this discussion I place the changes that occurred in Vi{egrad in the context of similar changes in surrounding municipalities. Before the war the population of Vi{egrad increased systematically from about 18,923 in 1948 to 21,199 in 1991 (Table 1). The municipality was relatively thinly populated in 1991 (47.3 persons per 1 sq.km), much of the population lived in the town of Vi{egrad. In 1991, as in 1981, the majority ethnic group was the Muslims (63.5% in 1991 and 62.1% in 1981). Serbs were the second largest group; their share was 33% and 31.8% in 1981 and 1991 respectively (Table 2). Table 1. Total Population of Vi{egrad Since 1948 1948 1953 1961 1971 1981 1991 18,923 21,566 24,557 25,389 23,201 21,199 Source: Stanovnistvo Bosne i Hercegovine. CROSTAT, Zagreb, Travanj 1995 Table 2. Ethnic Composition of Vi{egrad in 1981, 1991, and 1997 Year Total Croats Muslims Serbs Others 1981 23,201 60 14,397 7,648 1,096 1981 100.0 0.3 62.1 33.0 4.7 1991 21,199 32 13,471 6,743 953 1991 100.0 0.2 63.5 31.8 4.5 1997(18+) 9,241 60 3 8,861 317 1997 (18+) 100.0 0.6 0.0 95.9 3.4 Source: For 1981 and 1991: Stanovni{tvo Bosne i Hercegovine. CROSTAT, Zagreb, Travanj 1995. For 1997: The OSCE Voters Register Table 2 also shows the ethnic composition of Vi{egrad in 1997. The figures for 1997 do not entirely correspond to those for 1981 and 1991. For 1997 only the population at age 18 years or more is included, i.e. the eligible voters, while for 1981 and 1991 all age groups are shown. Moreover, the 1997 voters represent a sample of the whole population, while the 1981 and 1991 figures cover all citizens living in this area. Despite these deficiencies, the 1997 figures give a good impression of the ethnic structure in 1997. As we can see from Table 2, for Vi{egrad as a municipality, the pre-war majority group, Muslims, were entirely gone in 1997. The proportion of Serbs was substantially higher in 1997 than in 1991, whereas the remaining shares of Croats and Others stayed at similar levels as 1 September 2008 11

before the war. The Serbs became not only the dominant ethnic group in Vi{egrad, but also almost exclusively the only group remaining in this municipality. In order to produce a more accurate picture of changes in the ethnic composition it is necessary to analyse the population at age 18 or more years for both considered years. This is reported below. Figure 1a. Ethnic Composition in Vi{egrad: Pre- and Post-war Population 100 80 60 40 20 0 Serbs Muslims Croats Others 1991 1997 Figure 1b. Registered 1997 Voters Originally from Vi{egrad by Ethnicity and Place of Registration 100 Serbs 100 Muslims 80 80 60 60 40 40 20 20 0 This Mun. Other Mun. Out of country 0 This Mun. Other Mun. Out of country Croats Others 100 100 80 80 60 60 40 40 20 20 0 This Mun. Other Mun. Out of country 0 This Mun. Other Mun. Out of country The 1991 population of Vi{egrad was estimated at 21,199 individuals (Table 2), out of which 17,883 were at age 18 or more years in 1997 (Table 3a) and were eligible to vote. The eligible voters are the subjects of all analyses following in the next sections of this report. 1 September 2008 12

The population of Vi{egrad was in 1991 dominated by Muslims (about 62.5%; Figure 1a and Table 3a). The 1997 ethnic composition of the Vi{egrad population, as estimated from our sample of 9,241 registered voters, was completely different than the 1991 composition, i.e. it became dominated by Serbs in 95.9 per cent. Figure 1b and Table 3b further confirm that practically all Muslim voters who lived in Vi{egrad in 1991, registered to vote in 1997 in other municipalities in Bosnia and Herzegovina or abroad (some 6,798 or almost all out of the 6,799 registered). On the contrary to Muslims, Serbs from Vi{egrad registered mainly in the Vi{egrad municipality (3,704 or 96.9% out of 3,822 registered) in 1997. Croats were weakly represented among the 1997 voters, which may indicate that this group was largely absent in the area of Vi{egrad in 1997. Also Others were not numerous. Some 63% of Others registered to vote in 1997 outside the borders of the Vi{egrad municipality (22% in other municipalities in BH and 41% abroad). Table 3a. Ethnic Composition in Vi{egrad: Pre- and Post-war Population (18 Years of Age or Older) Ethnicity Absolute numbers Percentages 1991 Census 1997 Voters Register 1991 Census 1997 Voters Register Serbs Muslims Croats Others 5837 8861 11178 3 30 60 838 317 32.64 95.89 62.51 0.03 0.17 0.65 4.69 3.43 Total 17883 9241 100.00 100.00 Table 3b. Registered 1997 Voters Originally from Vi{egrad by Ethnicity and Place of Registration Ethnicity This municipality Absolute numbers Other municipalities Out of country Total Serbs Muslims Croats Others 3704 1 13 79 34 4717 0 47 84 2081 3 87 3822 6799 16 213 Ethnicity This municipality Percentages Other municipalities Out of country Total Serbs Muslims Croats Others 96.91 0.89 2.20 0.01 69.38 30.61 81.25 0.00 18.75 37.09 22.07 40.85 100.00 100.00 100.00 100.00 1 September 2008 13

The largest group of the internally displaced population from the Vi{egrad municipality were the Muslims. There were in total 4,717 registered voters whom I identified as registered in municipalities different than Vi{egrad. The 1997 place of registration of the displaced Muslim population is shown in Figure 1c. The Muslim population moved to areas located in the Federation of Bosnia and Herzegovina. Many Muslims fled to the Federal part of the Gora`de municipality, and to the Federal parts of Sarajevo (Centar, Stari Grad, Novi Grad, Novo Sarajevo, Ilid`a, Ilija{, Had`i}i, Vogo{}a) and surrounding municipalities such as Visoko, Kakanj, and Zenica. Figure 1c. Population Displacement from Vi{egrad (RS): Pre- and Post-war Population Serbs Muslims Region of origin Visegrad (RS) Size of population flow 0% - 0% 0% - 2.9% 2.9% - 5.9% 5.9% - 11.8% 11.8% - 29.4% Region of origin Visegrad (RS) Size of population flow 0% - 0.5% 0.5% - 3.1% 3.1% - 6.7% 6.7% - 17.3% 17.3% - 28.3% Others Region of origin Visegrad (RS) Size of population flow 0% - 0% 0% - 2.1% 2.1% - 4.3% 4.3% - 10.6% 10.6% - 25.5% Our results shown on the map above are very much in accordance with the distribution of displaced persons reported in 1998 by the UNHCR in Sarajevo (Table 4). 1 September 2008 14

Table 4. Distribution of Internally Displaced Persons from Vi{egrad Reported by UNHCR, 1998 Municipality Number of DPs Municipality Number of DPs Tuzla 139 Sarajevo Had`i}i 255 Breza 50 Sarajevo Ilid`a 601 Kakanj 129 Sarajevo Ilija{ 244 Visoko 853 Sarajevo Novi Grad 3576 Zenica 616 Novo Sarajevo 906 Fojnica 102 Sarajevo Stari Grad 227 Sarajevo Centar 516 Sarajevo Vogo{}a 384 Total 8598 Source: Raseljena lica po opcinama i kantonima (regijama) prebivali{ta i boravi{ta. Stanje: 31.12.1998. UNHCR Sarajevo, 1998. Table 5. Emigration from Vi{egrad, 1991-1997 Ethnicity Serbs Muslims Croats Others Croatia Number Percent Outmigration structure (out of country voters) FRY Number Percent Other countries Number Percent Total Number Percent 0 0.00 73 86.90 11 13.10 84 100.00 1 0.05 19 0.91 2061 99.04 2081 100.00 0 0.00 0 0.00 3 100.00 3 100.00 0 0.00 15 17.24 72 82.76 87 100.00 Source: The 1997 Voters Register and the 1991 Population Census for BH Table 5 summarises the emigration (i.e. out-migration) from the municipality of Vi{egrad. It is striking that out of 2,255 persons who left the municipality and in 1997 still lived in locations different from their pre-war place of residence which in addition were outside the borders of Bosnia and Herzegovina, the largest group (some 2,081) were Muslims. About 2,061 (99%) of all Muslim migrants went obviously to countries outside of the area of the former Yugoslavia. The remaining out-migration from Vi{egrad was of Serbs and Others (84 and 87 individuals respectively). Some 87% of the Serbs moved to FRY (practically to Serbia) and 83% of Others moved to countries outside the former Yugoslavia. 3.2 Changes in the Ethnic Composition in Selected Settlements in Vi{egrad, 1991-1998 In 1991 there were 159 settlements (administrative sub-units) in the Vi{egrad municipality. Most of them (93) had a clear Muslim majority (well above 50% Muslims), whereas 60 had a clear Serb majority. Only six settlements could be considered as mixed, i.e. having the difference between the 1 September 2008 15

shares of two major ethnic groups less than 15%. These settlements include Vi{egrad town and the villages: Koritnik, Pijavice (with hamlet Uzamnice), Prelovo, Sase and Velji Lug (with hamlets Kr~evine, Omar, [ipovac and @ili}i). The detailed list of all settlements in Vi{egrad with their ethnic composition in 1991, as reported in the Population Census, is attached in Annex B. The Annex also includes the estimated ethnic composition of selected settlements in 1998, based on the 1998 Voters Register (only for 1998 the information about voters settlements is provided). Note however, that as the Voters Register is only a sample of the post-war population, only those settlements are shown, for which the sample size is large enough, i.e. the number of observations exceeds 50. This condition is met for 15 settlements (Vi{egrad town and some larger villages, the full list is provided in Annex B). From this example it can be clearly seen that during the period 1991-1998 all of them became almost exclusively Serb, regardless their pre-war ethnic composition. 3.3 Selected Municipalities in the Vi{egrad Region, 1991-97 The objective of the analysis presented in this subsection is to compare the changes of the ethnic composition in Vi{egrad with the changes that occurred in the surrounding municipalities. The municipalities selected to study all entirely belong to Republika Srpska since 1995. These municipalities are: Bratunac, ^ajni~e, Han Pijesak, Mili}i, Rogatica, Rudo, Sokolac, Srebrenica and Vlasenica. Gora`de municipality which was split by the Dayton line and since 1995 consists of two parts (Gora`de-FBH and Srpsko Gora`de-RS), has been excluded from this study. I excluded Gora`de for the reason that in all split municipalities extremely large population movements occurred between the Serb and Federal areas. Such extreme changes are not directly comparable with the changes that occurred in non-split municipalities. It is also worth noting that Han Pijesak, Mili}i, (part of) Rogatica, Sokolac and Vlasenica belonged to the so-called Serbian Autonomous District of Romanija-Bira}, whereas ^ajni~e and Rudo to the Serbian Autonomous District of Herzegovina (as defined by the Assembly of Serbian People in BH on 21.11.1991). However, the municipalities with a considerable Muslim majority (Srebrenica, Vlasenica, Vi{egrad and part of Rogatica) were not assigned to any of the Serbian Autonomous Districts. To measure the size of changes in the ethnic composition in the mentioned municipalities and to 2 arrange the municipalities in a rank order, a statistical measure called the statistic (chi-square) was 2 used. The interpretation of this measure is straightforward: the bigger the value of, the more 2 dramatic changes in the ethnic composition occurred. On the basis of the calculated, a statistical test can be performed, indicating whether the obtained results show the true (significant) pattern or 2 could they have been obtained by chance. The values of the statistic showing the rank order of municipalities around Vi{egrad in terms of the size of ethnic changes are presented in Table 6. The statistical methodology underlying all the calculations and inference is described in Annex C. 1 September 2008 16

Table 6. Rank Order of Selected Municipalities in the Vi{egrad Region due to the 1991-1997 Change in Ethnic Composition Post-Dayton Municipality Serbs Serbs Muslims Muslims Croats Croats Others Others Chi-square 91 97-98 91 97-98 91 97-98 91 97-98 Statistic P-Value 105 - Srebrenica 24.7 96.3 73.2 0.1 0.1 0.5 2.0 3.1 21027.7 0.00000 104 - Bratunac 36.4 97.0 61.8 0.1 0.1 0.4 1.7 2.5 17744.7 0.00000 147 - Vi{egrad 32.6 95.9 62.5 0.0 0.2 0.6 4.7 3.4 17254.7 0.00000 146 - Rogatica 38.8 97.7 59.4 0.0 0.1 0.3 1.7 1.9 10812.5 0.00000 103 - Vlasenica 39.1 96.8 57.9 0.2 0.2 0.4 2.8 2.6 10555.3 0.00000 121 - Sokolac 66.1 97.9 32.5 0.0 0.1 0.3 1.2 1.8 4086.0 0.00000 185 - Mili}i 49.0 98.8 49.2 0.0 0.1 0.1 1.7 1.1 3770.0 0.00000 169 - ^ajni~e 53.0 97.3 44.5 0.0 0.1 0.1 2.5 2.6 2711.6 0.00000 170 - Rudo 70.1 97.5 27.5 0.0 0.1 0.2 2.4 2.3 1935.5 0.00000 123 - Han Pijesak 58.7 97.4 39.2 0.0 0.1 0.2 2.0 2.5 1620.5 0.00000 It can be seen that the most dramatic changes occurred in Srebrenica, Bratunac and Vi{egrad. Taking into consideration that in Srebrenica and Bratunac the changes are mainly related to a single event the fall of Srebrenica enclave in July 1995 (see also Section 4.2), it appears that Vi{egrad experienced the most dramatic changes in ethnic composition among the municipalities in the region. Moreover, the tests for all the municipalities proved that all changes are statistically significant, with the probability of obtaining such results by chance being practically equal 0. Summing up, changes in the ethnic composition in Vi{egrad reflect the pattern observed overall in the whole region, which became almost entirely Serb after the war. Finally, the intensity of these changes in Vi{egrad is substantially higher than of those in the surrounding municipalities (except Srebrenica and Bratunac). 1 September 2008 17

4. Basic Demographic Distributions of the Missing Persons in Vi{egrad 4.1 Vi{egrad Municipality In the ICRC list of missing persons (2005 release) I have found 705 individuals who disappeared in the Vi{egrad municipality. Of these persons, 560 have been identified in the 1991 Population Census (79.4%). 6 For the persons, whose identity was confirmed by linking with the respective Census records, I could provide basic demographic distributions such as the distribution by sex, time and place of disappearance and ethnicity. Of the persons identified, 542 declared themselves in 1991 as Muslims (96.8%) and 18 as Others Yugoslavs, other ethnicities and undeclared (3.2%). The basic demographic distributions are shown below in Figures 2 through 4 and in Tables 7 and 8. The complete list of persons missing in Vi{egrad municipality, being an excerpt from the 2005 ICRC list, is attached in Annex A. Figure 2. Number of Persons Missing in Vi{egrad, by Sex, Age and Ethnicity males 75+ 60-74 females 45-59 30-44 15-29 0-14 200 150 100 50 0 50 100 Muslim Other Unknow n Source: ICRC List of Missing Persons (2005) and 1991 Population Census for BH Table 7. Number of Persons Missing in Vi{egrad, by Sex, Age and Ethnicity Men Women Age Group Muslim Other Unknown Total Muslim Other Unknown Total 0-14 23 0 5 28 15 0 4 19 15-29 106 4 19 129 12 5 8 25 30-44 148 2 27 177 28 0 10 38 45-59 80 2 21 103 28 1 11 40 60-74 68 1 16 85 20 2 11 33 75+ 10 0 2 12 4 1 11 16 Total 435 9 90 534 107 9 55 171 Note: 1 Serb male, age 15-29, is included in "Others" Source: ICRC List of Missing Persons (2005) and 1991 Population Census for BH 1 September 2008 18

Figure 3. Number of Persons Missing in Vi{egrad, by Ethnicity and Month of Disappearance 350 300 250 200 150 100 50 0 Apr 92 Jun 92 Aug 92 Oct 92 Dec 92 Feb 93 Apr 93 Jun 93 Aug 93 Oct 93 Dec 93 Feb 94 Apr 94 Jun 94 Aug 94 Oct 94 Dec 94 Feb 95 Apr 95 Jun 95 Aug 95 Muslim Other Unknown Source: ICRC List of Missing Persons (2005) and 1991 Population Census for BH Figure 4. Number of Missing Persons from Vi{egrad, by Ethnicity and Date of Disappearance (in the Period April-June 1992) 35 30 25 20 15 10 5 0 1 April 5 April 9 April 13 April 17 April 21 April 25 April 29 April 3 May 7 May 11 May 15 May 19 May 23 May 27 May 31 May 4 June 8 June 12 June 16 June 20 June 24 June 28 June Muslim Other Unknown Source: ICRC List of Missing Persons (2005) and 1991 Population Census for BH 6 Records that remain unmatched are characterized by misspelling and incompleteness. Manual searching for these records in the 1991 Census one by one would likely bring more matches. Some records might be, however, too poor to be declared matched. 1 September 2008 19

Table 8. Distribution of Missing Persons in Vi{egrad, by Place of Disappearance (Only Places with 5 or More Missing Persons Are Shown) PLACE OF MISSSING NO. MISSING VISEGRAD 396 BIKAVAC 28 DRINSKO 21 DOBRUN 20 DUSCE 18 RODIC BRDO 14 MUSICI 13 PRELOVO 12 VELIKA GOSTILJA 12 SASE 10 MALA GOSTILJA 9 BAN POLJE 9 KLASNIK 8 KABERNIK 7 VUCINE 7 MEDEDA 6 KORITNIK 6 BABIN POTOK 6 BOSANSKA JAGODINA 6 SEGANJE 6 BARIMO 5 ZLATNIK 5 OTHER PLACES 81 ALL PLACES 705 Source: ICRC List of Missing Persons (2005) From Figure 2 and Table 7 it can be seen that the vast majority of persons missing in Vi{egrad were Muslim men, mainly aged 15 to 44 years (younger military age). Figures 3 and 4 show that the process of going missing was most intense in May and June 1992, especially around May 25 th, June 14 th and June 20 th 1992. Moreover, from Figure 3 it can be clearly seen that the disappearances of persons in Vi{egrad took place almost exclusively in late spring and early summer 1992. Table 8 shows that from all the places where people went missing in the municipality, most important was the town of Vi{egrad, where 396 persons disappeared (56%). 4.2 Selected Municipalities in the Vi{egrad Region The aim of the review presented in this subsection is to compare the number of persons missing in Vi{egrad with the numbers of persons who disappeared in the surrounding municipalities. The list of municipalities taken into account is the same as in the analysis presented in Section 3.3. For all these municipalities (including Vi{egrad) I have found that 9,719 individuals have been reported to the ICRC as missing during the war. Of these persons, 83.8% have been identified in the 1991 Population Census, out of those 98.1% had declared themselves in 1991 as Muslims. The detailed figures by municipality are presented in Table 9. 1 September 2008 20

Table 9. Distribution of Missing Persons in and around Vi{egrad, by Ethnicity and Municipality of Disappearance Municipality of Disappearance Croats Muslims Serbs Others Unknown Total Percent in 1991 Census Percent Muslims in 1991 BRATUNAC 2,972 15 28 489 3,504 86.0 98.6 CAJNICE 67 25 92 72.8 100.0 HAN PIJESAK 47 1 1 7 56 87.5 95.9 ROGATICA 2 353 7 4 102 468 78.2 97.0 RUDO 52 40 92 56.5 100.0 SOKOLAC 1 95 1 6 33 136 75.7 93.1 SREBRENICA 1 3,144 24 28 578 3,775 84.7 98.4 SREBRENIK 3 1 4 75.0 0.0 VISEGRAD 542 1 17 145 705 79.4 96.8 VLASENICA 1 713 5 11 157 887 82.3 97.8 TOTAL 5 7,985 57 95 1,577 9,719 83.8 98.1 Source: ICRC List of Missing Persons (2005) and 1991 Population Census for BH To eliminate the bias caused by uneven timing of the incidents in the municipalities around Vi{egrad (especially related to the fall of Srebrenica enclave in July 1995, what is reflected by the figures for Srebrenica and Bratunac), the totals from Table 9 are further broken down by the year of disappearance as shown in Table 10. Table 10. Distribution of Missing Persons in and around Vi{egrad, by Year and Municipality of Disappearance Municipality of Disappearance 1992 1993 1994 1995 Total BRATUNAC 372 43 4 3,085 3,504 CAJNICE 90 2 92 HAN PIJESAK 41 3 2 10 56 ROGATICA 343 6 6 113 468 RUDO 71 21 92 SOKOLAC 127 6 3 136 SREBRENICA 181 54 25 3,515 3,775 SREBRENIK 4 4 VISEGRAD 684 4 17 705 VLASENICA 619 66 6 196 887 TOTAL 2,528 205 50 6,936 9,719 Source: ICRC List of Missing Persons (2005) In 1992, out of the ten selected municipalities, most missing persons disappeared in Vi{egrad (684). The figures for 1992 are further shown on monthly basis in Table 11. 1 September 2008 21