International Criminal Tribunal for the Former Yugoslavia

Page 8386

1 Monday, 23 September 2002

2 [Open session]

3 [The accused entered court]

4 --- Upon commencing at 9.48 a.m.

5 JUDGE SCHOMBURG: Good morning, everybody. May I ask Madam

6 Registrar, please call the case.

7 THE REGISTRAR: Good morning. This is case number IT-97-24-T, the

8 Prosecutor versus Milomir Stakic.

9 JUDGE SCHOMBURG: Thank you. And the appearances, please.

10 MR. KOUMIJIAN: Nicholas Karper and -- Nicholas Koumjian and Ruth

11 Karper for the Prosecution.

12 JUDGE SCHOMBURG: I just wanted to congratulate.

13 MR. OSTOJIC: Good morning, Your Honour. John Ostojic on behalf

14 of Dr. Milomir Stakic. Thank you.

15 JUDGE SCHOMBURG: So I don't have the chance to congratulate

16 Mr. Lukic, having been elected as a secretary of the Defence Union, but

17 good to know that we have also a secretary in our room from now on.

18 And first of all, I am more than happy to know that, as you all

19 can see, once again we are sitting together in the composition of three

20 Judges, but I think it's necessary to start in an acceptable way for all

21 of us, and let us first clarify how to proceed today. It will not be

22 possible for Judge Fassi Fihri to attend the afternoon meeting, and there

23 are two possibilities: Either the parties accept, that, only for this

24 afternoon, we proceed under 15 bis, or we would proceed in the framework

25 of a Status Conference, as we did earlier, and then try to have an

Page 8387

1 overview on this overwhelming material we have received until Friday, late

2 in the evening, and then to introduce this in a kind of package solution

3 into the case, as we did it earlier.

4 I think the most facilitating way would indeed be to proceed with

5 some matters under Rule 15 bis, and then we would have to decide what to

6 do.

7 May I hear the observations by the parties, please.

8 MR. KOUMJIAN: Your Honour, we would hope that we could finish

9 Ms. Tabeau this morning. I hope that's not too optimistic. But if that's

10 the case, than I think it might be better for Mr. Sebire, who is a very

11 short witness - and actually, his preference is to further prepare - that

12 we could call him on Thursday and that we -- but if we have to proceed

13 this afternoon with Ms. Tabeau, I don't know if it's possible to break it

14 up into a trial and 15 bis procedure or just finish her tomorrow. But I

15 know Mr. Ostojic indicated -- I'm having trouble with names today --

16 indicated that he would like to talk to Dr. Stakic sometime.

17 MR. OSTOJIC: That's correct, Your Honour. We'll proceed in the

18 manner best for the Court and the honourable Judge Fassi Fihri. We did

19 receive today a copy of Madam Tabeau's report in B/C/S. I've tendered it

20 to Dr. Stakic. We are prepared to proceed and question her. I don't

21 believe that the questioning will take more than an hour and a half to

22 two, but again it all depends on the rapport that we can develop with her

23 during the question and answer session. I'd like to have a staff

24 conference. There are a number of issues that I'd like to discuss with

25 the Court and I think it's necessary both for purposes of scheduling as

Page 8388

1 well as projecting the next several witnesses and how we hope to proceed.

2 JUDGE SCHOMBURG: So I think it's a clear message that we proceed

3 with the hearing only this morning and then with the witnesses to come

4 tomorrow and the other days, and leave this afternoon for all these

5 procedural issues and try to start introducing and attaching exhibit

6 numbers to the new incoming documents. And if it's convenient, then

7 present it as a kind of package solution, once again, into the hearing

8 tomorrow.

9 Can we --

10 MR. KOUMJIAN: I just wanted to inform Your Honours of one other

11 matter, is that tomorrow there is a representative of the government in

12 question regarding the next witness planning to fly here and be present,

13 so if we don't finish for some reason the witness, we would ask to proceed

14 out of order with Mr. Brown so that to accommodate that special

15 representative.

16 MR. OSTOJIC: We're prepared, so we have no objection to that,

17 Your Honour.

18 JUDGE SCHOMBURG: Thank you. I think we have to be flexible in

19 the next five days and we proceed in the spirit shown until now, I have no

20 doubts that we will succeed. So therefore, for the registry, we proceed

21 with our hearing only until 1.00 this morning, and start at 2.30 a Status

22 Conference.

23 And then the parties may know that also tomorrow we sit in Trial

24 Chamber -- in courtroom 1. On Wednesday, we'll sit also the entire day in

25 courtroom 2, from 9.30 until 4.00. Yes, about 4.00. But the entire day

Page 8389

1 courtroom 2. This intimate atmosphere will facilitate the hearing maybe.

2 Then it's mandatory under Rule 70(E) to provide a report, and this

3 report will be filed, but for the reasons given earlier, I want to read

4 out slowly this report now in my capacity as Presiding Officer on the last

5 deposition taken.

6 It reads: "Report on deposition evidence under Rule 71(E) of

7 Rules on Procedure and Evidence to Trial Chamber II. Composed of Judge

8 Wolfgang Schomburg, presiding, Judge Mohamed Fassi Fihri,

9 Judge Volodymyr Vassylenko.

10 "On 11 September 2002, Trial Chamber II, acting under Rule 15 bis

11 of the Rules of Procedure and Evidence, in the absence of Judge Fassi

12 Fihri and with the consent of the parties and the accused, ordered that

13 deposition evidence pursuant to Rule 71(A) of the Rules be taken in

14 respect of certain witnesses. I was appointed Presiding Officer for that

15 purpose, and Judge Vassylenko was permitted to be present in the

16 courtroom. The procedure was to cease upon the return of Judge Fassi

17 Fihri to active duty.

18 "In accordance with the order for deposition evidence to be taken

19 pursuant to Rule 71 of 11 September 2002, the following depositions were

20 taken: Edward Vulliamy, from 16 to 18 September 2002; Dusan Baltic, 18 to

21 19 September 2002.

22 "A record was made of the depositions in the form of full

23 transcripts which have been transmitted to Trial Chamber and are to be

24 incorporated into the trial record. There were no objections raised

25 during the depositions by other party for decision by the Trial Chamber.

Page 8390

1 Documents tendered during the depositions were assigned provisional

2 reference numbers, and it is proposed that these be admitted with exhibit

3 numbers as follows: Edward Vulliamy, DP2, S 328, three pages

4 of notes; DP3, S329, video V000-0664; DP3-1, S329-1, transcript of

5 video; DP4, S330, video V000-0401; DP5-1 to DP5-8, S331-2 to S331-8,

6 these are notes regarding Mr. Kovacevic. DP6-1 to DP6-7, S332-1 to

7 S332-7, these are notes regarding Dr. Stakic. DP7, S333, video

8 V000-0662. DP8, S334, chapter 5 of "Seasons in Hell." DPD1, D25

9 "Shame of Omarska." DPD2, D26, pages 8 and 9 and 100 to 101 of the book

10 "Seasons in Hell." DPD3, D27, the article "Face to face with victims of

11 his horror." DPD4, D28 aide memoire, February 1996 meeting. Dusan

12 Baltic: DPJ1 --" maybe here there's a mistake. J 3. Shouldn't it read

13 something else? Document with 65 ter number 271, Madam Registrar.

14 THE REGISTRAR: It was labelled DPJ1 initially.

15 JUDGE SCHOMBURG: Yes. And then the next available J number would

16 be?


18 JUDGE SCHOMBURG: Not 1, but 13.

19 "DP9, S2-1, a map. DP10, S335, letter to Simo Drljaca, signed

20 by Mr. Baltic; DP11, S336, minutes from Crisis Staff, 13 to 14 May

21 1992."

22 May I ask the parties: Are there general observations on this

23 document? And especially because now under the Rules it's the time for

24 the question whether or not documents be admitted into evidence. Are

25 there any objections to the suggested proceedings? OTP, please.

Page 8391

1 MR. KOUMJIAN: No, Your Honour.

2 JUDGE SCHOMBURG: Defence, please.

3 MR. OSTOJIC: Generally we do not have an objection. I do have a

4 comment, if I may, Your Honour, or an observation. One of the exhibits,

5 at least from what I was able to follow, on the DPD list seems not to be

6 included in that, from memory, is the February 2002 article that was

7 written by Mr. Vulliamy in the Bosnian institute journal. I believe that

8 we asked some questions about that and that it was actually tendered, so

9 I'm just not sure in following the Court respectfully whether that's

10 included. I note that the Court did include the 1992 and 1996 articles

11 but we had, I believe at that time, moved for that article as well for

12 admission.

13 JUDGE SCHOMBURG: Correct. May I ask Madam Registrar: Maybe

14 there was an omission at that point in time, because we discussed it

15 whether or not it was an article from the Bosnian institute or, better, a

16 reprint from article having appeared earlier. Was it in the Bosnian

17 journal? I don't know. But no doubt it was. Could the parties agree

18 that -- in order that we don't have to repeat this entire report, you

19 tender this document now and the OTP agrees that this document be admitted

20 into evidence?

21 MR. KOUMJIAN: We have no objection. I think it was a reprint

22 from a Guardian article.

23 JUDGE SCHOMBURG: Sorry. From Guardian, yes, right. Yes. We all

24 know what we are discussing, and I can see Madam Registrar has a report

25 already available. Could we please hear what would be the number of this

Page 8392

1 exhibit.

2 THE REGISTRAR: It will be D29, Your Honour.

3 JUDGE SCHOMBURG: D29. And please both parties, don't hesitate,

4 if you identify that something is not included, let me immediately know

5 and then we proceed on this basis. But this is not -- forms not part of

6 this report. It's now admitted today into evidence, and it will be

7 distributed.

8 Judge Fassi Fihri in part is already and will be later on totally

9 informed when he's had the time to read all the documents and summaries

10 and transcripts on what was going on, both under Rule 71 and under Rule 15

11 bis, that as soon as possible he also will be aware of we discussed during

12 the last three weeks and what was the outcome. Are there any further

13 objections or observations to this report?

14 MR. OSTOJIC: No, Your Honour.

15 MR. KOUMJIAN: No, Mr. President.

16 JUDGE SCHOMBURG: Thank you. Then we can proceed this way: This

17 report will be filed and, as mentioned, if you take it or regard it as

18 necessary, please tell me when it's on the admission of additional

19 evidence. We'll do it this way. If one of the parties does regard it as

20 necessary to comment on the one or other evidence taken in the absence of

21 Judge Fassi Fihri, it's the time today or tomorrow to address this issue

22 or give comments on this evidence in order to have also available for

23 Judge Fassi Fihri not only the assessment of the colleagues and the legal

24 officers but also your observations and your assessments of the evidence

25 taken in the absence of Judge Fassi Fihri.

Page 8393

1 THE REGISTRAR: Sorry to interrupt. There is a confusion. The

2 article from the Bosnian institute was already marked, D27. So for the

3 record, D29 is not attributed to this article.


5 THE REGISTRAR: Yes, Your Honour.

6 JUDGE SCHOMBURG: The article, "Face to face with the victims of

7 his horror." This would be in fact this Bosnian institute article. Do

8 you agree?

9 MR. OSTOJIC: I would agree. I have no reason to doubt Madam

10 Registrar. We'll go over it, if we may, because there were actually four

11 particular items, so if I would just have that as an option during the

12 break and then we'll clarify. Thank you, Your Honour.

13 JUDGE SCHOMBURG: We have an entire week on these issues.

14 Anything else to be changed what we know already? No? I can see

15 no -- so then may I ask the usher to escort today's witness into the

16 courtroom.

17 Can we agree that this is regarding Ms. Tabeau from now on as a

18 new witness. This is not a continuation of the former statement;

19 correct?

20 MR. KOUMJIAN: Correct

21 [The witness entered court]

22 JUDGE SCHOMBURG: Good morning, Madam Tabeau. Can you hear me.



25 JUDGE SCHOMBURG: As this is not a continuation of your former

Page 8394

1 testimony in this case, may I ask first of all that we please hear your

2 solemn declaration.

3 THE WITNESS: I solemnly declare that I will speak the truth, the

4 whole truth, and nothing but the truth.


6 JUDGE SCHOMBURG: Thank you very much. Please be seated.

7 Mr. Koumjian

8 Examined by Mr. Koumjian:

9 Q. Good morning, Dr. Tabeau. You've previously testified and have

10 told the Court what your qualifications and training and experience is; is

11 that correct?

12 A. Yes.

13 Q. Did you in your work for the Office of the Prosecutor prepare a

14 report entitled "Basic demographic characteristics and socio-economic

15 status of missing and killed persons from Prijedor"?

16 A. Yes.

17 Q. 30 April through 30 September 1992?

18 A. Yes.

19 Q. And is that report dated the 9th of September of 2002?

20 A. Yes.

21 Q. And I believe the registry has a copy of that. May that be given

22 an exhibit number. We will tender that report.

23 THE REGISTRAR: This is exhibit S 337.


25 Q. Doctor, just to remind you, because of the translation, and

Page 8395

1 particularly since we're talking about some technical matters, if you

2 would pause after I've finished a question before you begin your answer,

3 so that the interpreters can catch up.

4 Dr. Tabeau in going over your report that has been submitted, is

5 it correct that you have identified certain areas that you wish to have

6 corrected in the report?

7 A. Yes.

8 Q. Starting at page 3 in the section entitled "2.1, Absolute

9 Numbers"?

10 A. Yes, this is the section.

11 Q. And specifically, the second paragraph lists certain figures for

12 the ICRC for what's listed as KNP, or the Prijedor Book of Missing Persons

13 and for the Exhumations and Proof of Death database; is that correct?

14 A. Yes. These numbers should be corrected [sic].

15 Q. Please, again pause a few seconds before you begin your answer.

16 The ICRC numbers are correct; is that true?

17 A. Yes. These numbers are correct.

18 Q. For the Book of Missing Persons, the KNP, can you read the

19 correct figures?

20 A. The first figure should be 1.088 instead of 1.092, and the second

21 figure should be 1.953 instead of 1.122.

22 Q. Can you give the correct figures now for the Exumations and Proof

23 of Death database, the EXH figures?

24 A. The first figure should be 638 instead of 960, and the second

25 figure should be 1.122 instead of 1.953.

Page 8396

1 Q. So in fact, of the total linked records, you simply had -- you had

2 misplaced -- you had put the exhumation records for the Book of Missing

3 Persons and vice versa, and those need to be reversed; is that correct?

4 A. Yes. This is one correction. But also the numbers, total numbers

5 reported for KNP and for exhumations are slightly different, the new

6 numbers, the other totals were totals for the whole 1992.

7 Q. Now, in this paragraph it lists these totals, but further on in

8 your report you use figures to make your calculations. Is it correct that

9 the figures you used to make your calculations are the correct figures,

10 the numbers you just gave, and that they simply were listed incorrectly in

11 this second paragraph under 2.1?

12 A. Yes, this is correct. The analysis is based on correct numbers

13 and only in the discussion in the text some mistakes were made.

14 Q. On page 7 of your report there appear two figures, figures 2 and

15 3, and they indicate that these are rates per 100.000 population in ARK.

16 Do these figures relate to the Autonomous Region of Krajina or to

17 Prijedor?

18 A. Both figures are made for Prijedor. The reference population is

19 the one of Prijedor and not the one of the Autonomous Region of Krajina.

20 Q. So to correct that, would it be correct on the headings for

21 figures 2 and figures 3 to replace the word "ARK" with "Prijedor"?

22 A. Yes, this is correct.

23 Q. Going now to page 11 of your report, in the paragraph -- I believe

24 it's the third paragraph, towards the bottom, you list categories of

25 occupations, and the sentence reads: "It allowed us to classify the

Page 8397

1 population among the following five categories," and then you listed four

2 categories. Was one category missing from that list?

3 A. Yes. One category is missing from this list. This is the

4 category of employees that should be mentioned as the second category

5 after inactive population.

6 Q. Thank you. And going to the second-to-last line on page 11, you

7 indicate that in Prijedor there were a total of 665 [sic] men and 65 women

8 who went missing or were killed during the indictment period. Does the

9 figure for men need to be corrected?

10 A. The figure for men should be 1.666 instead of 1.665. Just a

11 difference of one.

12 Q. Again, is it correct that you used the correct figures, in other

13 words, the 1.666 in your calculations and that the error is simply in the

14 text on page 11, as far as the number being wrong?

15 A. It is again the same situation. Error in the text and correct

16 figures in the analysis.

17 Q. I want to go over the sources that you used to make this report.

18 Is it correct that you utilised three different databases?

19 A. Yes. I used three sources. One first source was the list of

20 missing persons completed by International Committee of the Red Cross,

21 ICRC list of missing persons. Second one, is the Prijedor Knjiga Nestalih

22 the Book of Missing from Prijedor. And the third source, is the

23 Exhumations and Proof of Death database. Sometimes I call this database

24 exhumations database --

25 THE REGISTRAR: Sorry to interrupt, we have a problem with the

Page 8398

1 transcript. If we could pause for a second.

2 JUDGE SCHOMBURG: [Microphone not activated]

3 THE INTERPRETER: Microphone, please.

4 JUDGE SCHOMBURG: Will there be an additional or substitution of

5 this report, including the new numbers for the purpose of the registry?

6 MR. KOUMJIAN: Yes, that can be done, Your Honour, easily.

7 JUDGE SCHOMBURG: We would appreciate that at least in that what

8 we have under S337 reads correctly. Unfortunately, we have to wait until

9 the transcript works again.

10 THE REGISTRAR: Keep going and it will be fine.

11 MR. KOUMJIAN: Just to clarify, then, we don't have to go back to

12 the point where it froze and go over those questions again? Thank you.

13 Q. I believe, if my memory is correct, that we were just talking

14 about the sources. You mentioned two, but the first one was a list of

15 missing persons by the ICRC. Can you tell us the date of the version of

16 that list that you have?

17 A. The version we used for this report is the fourth version of the

18 ICRC list. It was published in June 1998, and for the whole Bosnia and

19 Herzegovina, the list included approximately 19.000 names.

20 Q. And you were just telling us about the third list, the Exumations

21 and Proof of Death that you indicated you sometimes referred to in your

22 report as the exhumation database. Would it be correct --

23 A. Yes. This is the third list that we used for this report.

24 Q. Would be it correct that this is a database that you utilised that

25 was put together --

Page 8399

1 JUDGE SCHOMBURG: To be fair to all participants, I just learned

2 that the transcript will stay frozen until five minutes. No doubt then we

3 have -- we will keep up with everything what is reported until now. But

4 we haven't available immediately the text. May we nevertheless proceed?

5 I can see no objections.




9 Q. The Exhumations and Proof of Death database, is it correct that

10 this was a database compiled by Nicolas Sebire of the Office of the

11 Prosecutor from court rulings on proofs of deaths, declarations of death

12 in Bosnia, and lists of persons exhumed and identified in exhumations

13 conducted in Bosnia?

14 A. Yes. This is the database established under the supervision, say,

15 of Nicolas Sebire OTP investigator.

16 Q. Just to clarify, in your report, when you list individuals

17 exhumed, is it correct that because your report, it's necessary to

18 identify those individuals, you are only referring to individuals exhumed

19 and identified or those who have had declarations of death? In other

20 words, you are not utilizing bodies exhumed, but not identified; correct?

21 A. Yes, this is correct. The report is very specific regarding the

22 death, date of death, and name of the person who went missing or was

23 killed. We needed both items: Date of death and the name. All persons

24 whose name was unknown or date of death was unknown or incomplete were

25 excluded from our study. The numbers are quite considerable.

Page 8400

1 Q. In this project for this report, did you again utilise the

2 technique of linking the individuals identified to the 1991 census in the

3 same or similar manner to which you did to prepare your earlier report

4 regarding the changes in the population demographics of Prijedor?

5 A. We used the same technique, matching, individual matching, with

6 the census, all sources, the three sources I mentioned were linked with

7 that census. Moreover, the sources were compared with each other in order

8 to exclude duplicate records, and we also compared our lists of missing

9 persons and killed persons with the voters register, and in this way we

10 could exclude records inconsistent, those reported missing or dead but

11 still included in the lists of voters.

12 Q. So would it be correct, then, that among the individuals excluded

13 from your analysis would be individuals that you could not for some reason

14 link to the 1991 census, perhaps because the date of birth was not

15 available or other information sufficient to positively identify the

16 person?

17 A. Yes. Certain number of records from the lists of missing or dead

18 were not linked with the census. Approximately, say, 20 per cent, for

19 different reasons, one of them being the reasons you mentioned, another

20 one of just spelling mistakes in the names that make it very difficult to

21 link different lists with each other.

22 Q. You indicated that you used only those individuals who it could be

23 determined had died between the 30th of April and the 30th of September,

24 1992. Is that correct, that you excluded all others who you had a

25 different date of death for, a date of death outside of that period?

Page 8401

1 A. Yes. The date had to be within the mentioned interval in order to

2 include a record, given record in the analysis. Unfortunately, there were

3 dates of death that were incomplete, so all deaths, for instance, from

4 1992 where the month of death was missing, to be excluded as well.

5 Q. So when you received an incomplete date of death or no date of

6 death, did you -- that person was then excluded from your database; is

7 that correct?

8 A. Yes. Such persons were excluded from the databases used in this

9 study.

10 Q. Then you indicated that you did a second exercise to remove

11 duplicates from among the three databases; is that correct?

12 A. Yes. We merged the lists with each other and looked -- compared

13 the records on these merged lists, and once we saw that one record was

14 repeated more than one time, so all the duplicated records were excluded.

15 Q. Excluding all those for whom you could not identify and link to

16 the 1991 census, those persons whose dates of death could not be

17 determined or it could not be determined that they fell within the

18 indictment period and removing all duplicates, is it correct that you

19 ended up with a total of 1.731 individuals from Prijedor who were killed

20 or went missing during the indictment period?

21 A. Yes. This is the total we ended with.

22 Q. Looking at page 6 of your report, of that total of all

23 ethnicities, looking at table 2 of page 6, of the 1.731 individuals, is

24 it correct that 1.651, or 95.4 per cent, were Muslims?

25 A. Yes, this is correct.

Page 8402

1 Q. And again I believe you testified concerning the ethnicity

2 categorisation in your first testimony, is all of the ethnicities listed

3 based upon a person's self-reporting in the 1991 census?

4 A. Yes, this is correct. Ethnicities -- an item taken from the 1991

5 population census.

6 Q. And to remind us, is it correct that the category "others" that

7 you have listed includes anyone who identified themselves as being a

8 Yugoslav, for example, or of other ethnicities in Prijedor, such as

9 Ukranian or Roma, they would all be listed under the "others" category or

10 "check"; is that correct?

11 A. Yes, this is correct.

12 Q. Of the 1.731 who died during the indictment period, how many were

13 of Serbian ethnicity?

14 A. There were only four persons of Serbian ethnicity that we found in

15 this list of 1.731.

16 Q. I'd like to now turn to page 10 of your report. Is it correct

17 that you did a similar analysis for other municipalities including the

18 Autonomous Region of Krajina and that the results of these municipalities

19 - I believe there are - I believe 19 for the Autonomous Region of Krajina

20 listed here, these are presented in table 4 on page 10 of your report.

21 A. Yes. We see, in table 4, 19 municipalities from the ARK

22 Autonomous Region of Krajina, Prijedor being one of them.

23 Q. This table indicates that 3.010 individuals appear in your

24 database. Is it correct, first of all - I should caution - is it correct

25 that this database is slightly different in that it covered the period

Page 8403

1 30th of April through 31st of December 1992?

2 A. I think that the period is just the year 1992. This is the whole

3 1992. And the sources are slightly different. Knjiga Nestalih, for

4 instance, is not used here for Prijedor. Only the ICRC list of missing

5 persons and Exumations and Proof of Death database are used.

6 Q. Is it correct that of those in this database who were identified

7 as having died in 1992, 58 per cent were from the municipality of

8 Prijedor?

9 A. Yes, this is correct.

10 Q. And now I want to discuss other exercises, analyses that you did.

11 Did you then take the list of those persons - there's 1.731 individuals

12 that you identified as having died in 1991 - excuse me, during the

13 indictment period 30th of April through 31 December 1992, and attempt to

14 do some analysis of --

15 A. Excuse me. I don't have anything on my screen. Perhaps someone

16 can help me. Oh, I see. Thank you.

17 Q. Did you look at those individuals and, using data from the 1991

18 census, analyse the educational level, the level of education of those

19 individuals, and also the occupational group in which those 1.731

20 individuals belonged?

21 A. Yes. We analysed the -- those missing and killed by educational

22 level and also occupational status.

23 Q. Can you describe how you did that? Can you just briefly explain

24 it? First for education.

25 A. We analysed the census item educational attainment by

Page 8404

1 restructuring slightly the categories reported by the respondents. We

2 analysed variables that had the following categories as responses. First

3 category, no education at all; second, incomplete elementary school;

4 third, elementary school; fourth, middle school, which is secondary

5 school; fifth, higher or university education; and two categories in

6 school, meaning continuing education and missing values were excluded from

7 the analysis. It is also important to note that we only looked at a

8 subgroup of the population, at the group of those at age 18 or more years,

9 in order to minimise the bias related to the fact that age, as such, can

10 imply that a person is in education still. We wanted to use the same 8

11 classification in both analysis for education and also occupational

12 status. Persons at age 18 or more can become occupationally active, so

13 this was a sort of compromise that allowed us to present, to complete a

14 comparable analysis for the same population group.

15 With occupational status, we analysed two census items. The first

16 one was occupational activity. The item called ZAN. This is a variable

17 name reported in our data files. The variable occupational activity

18 allowed us to distinguish between those occupationally active and those

19 inactive. And the second item we included in our analysis was

20 occupational position. This variable called PZAN in the data file allowed

21 us to distinguish between employees, owners, self-employed, and those who

22 helped the self-employed, the household members who worked with them.

23 We constructed a new variable. This new variable included all

24 occupationally inactive as one population group, and then it included

25 employees, owners self-employed, and their help, as the following

Page 8405

1 categories. We called this new variable "occupational status." So

2 starting from those inactive and ending with variable ended with the two

3 upper status category, categories, first one being owners, second one

4 self-employed, with their help.

5 For these two variables we analysed risks of going missing or

6 being killed. The results of this analysis are presented in section 3 of

7 my report, first on page 12 and 13 we see the results of the analysis by

8 education, and on page 14 to 16 we see the results by occupational status.

9 Q. Would it be correct that in your analysis you did not find a link

10 overall for non-Serbs for the correlation between educational levels and

11 the risk of going missing or being killed in 1992?

12 A. Generally we didn't see a systematic pattern of increasing risks

13 of going missing or being killed associated with the increase in

14 educational categories. The educational categories like secondary school,

15 those who completed secondary school or post-secondary education, didn't

16 have unquestionably the highest risks of being killed or going missing.

17 So this is the answer to your question: We could not see an obvious

18 pattern of targeting related to the educational attainment of the Prijedor

19 population.

20 Q. Would it be correct that that pattern, however, does emerge if you

21 took Croats alone as a single group?

22 A. Yes, indeed. The simplest way to get an impression of how such a

23 pattern should look like and how it looks like for Croats is the figure 4A

24 and 4B on page 12. The blue line, I hope that you can see colours.

25 Not -- the line which is --

Page 8406

1 Q. I think we all have coloured copies.

2 A. You do. Okay. So the blue line is the one for Croats, and then

3 you can see how this line increases with the increase in educational

4 categories. So those with no education at all or incomplete elementary

5 education have the lowest, very small risks of going missing, being

6 killed, and those with post-secondary, who completed post-secondary

7 education, have unquestionably highest risks. The line for Muslims, the

8 green line, the uppermost line, is different. It is clear that those

9 with post-secondary education do not have the highest risks of going

10 missing, being killed. There is still certain increase in the risks with

11 the increase in education and education level observed for Muslims as

12 well.

13 Q. And just to remind us: Again, these are the risks of the overall

14 population as it existed in 1991. So, for example, if people with

15 post-secondary education were perhaps smart enough to leave Prijedor prior

16 to being detained, they would show up on your report simply as someone who

17 was -- they would lower the risk of death of that group; correct?

18 A. Yes. This would certainly lower the risk in this group.

19 Q. You were not able to, for example, look at the solely the persons

20 detained in camps to see the link between education or occupation and the

21 risk of being killed; correct?

22 A. Well, we did not this type of statistical information at our

23 disposal, and the purpose of this report was to analyse missing and killed

24 population from Prijedor.

25 Q. I now want to turn to the statistics that you mentioned regarding

Page 8407

1 occupational status. Is it correct, if we look on page 15, at table 6A,

2 And at page 16 at table 6B, the difference between these is that table 6A

3 includes men and women and table 6B is only men; is that correct?

4 A. Yes, this is the difference.

5 Q. Looking at the figures under the last portions of each table,

6 deaths per 100.000 population, would it be correct to interpret these

7 figures as -- if you divide the figure by 1.000, this is the percentage of

8 these persons who were identified in the 1991 census who were also

9 identified as having been killed in Prijedor within the indictment

10 period? Is that correct?

11 A. Yes.

12 Q. So, for example, if we take the figures for Muslims on table 6A,

13 deaths per 100.000, would it be correct, then, that among the inactive

14 occupational group, 2.934 per cent were killed or went missing in the

15 indictment period?

16 A. Yes, this is correct.

17 Q. Among employees, 5.359 per cent were killed or went missing in the

18 indictment period?

19 A. Yes, correct.

20 Q. For owners, approximately 6.7 per cent; correct?

21 A. Yes, correct.

22 Q. For self-employed, approximately 6.64 per cent?

23 A. Yes.

24 Q. And for those with self-employed with help -- and by that -

25 correct me if I'm wrong - "self-employed with help" means that person has

Page 8408

1 employees that they pay or has employees that assist them; is that

2 correct?

3 A. Not exactly. The "help" are other household members working

4 together with the self-employed person and not necessarily being paid.

5 Q. Thank you. Among this group, 9.56 per cent were killed or went

6 missing in the indictment period, if we look at men and women as a group

7 in Prijedor?

8 A. Yes, this is correct. But we generally analysed these two last

9 categories - "self-employed" and "help to self-employed" - together as one

10 category, because these two groups belong together. This is just

11 self-employed and their families who helped them in their work, so they

12 worked just together.

13 Q. If we now turn to the next table, 6B, on page 16, which again is

14 limited to men only, only to males. And if we again look at the Muslim

15 population, would it be correct that - I'll go through them from

16 "inactive" up to "help to self-employed" - the figures increased, the

17 percentage, the risk -- the percentage of those from the 1991 census

18 identified who were also identified as having been killed in the

19 indictment period went from inactive 6.8 per cent, employees 10 per cent,

20 owners 11.6 per cent, self-employed 11.3 per cent, and help with

21 self-employed 26 per cent. Is that correct?

22 A. Yes. Only the first figure should be 6.9 per cent.

23 Q. 6.9. Thank you.

24 A. Approximately.

25 THE REGISTRAR: Please remember to pause. Thank you.

Page 8409


2 Q. I now just want to briefly touch upon your regression analysis.

3 In the regression analysis, you looked at these factors of occupation.

4 Let's look at table 7 on page 17. You did a logistical regression in

5 order to determine the effect statistically of being a member of an

6 ethnicity and a particular occupational group upon the risk of being

7 killed or going missing during the indictment period; correct?

8 A. Yes.

9 Q. And is it also correct that you used as a baseline the group of

10 inactive others?

11 A. Yes. This is the reference category used to compare all other

12 categories to show how many times the risks observed for other categories

13 are higher or lower in relation to the reference.

14 Q. And just to again remind us: The "others" category would include

15 not only other ethnicities, such as Albanian, Ukranian, Czech, but it

16 would also include persons who declared themselves as Yugoslavs, perhaps a

17 Muslim married to a Serb; correct?

18 A. Yes. "Others" contains first of all Yugoslavs. This is the major

19 component of this ethnic group. And the other categories are as you

20 mentioned

21 Q. So if we look at the regression analysis -- let's take it for men

22 and women. If we look at Muslims -- by the way, the significance level,

23 the lower it is, the more significant the result. In other words, the

24 harder it is, the more unlikely that this would have been obtained by

25 chance; correct?

Page 8410

1 A. Yes, this is correct.

2 Q. As far as the exponent on the regression analysis, would it be

3 correct that this would indicate that, for example, Muslims self-employed

4 with help, your analysis showed that these persons were approximately six

5 times more likely to be killed or go missing during the indictment period

6 than the baseline group, the others inactive; correct?

7 A. Yes. This is how these figures should be interpreted exactly.

8 Q. You did not compare Muslims in this analysis to Serbs directly;

9 correct?

10 A. No.

11 Q. However, you did compare Serbs to the baseline group; correct?

12 A. Yes. Yes. Correct.

13 Q. And your figures for Serbs indicate that they were much, much less

14 likely to go missing or be killed in the indictment period than those in

15 the others inactive category?

16 A. Yes, because the estimated parameter is lower than 1. So this

17 simply means a risk that is lower than the risk observed for the reference

18 category.

19 Q. So if in fact you had used, for example, Serbs, employees, as the

20 baseline, the exponent for Muslims would have been much higher than it is

21 when you used the others category; is that correct?

22 A. Yes, this is the implication.

23 Q. Now, did you also attempt to make some estimate from your data of

24 the total number of persons who went missing or were killed in Prijedor in

25 1992 using a statistical tool known as the capture-recapture method?

Page 8411

1 A. Yes, we did. In this way, we wanted to address the issue of the

2 character of our total of 1.731 missing and killed persons. This is an at

3 least number. This is a minimum number, and we must be very clear about

4 that. We -- by estimating the capture-recapture unknown total number

5 wanted to stress that the actual number is much higher than what we were

6 able to produce on the basis of the three sources used in this study.

7 And the estimated total is higher than 1.731. The total estimated is

8 2.190, and we believe it is still an underestimation of the true total,

9 that we are unable to present in terms of lists of persons.

10 MR. KOUMJIAN: Your Honour, I'll ask, Mr. President, for your

11 guidance. Would it be possible for me to have an easel, which is just

12 outside the courtroom, be brought in so I could put in some figures and

13 some demonstrations in front of the Court?

14 JUDGE SCHOMBURG: Please do so.

15 May I use the time, because we have to make a decision today on

16 the addendum I don't want to mention, the one of a sentence. The addendum

17 to Prosecution's false motion for specific protective measures. Can the

18 OTP agree that when it reads that persons be present in court during the

19 testimony, it reads: "In their capacity as amici curiae."

20 MR. KOUMJIAN: Yes. I think though, Your Honour, to be accurate,

21 the person will be representing the interests of his own government as

22 amici curiae.

23 JUDGE SCHOMBURG: Yes. And finally, that because it's not

24 mentioned in the conclusion in B5, also there it goes without saying, as

25 it was said before, the Trial Chamber will then rule on the submissions.

Page 8412


2 JUDGE SCHOMBURG: Observations by the Defence? Is it correct for

3 you?

4 MR. OSTOJIC: Well, we don't consider that individual as an

5 amicus curiae. We consider him in a totally different light. So we do

6 not object to that categorization at all.

7 JUDGE SCHOMBURG: Yes. Indeed we would appreciate and then on

8 this basis it's -- it is agreed as requested. Thank you that we could use

9 the meantime.

10 MR. KOUMJIAN: May I proceed, Mr. President?



13 Q. Ms. Tabeau, I just would like try to explain this

14 capture-recapture method using some very simple figures so we can get an

15 intuitive sense of what that is. Let's say we were trying to measure a

16 small town which has an unknown population. We don't know it, but the

17 real population is 1.000 persons. And we had two samples. Someone had

18 counted everyone wearing blue and they found that 400 individuals were

19 wearing blue. Someone had also counted the number of persons with college

20 degrees and found 100 persons had college degrees. But we don't know when

21 we start just from these two samples is that about 40 per cent of people

22 in this town wear blue and about 10 per cent from college degrees;

23 correct?

24 A. This is what we know, yes.

25 Q. We would expect the overlap, in other words, the two samples, to

Page 8413

1 include each other, that a person has both a college degree and blue, to

2 be 40 persons; correct? So in other words, about 40 per cent of the

3 college people are wearing blue and about 10 per cent of the people with

4 blue have a college education, so the overlap is 40. Again, we don't know

5 those percentages now. All we know is we have these two samples. If we

6 were using only these two samples, in order to estimate the total

7 population of the town, what would the formula be?

8 A. For the capture-recapture method?

9 Q. Yes.

10 A. So the total would be the size of the first sample, which is 40.

11 Q. 400.

12 A. Sorry. 400. The other one is -- the other sample is 100.

13 Q. And would you multiply those together?

14 A. Yes, yes, yes. You would multiply the two and divide it by the

15 overlap, by the size of the overlap, yes.

16 Q. If you did that, you would come out with 1.000 as the

17 population --

18 A. Yes.

19 Q. -- of the town?

20 A. Yes. Yes.

21 Q. Now, the capture-recapture method does have a premise that the

22 samples are independent of each other, as you state in your report;

23 correct?

24 A. Yes.

25 Q. Because if they are dependent, then if one is dependent on the

Page 8414

1 other, then it could result in an underestimate of the total population;

2 correct?

3 A. Yes. So the estimate obtained under the independent assumption

4 uses the formula as the one you mentioned. While it is not necessarily

5 the case when the samples are dependent.

6 Q. I want to explain, give an example, of how an underestimation

7 could occur. If we had first that same sample of 400 with blue and we had

8 a second sample of the number of persons wearing denim in the town, we had

9 100 persons wearing denim, in fact we know the two are not really

10 independent, because most people -- most denim is blue, maybe 80 per cent;

11 is that correct? Let's assume it's 80 per cent. So if 80 per cent were

12 of denim was blue, then we would expect the overlap to be 80; correct?

13 A. Yes.

14 Q. And if we used your same technique, the capture-recapture method,

15 then the formula would be 400 times 100 over 80, or only 500?

16 A. Yes.

17 Q. You would underestimate by one half; correct?

18 A. Yes. Yes.

19 MR. KOUMJIAN: Okay. I'm done with the easel.

20 A. The conclusion is that statistically speaking if samples are

21 dependent than the estimate produced under the independent assumption does

22 indeed underestimate the total number, and this can be shown statistically

23 in a sort of proof as well.

24 Q. The three databases that you used were the ICRC list, the Prijedor

25 Book of Missing, and the Exumations and Proof of Death database, that is,

Page 8415

1 persons exhumed and identified and those for whom a family member had gone

2 to court and obtained a declaration of death; correct?

3 A. Yes.

4 Q. Would you agree with me that all three of these would

5 systematically exclude situations where, for example, all members of the

6 family had been killed or where the only surviving members of the family

7 were perhaps out of Bosnia and not able to report their family members

8 missing, did not come to make an identification at an exhumation or go to

9 court to get a declaration of death?

10 A. Yes, of course these situations are excluded, these reports are

11 excluded from the lists used in this report. But not only these reports

12 are excluded. I said in the beginning of this testimony that there were

13 many reports in the lists we used that had to be excluded because of date

14 of death or missing was incomplete or missing altogether, or the name of

15 person was not given. We couldn't use these records in linking these

16 records were very, very frequent. For instance, in Knjiga Nestalih, for

17 Prijedor, in total this list included 3.146 records, of which 1.350 did

18 not included the year even of death. So we -- the year was completely

19 unknown, which is a huge portion of the list that could not be analysed in

20 our study. With exhumations, the year was unknown for 652 records in the

21 whole database. Out of a total of about 3.000 records in the database,

22 which is again a considerable portion of information that was not used in

23 our study. So this is another reason -- the excluded records is another

24 reason that the figures used for the capture-recapture method are very

25 low, and of course the result is likely also not as it should be if the

Page 8416

1 figures would be more complete.

2 Q. I believe you indicated it should be seen as a minimum, very

3 conservative estimate; correct?

4 A. Yes. This is what I suggested, and I agree, fully agree with

5 that.

6 Q. I'd like to turn just to page 4 of your report, figure 1, for a

7 moment. We've talked about the effect of there being a positive

8 correlation among the databases in getting an estimate that is possibly

9 too low. If we look at the overlap on the databases, is it correct that

10 every person who is exhumed and identified, or that individual was

11 declared dead because family members went to court, appeared either in the

12 Book of Missing or in the ICRC list, or both? Is that correct?

13 A. Yes, this is correct.

14 Q. And wouldn't that indicate that the process of a family member

15 coming to make an identification or going to court to get a declaration of

16 death, that that family member is also likely to report the person missing

17 in one or both of these sources? Correct?

18 A. This is what we see here in this data, in many, many instances,

19 and this is the logical, I think, behaviour of those who lost their loved

20 ones.

21 Q. Now, if we chose -- I understand the data normally gets better the

22 more sources you use; correct? You normally want to have as many

23 different databases as possible, but it's also necessary for the accuracy

24 of the formula for them to be independent; correct?

25 A. Well, we -- first of all, we have to use the best possible

Page 8417

1 information that is available, but also we have to remember that for this

2 method sources should be independent, and the probability of being

3 reported in every one of the sources for every individual is the same. So

4 these are the two assumptions of the method. In this case, we intuitively

5 know that the sources were not independent but likely dependent, but it is

6 only intuition. At this stage, when the lists are completed we cannot say

7 anything about how these samples were drawn, and we cannot measure the

8 dependency of the samples. This is what I can answer to this question.

9 Q. If we took the capture-recapture method and excluded the

10 exhumation and court declaration database, we would then use the same

11 formula we did for the people wearing blue in the small town. We would

12 multiply the total samples of the Book of Missing Persons times the ICRC

13 list and divide by the overlap; correct?

14 A. Yes. This would be the formula, as the one you used already.

15 Q. Yes. So that would be the KNP sample times the ICRC sample,

16 divided by overlap. And the overlap, looking at the figure 1, would be

17 157 plus 263, or 400. Is that correct? 420. Excuse me.

18 A. 428.

19 Q. 157 plus 263, 420. Correct?

20 A. Yes. Sorry. This is 263. Yes.

21 Q. So the total would be -- the formula would be 1.092 times 1.063

22 divided by 420. And I believe that comes out to 2.763. Would that be the

23 formula that you would use if you excluded the exhumation and court

24 declaration database?

25 A. It is -- I didn't make the calculation, but it seems that you did

Page 8418

1 it correctly, and if the result is correct, this is what we would obtain,

2 a much higher number than what we were able to produce.

3 Q. Thank you. I have no further -- excuse me.

4 When you testified last time, His Honour Mr. President asked if

5 you were preparing a report on all of Bosnia, and is it correct you did

6 prepare and submit a report for the Krajisnik/Plavsic case dated the 20th

7 of July?

8 A. Yes. I completed the report and submitted it, I believe also to

9 the registrar of this case, and I hope Your Honours received copies of

10 this report.

11 MR. KOUMJIAN: Would Your Honour like me to mark that report?

12 JUDGE SCHOMBURG: It's now tendered as S338.

13 MR. KOUMJIAN: I have no further questions. Thank you.

14 JUDGE SCHOMBURG: Are there any objections against the admission

15 into evidence of S337 and S338.

16 MR. OSTOJIC: No to the former, yes to the latter. Namely, we do

17 object to Exhibit S338, Your Honour.

18 JUDGE SCHOMBURG: May I hear the reasons?

19 MR. OSTOJIC: It's a report that deals with items that are outside

20 the parameters of the fourth amended indictment against the accused

21 Dr. Milomir Stakic. It's a report that I think is prepared - with all due

22 respect to Dr. Tabeau who is sitting here - which is skewed, that we are

23 not, and only having received the report today, although there is some

24 disagreement as to when it was tendered to cross-examine around this

25 report. The indictment clearly against Dr. Stakic involves Prijedor. We

Page 8419

1 have a report on that. We have questioned her once and we will hopefully

2 do so again this morning. We don't believe that it's relevant to the

3 indictment. There's been no explanation as to why this report should not

4 have been included with her prior report, and how it now has any bearing

5 or relevance to Dr. Stakic.

6 JUDGE SCHOMBURG: Thank you. No doubt -- yes, please.

7 MR. KOUMJIAN: Ms. Karper informs me that our records show the

8 Defence got this report on the 2nd of August. It was written -- it's

9 dated the 28th of July and our records show it was delivered to the

10 Defence. I know Mr. Ostojic wasn't here the 2nd of August, but on the 2nd

11 of August.

12 JUDGE SCHOMBURG: But the reasons given as maybe obstacles to the

13 admission into evidence, the Trial Chamber will decide on S338 only

14 later.

15 As regards S337, it's admitted into evidence in the form we have

16 it now updated, including the corrections to be found in today's

17 transcript, and once again, I kindly ask the witness to give at least one

18 original with these corrections to the registry.

19 THE WITNESS: Of course.

20 JUDGE SCHOMBURG: The trial stays adjourned until 11.40.

21 --- Recess taken at 11.11 a.m.

22 --- On resuming at 11.46 a.m.

23 MR. KOUMJIAN: Mr. President, just the last time Ms. Tabeau

24 testified, Dr. Tabeau testified, there was one document, S227, which was

25 partially illegible, listing the census breakdown by municipality. We

Page 8420

1 found a much better copy of that, and it's not the same page, but the same

2 census results, and we have that to distribute or to substitute for S227.

3 Dr. Tabeau actually provided it to us.

4 JUDGE SCHOMBURG: I think for the purposes of clarification, we

5 should have this as an addendum to S227 and be marked as 227-1.

6 Objections?

7 MR. OSTOJIC: No, Your Honour.

8 JUDGE SCHOMBURG: Admitted into evidence S227-1.

9 May I ask the Defence to start with the cross-examination.

10 MR. OSTOJIC: Thank you, Your Honour.

11 Cross-examined by Mr. Ostojic:

12 Q. Good morning, Dr. Tabeau. Again, my name is John Ostojic, on

13 behalf of Dr. Stakic. I read your report, and thank you for providing it.

14 I think I understand it. I'm a little frustrated by a couple of items,

15 both in your report and during your testimony on direct examination

16 involving the socio-economic status of individuals in the Prijedor area

17 within the time period of April 30th through September 30th of 1992. Mr.

18 Koumjian from the OTP, specifically on page 14, line 20, seemed to

19 indicate, and I think your testimony was that there was a 20 per cent

20 range that you would attribute to persons who were excluded from the

21 numbers in your calculations. Correct? Meaning, in essence, that the

22 number is understated; correct?

23 A. Meaning that the list used for the capture-recapture method and

24 other analyses are incomplete, lower than the total number of records

25 reported by approximately 20 per cent.

Page 8421

1 Q. And that's where the frustrating part comes in. If you look at

2 the capture-recapture method that you utilise in your annex, in fact the

3 numbers statistically as you've computed them indicate that there's an

4 excess of a 25 per cent overstatement of those numbers using the

5 capture-recapture method specifically, if you look at the real raw data,

6 as I call it, of 1.700 approximate individuals and if you calculate that

7 or compare it to the capture-recapture method, you get approximately 2.100

8 of such individuals; correct?

9 A. Yes. This is the new total, approximately, 2.100 and something.

10 Q. So in fact by utilizing that method or, as you call it, a tool,

11 in your methodology, the capture-recapture method there's actually an

12 overstatement of the exact number based upon your very own calculations,

13 correct, of more than 25 per cent?

14 A. Well, I am a little bit confused. So you are saying that this

15 number, the 2.100, or 2.190, is an overestimation you are saying, because

16 of the unlinked record not included in the calculation? This is correct?

17 Q. In part, yes. But what I'm saying is that it's been well

18 established that utilizing the tool such as the capture-recapture method

19 will provide a statistician or a scientist or a demographer such as

20 yourself an overstatement of the true picture in which you're trying to

21 analyse.

22 A. This method is used to estimate the unreported events, and this is

23 why we applied this method. The lists that we had at our disposal were

24 incomplete, all of them. We knew that. And they were incomplete because

25 of the 20 per cent you mentioned, the unlinked records, and also because

Page 8422

1 of excluded records, the records with incomplete or missing date of death

2 or unknown names. Generally, the lists were incomplete because certain

3 records were not reported at all.

4 Q. So based upon a reasonable degree of scientific, statistical, and

5 demographic certainty, can you tell me which number would be considered,

6 scientifically speaking, more accurate for purposes of determining the

7 qualitative criteria necessary in this case, the 1731, the 2.789, or the

8 2.130 that comes out of the capture-recapture method analysis that you

9 provide in your annex?

10 A. I believe that the number 2.789 is a number that includes

11 duplicates, and we should not look at this number at all, because this is

12 our least made of three sources by combining the sources, and nothing

13 more than that. The number 1.739 is a list that doesn't include any

14 duplicates, and this is a list that we can produce by showing names and

15 other details, date of death, many other details about the victims, and

16 this is an absolute minimum we should think of in this case. And the

17 2.000 estimate capture-recapture estimate, the 2.190 is that what we at

18 least should expect as to the total number of victims, missing persons,

19 killed persons, from Prijedor.

20 Q. And if I may, the difference between 1731 and this 2190, do you

21 make any assumptions as to the ethnicity of those missing people?

22 A. Not at all.

23 Q. Would you logically then take the percentages that you have

24 prepared on table 4 -- strike that. Table 2 on page 6 of your report in

25 order to make that determination?

Page 8423

1 A. You could do that.

2 Q. Let me ask you this: When you do, as Mr. Koumjian did on this

3 easel here, he actually showed you the basic, precise way of doing it on

4 the first page. The second page he made an assumption which indicated

5 that there was an underestimation of the total population 500. What would

6 be the formula, if you will, to show that it was an overestimation of the

7 total amount of victims, if we can call them that?

8 A. Well, there are certainly many formulas that could be used for the

9 dependent samples. The formulas would be specified on the basis of the

10 type of dependency between the samples. And as I said, at this stage,

11 when the lists are available, as they are, I am unable to say anything

12 about the dependency of the samples. So we believe that by using the

13 formula for independent samples, we are not fully correct, but still we

14 cannot say anything more about a better estimate.

15 Q. But is it true, Doctor, that in making your report and in reaching

16 some of your conclusions, in the socio-economic status of the two

17 categories that you were asked to review, namely, educational attainment

18 and occupation, that you used as a dependent source once again the 1991

19 census? Correct?

20 A. We used 1991 census as the source for this part of the report,

21 together with the three lists of missing and killed persons.

22 Q. Now, we discussed at another portion of your testimony in

23 connection with the separate aspect of your report the utter unreliability

24 and untrustworthiness of reliance upon censuses by virtue of the fact that

25 they have control mechanisms that were not utilised, among other things,

Page 8424

1 the flaws, if you will, in a census. Did you, Dr. Tabeau, take into

2 account any of those flaws in creating a margin of error upon relying on

3 the census to reach those two conclusions with respect to educational

4 attainment and occupation?

5 A. Well, after the discussion about the reliability of the census, I

6 actually made quite some efforts to learn more about how the census was

7 conducted and whether the way of conducting the census was as they should

8 be. And I must say, after having requested two reports from people who

9 were involved in the census, I believe that the census was conducted very

10 much properly as they should be conducted in every country, and I can

11 present to this Chamber reports of these local authorities, of those who

12 were responsible and who were involved in the census in Bosnia and

13 Herzegovina, from which it is clear that criteria of reliability were

14 satisfied in the census as well.

15 Well, you asked about -- the second part of your question, about

16 margin of error with respect to educational attainment and occupation. I

17 believe do you mean confidential intervals for the rates of risks

18 presented or things like that? What are you -- do you think about with

19 this -- in this case.

20 Q. I had a lot of thoughts on it. We can discuss it either way. In

21 your last testimony, for purposes of our discussions, basically, and with

22 all due respect, agreed that the confidence levels and the margin of

23 errors -- margin of error can be used interchangeably. So for our

24 purposes, if you can share with me what the confidence levels were, what

25 the margin of error for utilising a census as a dependent source in making

Page 8425

1 your determination that there was no targeting based upon the educational

2 attainment category, then that there was supposedly targeting based upon

3 the occupational socio-economic status report that you did.

4 A. We did not include any confidence intervals for the risks of

5 missing or being killed, for any of these two items, not for educational

6 attainment, not for the second variable, occupational status. But by

7 estimating the regression models, we actually proved that the parameters

8 for Muslims were all significant and the pattern of increasing risks with

9 the increase associated with the increase in the categories of the

10 occupational status is also seen there. So this is some way of speaking

11 about reliability of these estimates.

12 Q. We'll get back to the logistic regression model that you used,

13 because I do have actually a couple of questions in connection with that.

14 Respectfully, I dispute the way in which you in essence linked the two

15 categories, but we'll get to that in due time.

16 Is it correct that in your report you state that for purposes of

17 occupational status, that each and every person who filled out the census

18 specifically, and all of them, filled out the category as to what their

19 occupation is?

20 A. Not all of them. There were two items in that census related to

21 occupation, and this is explained on page 11. The first item called

22 "Occupational Activity" was addressed to all participants of that census.

23 Every respondent had to answer this question. And occupational activity,

24 the ZAN variable, only served to distinguish between those active and

25 inactive. And for those inactive, there were several possibilities

Page 8426












12 Blank pages inserted to ensure pagination corresponds between the French and

13 English transcripts. Pages 8426 to 8434.













Page 8435

1 included as responses to this question. It was only a matter of choosing

2 one by a given respondent. And the other variable, occupational position,

3 PZAN variable was only addressed to those occupationally active to a

4 subgroup of the respondents included in that census.

5 Q. But my question is: For the ZAN category, the occupational

6 activity, it's your position that the census information in your

7 possession reflects that each and every household had completed the

8 census, answered that category; correct?

9 A. Well, this is what I know about it.

10 Q. Well, did you do any independent review of the actual census,

11 since you have that data, to determine what percentage of a whole was this

12 category completely filled out or not filled out? Because I find it

13 personally incredible that in a census and that doesn't matter, it may

14 very well have been completed by all, but in your experience have you ever

15 found that the census form on a specific category other than name and

16 address was completed totally by each and every participant in the census?

17 A. Well, it is usually the case that everybody respondent has to

18 respond to all questions included in the questionnaire, and this is the

19 responsibility of the interviewer to take care of this. And going back to

20 the question about the categories reported in that census for this

21 particular variable, for ZAN variable, we made some frequency tables,

22 checking the responses, and as far as I remember, I cannot show you right

23 now this frequency distribution, there was nothing striking in the

24 distribution. There were categories, for instance, for children, that

25 they were children in school, and therefore inactive, or there was a group

Page 8436

1 of household keepers that were reported as inactive, and the reason was

2 they were keeping household, you know. And this type of answers. There

3 were some special groups, like imprisoned population, who was

4 occupationally inactive because imprisoned. There were handicapped

5 persons in institutions, inactive because handicapped. There were people

6 retired, inactive because retired, et cetera. So this was not only

7 distinction between active and inactive in general sense that this

8 question was an opportunity to be specific about what type of inactivity

9 was linked with a given person.

10 Q. And those tables would be readily available if the Court would

11 permit us to have a copy of that, to show the frequency of all the line

12 items that may have been filled out within the 1991 census form that you

13 provided us last time?

14 A. If requested by the Chamber, we would make, of course, such

15 tabulations.

16 Q. If I could direct your attention to page 6, table 2, please.

17 A. Yes.

18 Q. Is it your testimony, based upon this table and the information

19 that you reviewed, that in fact only 4 ethnic Serbs were killed during the

20 period April 30th, 1992 through September 30th, 1992?

21 A. This is this number for Serbs killed or missing were found in the

22 three lists used for this report.

23 Q. But did the OTP at any time share with you some reports that they

24 have in their possession that we'll see from other witnesses, possibly

25 even tomorrow, that in fact that number is far greater than the number 4?

Page 8437

1 Did they share that with you?

2 A. No, not that I'm aware of at this very moment.

3 Q. And in fact, if you look at your statistical calculations in

4 presenting the percentages specifically relating to the Muslims, where you

5 have 95.4 per cent of the total, that number is, in my opinion - and you

6 correct me if I'm wrong - skewed because if the ethnic Serbs that were

7 killed within that period is greater, what would happen to the percentage

8 as a whole of the Muslims in that category? The percentage would go down;

9 correct?

10 A. All percentages would change slightly, depending on the number of

11 Serbs, new Serbs, records, included in the database. It would be lower,

12 of course.

13 Q. And why do you say "slightly"? Do you have an opinion as to --

14 A. Well, I don't know how many Serbs would be included, additionally,

15 into the database, but -- well, this is my first reaction. So I haven't

16 found many Serbs, so I will be very surprised if rapidly I would be

17 presented with a list where hundreds or thousands of Serbs would be

18 included.

19 Q. If I may ask you this: Within your report, do you include any

20 ethnic Muslims who may have died outside the territory of Prijedor but who

21 were citizens of the Prijedor municipality?

22 A. Yes. There were a few such persons. It is related to the fact

23 that actually for this type of analysis, two different perspectives can be

24 taken. One is the perspective the municipality of origin, or region of

25 origin, or region meaning the 1991 census. Or the other perspective is

Page 8438

1 the municipality of arrival, so receiving municipalities. We made this

2 report using the perspective of municipality of origin because we wanted

3 to present relative measures of risk, and in this case we should think of

4 relating the number of victims to the 1991 census population. And

5 alternatively, we could think of all victims who went missing or died in

6 Prijedor without taking into account the municipality, their municipality

7 of origin in 1991. And these two numbers are slightly different, of

8 course, because these are two different perspectives.

9 Q. Would you agree with me that if you take your number of 1.731, and

10 that if indeed it includes ethnic Muslims who may have died outside the

11 territory of the Prijedor municipality, although within that time frame,

12 that statistically speaking, your estimate of 1731 would be an

13 overstatement of the number of victims missing or dead? Correct?

14 A. It would be, but I want to stress that for deaths, we used the

15 Exhumations and Proof of Death databases. Those were all based on

16 information collected for a particular reason, Autonomous Region of

17 Krajina, Prijedor being one of the most important municipalities in this

18 region. And in Prijedor, the highest number of mass graves were found,

19 and the bodies were exhumed from these mass graves. So there is no --

20 very little doubt about the place of death from this database. Regarding

21 the list of missing persons, we have only little information about place

22 of death, because the item reported in these databases, the location item,

23 is the place where the person was last seen. But again, it was for the

24 vast majority of cases that it was -- they were seen in Prijedor, in

25 Prijedor, and in at most the surrounding municipalities of Prijedor.

Page 8439

1 Q. I'm not asking for the far majority of it, but I think if you take

2 a calculation such as you did in utilizing a number of 1.731, and that if

3 indeed you're trying to give us an objective analysis to tell us that of

4 these people that are certain computations, methodology that we would use

5 to determine that there was targeting it would be an honest assessment on

6 your part to state, whether in a footnote or directly in the text of the

7 report, to tell us that in fact this report does include a certain number

8 of individuals whose origin are from Prijedor but who died outside of the

9 municipality of Prijedor. Wouldn't you agree with me?

10 A. I believe I made such a note in the report. If you allow me to

11 find it. I am very consistent through the whole report by saying that we

12 analysed those from Prijedor municipality, those victims from Prijedor

13 municipality. As far as I remember, I made such a note, where I

14 explicitly explained this issue.

15 Q. Would you be kind enough to direct my attention to it?

16 A. Yes.

17 Q. Thank you.

18 A. Yes. This is on page 5, in section 2.2, second paragraph in this

19 section, where we read: "In this section, we take the perspective of the

20 1991 place of residence of the victims and analyse all those victims who

21 in 1991 lived in the municipality in Prijedor -- of Prijedor by relating

22 the number of victims originating from Prijedor to the 1991 population of

23 Prijedor, we can produce measures of the risk of being killed or going

24 missing for the 1991 Prijedor population."

25 And the footnote 6 explains:

Page 8440

1 "Note that the place of death is available in our sources as the

2 place where victims were last seen as place of disappearance, or as

3 location of mass graves, or place of death or disappearance, indicated to

4 court by the relatives of victims. A vast majority of places were

5 consistently reported to be in Prijedor or surroundings of Prijedor."

6 Q. It's precisely my point. What is the vast majority? You're

7 giving us a number of 1.731 as a concrete number. In your report, knowing

8 just by virtue of these last several questions that there are instances

9 where the number would necessarily be decreased if it's established,

10 either by way of the exhumation report that Mr. Sebire put together by the

11 OTP or any of the other sources that you utilised that in fact people

12 whose origin in 1991 was in the Prijedor municipality may have died or did

13 in fact die outside of the Prijedor municipality, what is the percentage?

14 I understand that it's the vast majority, but as a statistician, by

15 providing me with a margin of error on this particular number, 1.731,

16 perhaps we can gain an appreciation for the relevance in a qualitative

17 term in your analysis.

18 A. Yes. I believe I made certain calculations to assess this

19 percentage. It was an important question, while making this report. And

20 I believe it was a percentage which was far higher than 90 per cent, far

21 higher than that. And I want to stress that it is extremely difficult to

22 be very specific in making these type of calculations, as I present in

23 this report, because of the problems of the data deficiencies. If a

24 person is found in a mass grave in the municipality neighbouring to

25 Prijedor, does it mean that the person died in Prijedor or in this

Page 8441

1 municipality? So it is a very, very difficult to be a hundred per cent

2 specific about the place of death. But it is likely -- what we know for

3 sure, it is that this person used to live in Prijedor and died. So the

4 risks we calculated are for the Prijedor -- 1991 Prijedor population, and

5 from this point of view, these are very correct measures.

6 Q. So the confidence interval, if you will, is somewhere in the 90

7 percentile range, is that correct?

8 A. It is -- confidence interval is a statistical term, if you would

9 ask me to present a confidence interval, I would have to calculate it and

10 then I would tell you it is a 90 per cent confidence interval, 95 per cent

11 confidence interval. 90 per cent I mentioned is a measure, a fraction of

12 those whose place of death or disappearance or location of mass grave was

13 within Prijedor municipality.

14 Q. When you say "problems with the sources," as you just did in your

15 testimony, you're really referencing flaws and inaccuracies upon relying

16 upon sources such as the 1991 census, the ICRC report, Knjiga Nestalih, as

17 well as the exhumation database; correct?

18 A. Well, first of all, I don't like that you put census together with

19 the sources which include information about missing persons and death

20 persons. Census is a different thing, and it is a very reliable, complete

21 source of information. With the sources like exhumations, we have this

22 difficulty that it is -- the place of death is reported just as the

23 location of the mass grave, still located, most of them, within Prijedor

24 municipality. And there are deficiencies in the sense that many items

25 are incomplete or missing in this list of missing persons or dead persons,

Page 8442

1 but the difficulty I am speaking about right now is that it is not a

2 matter of how the figures or the items were reported. It is sometimes

3 hardly possible to report the place of death as such.

4 Q. In addition to the situation that I mention of, namely, that

5 individuals whose origin may have been in the Prijedor municipality and

6 they may have died in a municipality outside of Prijedor, you mention two

7 in your report on page 8 as the morbidity control mechanism as well as

8 the - as you coin it - the accidental hump mechanism; correct?

9 A. Yes. May I say something about accidental hump, observed in each

10 pattern of mortality for populations living under peace, normal times, not

11 in conflict situations.

12 Q. The accidental hump actually means, as a layperson such as myself,

13 that there's a certain category within the population that sustains a

14 death as an end result by virtue of an accident; correct?

15 A. Accidental hump means that this is an increase in mortality rate

16 that is caused by certain factors related to external causes of death.

17 Q. Not natural?

18 A. Speaking medically. Not natural causes of death, if we can

19 make this distinction, yes.

20 Q. The other section of the control mechanism that you use is the

21 mortality table, because every population certainly has that, given any

22 period of time examining any population, there will be some deaths from,

23 as we've just described, natural causes; correct?

24 A. Yes. Natural causes are causes of death related to the ageing

25 process of individuals, and these natural causes of death can be seen in

Page 8443

1 increasing death rates, increasing with age, because of age-related

2 diseases, for instance.

3 Q. But looking at those two additional factors, what would your

4 confidence level be of this section that you've identified as 1.731 people

5 missing or dead in Prijedor within that time period?

6 A. I can not answer this question because it is impossible to answer

7 this question.

8 Q. And why is that?

9 A. Why? Because these are two completely different things,

10 confidence interval for the total number of 1.731, and the age pattern of

11 mortality observed for populations living under peace.

12 Q. So let me just understand it with respect to those three control

13 mechanisms, as I call them, to determine the accuracy and the reliability

14 of the data or the figure of 1.731 that you used. If a group of Muslims,

15 ethnic Muslims, respectfully, were to have gone to a different

16 municipality, let's say even to a different republic or country, Croatia,

17 to fight, their origin is in Prijedor, they died during battle in

18 Croatia. Your report, in essence, the 1.731, includes those individuals

19 that may have perished outside the Prijedor municipality; correct?

20 A. Well, I think you -- your suggestion is very extreme. I couldn't

21 have information about those who died in Croatia or other countries

22 because the sources I use for this report are related to the population

23 who went missing or died, was killed, in the area, Prijedor area and ARK

24 area. So these sources, especially Exhumations and Proof of Death

25 databases, were established to estimate the scale of killing process in

Page 8444

1 this particular area. Missing persons, on the other hand, Knjiga

2 Nestalih, was completed in Prijedor, was initiated in Prijedor and

3 coordinated from Prijedor, so it was basically established for the

4 Prijedor population. ICRC list of missing persons has the most -- the

5 broadest coverage of the three sources. It is the part of the list we

6 used for this report. It's just records for Prijedor. So people who

7 reported last seen in Prijedor and close surroundings of Prijedor were

8 just taken. Only those who lived in Prijedor in 1991. So it is not as

9 you are saying, that any death or missing person is included in our

10 sources. The selection of sources is the basis for producing a

11 dishonourable figure. This is how we should see it.

12 Q. I have two questions in connection with your last answer, but

13 before I get to that, my question is: So your analysis and your figures

14 are as reliable and accurate as the sources upon which you relied on;

15 right?

16 A. Well, in terms of coverage and the time period, geographical

17 coverage and time, calendar time coverage. These are very much relevant

18 sources from the point of view of this indictment.

19 Q. Well, you mention, and I know -- I didn't take it that personally,

20 but you mentioned that my analysis or my suggestion was extreme by virtue

21 of the fact that if individuals who were from the origin of Prijedor go to

22 battle and fight in Croatia and die, how could it be possible that they

23 would be included within any of the data sources upon which you relied.

24 Dr. Tabeau, it would be possible because, in my opinion, if you look at

25 the exhumation data or proof of death, the people who would be reporting,

Page 8445

1 these individuals who had become deceased, are people from what origin?

2 From the origin we said went to Croatia or Prijedor; correct?

3 A. The origin was, of course, Prijedor.

4 Q. So it's not extreme to suggest that if a family of, let's say, 20

5 sent their young men to battle in Croatia and they all perished, the

6 family members wouldn't go for Croatia to report the death, would they?

7 They would report it in the municipality of Prijedor, wouldn't they?

8 A. They would report in the municipality of Prijedor, indeed,

9 especially if the report would be made to the organs like vital statistics

10 administration in this particular municipality that we are speaking about

11 court declarations and exhumations. I cannot imagine that a person who

12 died in a battle in Croatia would be found in mass graves in Prijedor or

13 reported by a family to court, in Sanski Most or Kljuc. I cannot fully

14 agree with this.

15 Q. And you don't have to, and I'm not going to suggest to you.

16 Mr. Sebire's report clearly indicates that of the exhumations the 261

17 conducted of both the mass and individual graves, that in fact there were

18 600 or so that were individually identified, which is approximately 1.100

19 less than the data that you've provided. So in essence, we don't know the

20 answer to that question, and I don't think anyone can get the answer to

21 that question. But I'm suggesting to you is that the assumption that

22 you're making relying upon that data, it's incorrect and because it's

23 unreliable and inaccurate, that there should be a margin of error within

24 your report to suggest that this number of 1.731 is indeed skewed by the

25 percentage of your confidence interval. That's all I'm suggesting.

Page 8446

1 A. Right. By presenting the capture-recapture estimate, we first of

2 all say that there are -- there were unreported deaths and missing persons

3 that should be included in a report like this one. And secondly, we also

4 estimated confidence intervals. So the confidence intervals are included

5 in this report for this estimated total. And then you can see that we

6 have two numbers, 2.115 to 2.265. This is the confidence interval for the

7 total 2.190.

8 Q. And again, I'm suggesting to you that utilizing tools such as the

9 capture-recapture method is inaccurate because it gives an overestimation

10 of the numbers, and in fact if you even compare the methodology used by

11 the capture-recapture method, you yourself will find that your number of

12 1.731 would be a closer number to the true figures than by utilizing this

13 tool known as capture-recapture by 25 per cent, if not greater.

14 A. Well, first of all, I cannot agree with you that the

15 capture-recapture estimate is an overestimation. I don't think it is an

16 overestimation. It is an underestimation of the unknown true total. We

17 don't know the true total. I don't know, and no one knows. I believe

18 that the input used for the capture and recapture is very close to the

19 number that we should think of -- a minimum number for Prijedor that we

20 should think of, a minimum number of victims.

21 Q. You're familiar, then, in fact, the capture-recapture method is

22 considered outdated since its initial theoretical coming out, if you will,

23 in the statistical area, coming out in 1940, and you'll also agree with

24 me, will you not, that that method, capture-recapture, was utilised only

25 for purposes of counting and calculating what the wildlife animals would

Page 8447

1 be in certain areas? True or false?

2 A. I disagree that it is outdated. I think that we can find hundreds

3 of variants of this method, including -- in our models, very modern

4 technique that did not exist by 1940. So we could have used a more

5 complex approach which would be certainly less clear to you here in this

6 courtroom, perhaps also to us less clear. At least now by presenting a

7 simple method, we all understand a little bit what are we talking about.

8 And as I said, the intention is to speak about through numbers that are

9 different from minimum numbers that we can show black on white, you know,

10 like the number 1.731.

11 Q. And I'm trying to understand it. Let me direct your attention to

12 table 6A if I may on page 15. Although our pages are not numbered. Page

13 15, I think, table 6A.

14 A. Yes.

15 Q. Under this occupational status category that you have, namely, the

16 two that I'm going to focus on for our purposes here, the self-employed

17 and the help to self-employed, from what I understand from reading your

18 report if I may, self-employed would actually be, for example, in a

19 household, the father of the household, the husband of the household;

20 correct?

21 A. It depends on what the person would be doing. It is mainly, I

22 believe, farmers who are included as self-employed. And the aid, the help

23 to self-employed, would be the families of the farmers.

24 Q. So just if you could follow with this table for me, under

25 self-employed, you have that there were 110 Muslims. Just under that

Page 8448

1 category, out of 1.657; correct?

2 A. Where this total comes from?

3 Q. The next category in your table, 1991 population.

4 A. Yes.

5 Q. Then under the category "Muslims." Do you see that?

6 A. Well, I see that there were 110 Muslims, self-employed Muslims,

7 who died or went missing, out of the total of 1.609 of all Muslims who

8 died or were killed.

9 Q. Right.

10 A. So this is how it should be read.

11 Q. Thank you. I'm sorry. And the next category, you have 13 Muslims

12 with help to self --

13 A. Yes.

14 Q. Self-employed, and out of those, you also include 136. So out of

15 those 136 that in the 1991 census identified themselves as help with

16 self-employed, you've calculated that 13 of them have died; correct?

17 A. Yes.

18 Q. Now, if you compare that -- and this table is for both men and

19 women; correct?

20 A. Yes.

21 Q. And are we making an assumption here, since we separated the two

22 categories, that the 13 individuals under the category of help to

23 self-employed were women or men?

24 A. This is both men and women, the certain individuals,

25 Q. Now, if you look on your next table, the 13 individuals --

Page 8449

1 A. Yes. It is only men.

2 Q. By comparing the next table on table 6B, the same 13 number under

3 the category of help to self-employed appears; correct?

4 A. Yes.

5 Q. So in fact we can conclude by looking at those two reports that it

6 is only men; correct?

7 A. Yes.

8 Q. So when I asked you whether this category of help to self-employed

9 included men or women, you said both, but that only of those that

10 identified themselves as men have actually perished. Would that be a fair

11 read on it?

12 A. It is also about the population as risk. So if you compare the

13 1991 population in these two tables, then you see that for Muslims in the

14 first table, the total is 136, and in the other table it is 50. So the

15 population at risk is used to calculate the rate, and this is why the

16 rates are different. But in fact, it is men, the rate for men, which

17 should be considered. But I think that it is good to realise that we did

18 not analyse the help to self-employed as a separate category. This is a

19 category that was analysed together with self-employed.

20 Q. I realise that, and we'll get to that in a second when we do the

21 logistical regression model that you use in your annex. Just so I could

22 have clarification for my purposes. When you -- in each of your tables or

23 in most of them, you have a category that says death per 100.000; correct?

24 A. Yes.

25 Q. Like you do on table 6A. When you have that, although I think

Page 8450

1 it's fundamental, I apologise for asking you that, in essence it's giving

2 us a percentage per 100,000; correct?

3 A. No. It is a number per 100.000. If you divide these numbers in

4 the table, the rates, deaths, per 1.000, by 1.000, then you end with

5 percentages.

6 Q. Help me understand it, then. Rook on table 6A under "owners."

7 Let's take a different category. You have under those individuals of

8 ethnic background being Muslim who died under the category of owners, 32

9 died out of 479; correct?

10 A. Yes.

11 Q. And if you continue to read your table on the next section, deaths

12 per 100.000 population, you have 6681; correct?

13 A. Yes.

14 Q. In fact, by utilising the term 100,000, all you have to do is

15 move the decimal point three intervals to your left and you get the exact

16 percentage. So if I were to ask you what is the percentage under the

17 category of owners that died vis-a-vis the population in 1981 all you

18 would have to do is to look on this section and say 6.6; correct?

19 A. Yes, something like that.

20 Q. Well, it's that exactly, isn't it?

21 A. Yes, it is.

22 Q. Now, my problem with understanding --

23 THE INTERPRETER: Please slow down.

24 JUDGE SCHOMBURG: May I ask the parties first to slow down and not

25 to overlap. Thank you.

Page 8451

1 MR. OSTOJIC: Thank you, Your Honour.

2 Q. In examining your logistic regression model and reaching the

3 conclusion that there seems to be a targeting of occupational status in

4 the Prijedor municipality within the time frame, am I correct in

5 understanding that what you did was use as comparing the two the inactive

6 status of one group and that of two other groups correct, namely, the

7 group of self-employed and help to self-employed? Correct?

8 A. Well, partly correct, partly incorrect.

9 Q. Tell me the part that I was correct on first.

10 A. Correct is that two other groups, ethnic groups, were compared

11 with inactive others, but we analysed the data by two variables at the

12 same time, by ethnicity and occupational status. So the comparison we

13 made is comparing ethnic groups, like Muslims and Serbs, by occupational

14 position with professionally inactive others. So it is -- each ethnic

15 group is analysed at the same time by occupational status.

16 Q. Yes, but when you gave the testimony, I think during your direct

17 examination and your report, that there was a four times more likelihood

18 that someone who has an occupational status such as self-employed or

19 employed with help and those with an inactive status, really what you're

20 saying is that you did not include in that variable the other two

21 categories that you identified in your report of occupational status;

22 correct.

23 A. In the regression model, the rates or risks of, say, Muslims and

24 Serbs in certain occupational status, categories, are compared with the

25 risk for inactive others, inactive others, inactive others. So this is

Page 8452

1 the reference. So for Muslims, for instance, if we refer to table 7 on

2 page 17, and we see Muslims employees, then next category is Muslims

3 owners, and next one, self-employed and help, and we then look for men and

4 women at the XB values, we see the numbers, approximately, 4.7, 5.9,

5 6.1. These are the numbers. So these numbers tell us how many times the

6 risk of, for instance, Muslims employees were higher than the risk for

7 inactive others. It is just a comparison of these two groups, one being a

8 reference, the same reference for all other categories.

9 Q. But you did not, if I may, you did not make that same calculation

10 with respect to each of the occupational status, namely, employees,

11 owners, or self-employed and help; correct?

12 A. Separately, you mean, as such.

13 Q. Yes. Using the --

14 A. This model that we are discussing is a so-called interaction

15 model. This means that we look at the effects of being of a given

16 ethnicity and, at the same time, of belonging to a particular occupational

17 group. So it's answer to your question, whether I looked separately at

18 the various occupational categories: No, I didn't, not in this model.

19 Q. Does your report or conclusions in any way give us an indication

20 of what the intellectuals, if you will, and looking only at occupational

21 status - we'll get to educational attainment in a second - the

22 intellectuals or leadership of the community, whether or not they were

23 targeted?

24 A. Well, it is -- generally, if you make an analysis by

25 socio-economic status, in whatever way the status is measured, you

Page 8453

1 basically want to identify patterns in intensity of a process, like

2 missing or being killed, for the particular categories of the measure used

3 in the study. And, well, how to measure whether or one is a member of

4 the elite or not is of course a very difficult question. One can think

5 of intellectual elite, and then the education is probably not such a bad

6 measure. You can also think that elite are those who cumulated certain

7 material goods more relatively to others. And then you can think of

8 distinguishing owners from employees or inactive, whose income is

9 certainly lowest among all occupational groups.

10 Q. All I'm suggesting is that you did not do any such calculation or

11 analysis; correct?

12 A. You mean the risks of being killed or going missing for the elite

13 from Prijedor? This is your question?

14 Q. Correct.

15 A. I believe -- I didn't certainly attempt to measure the risk for

16 the elite because the elite can be defined in many, many various ways, and

17 I didn't make any study of whether my definition is a proper one, whether

18 it really measures being a member of the elite in Prijedor. I don't think

19 so.

20 Q. In part, I agree with you. But in your answer earlier you did say

21 that you looked as factor education, and that can be considered. Your

22 report does address at least arguably education as a possible factor or

23 educational attainment to determine whether or not intellectuals or the

24 elite were targeted, and your conclusion is, based upon a reasonable

25 degree of scientific, statistical, and demographic certainty, that you

Page 8454

1 could not find any such pattern; correct?

2 A. Not for education.

3 Q. My question is for occupation. The same question: You did not in

4 your report, based upon the four or five categories under occupational

5 socio-economic status were you able to determine a pattern based upon

6 whether or not intellectuals were targeted; correct?

7 A. Based on occupation, I believe I did find a pattern, systematic

8 pattern of a sort of selection.

9 Q. What do you mean by sort of selection?

10 A. Well, if you observe, if you see, if you obtain risks, increasing

11 risks associated with the increasing categories in the occupational

12 status, then it is quite obvious that this is the pattern, and we -- I

13 actually saw this pattern, not only for Muslims but also for other ethnic

14 non-Serb groups.

15 Q. But from my understanding, you did not use owners in your

16 logistical regression table and compare it to the other categories. What

17 you did is actually use, which is reasonable in my opinion, because it's a

18 rural area, you used two separate categories, namely, self-employed, which

19 I think we can agree means farmers; correct?

20 A. Yes, self-employed with help, farmers and their families.

21 Q. Right. And that was my next question. "With help" really means

22 the farmer with the people within that same household assisting him and

23 maintaining and cultivating the land; correct?

24 A. Yes.

25 Q. So it's that group that you've identified that was targeted;

Page 8455

1 correct?

2 A. Not only that group, but two groups, owners and self-employed with

3 help. I think I'm speaking about these two group. These two groups are

4 those who have the upper occupational status, and I believe possessed

5 certain material, certain property, simply, property. This is what I'm

6 saying.

7 Q. And I want to just understand it. So an employee, for example,

8 like a president of a municipal assembly, if he's within the category of

9 employee, you would not be able to quantify for us whether or not that

10 person was targeted; correct, because, utilizing your logistical

11 regression model, you totally excluded that category of socio-economic

12 status under occupations, that category; correct?

13 A. The part about logistical regression is incorrect. How the

14 person, the president, would be classified, whether an employee, he would

15 be an employee or an owner, it depends. If he owned, being the president,

16 also a private firm or a shop or a farm, he would be probably reported not

17 as an employee but as an owner or a self-employed -- self-employed, I

18 don't think so. As an owner, I believe.

19 Q. Or he might be identified as both, and then we would have the

20 problem that we discussed before, and that is of double-counting; correct?

21 A. No, no, no. These type of situations, double answers given to the

22 same question, I don't think so.

23 Q. So what assumptions did you make, if you could quantify them for

24 me, that within the categories that you looked under self-employed and

25 help to self-employed were these individuals also under the category of

Page 8456

1 employees?

2 A. Well, it is -- I don't think so. These persons who reported

3 themselves as self-employed or helping self-employed persons would be

4 reported as self-employed or help. However, it is possible that the

5 person helping a self-employed person reported himself or herself as

6 inactive on the different item, the ZAN variable. We actually believe

7 that this is very well possible because these two items show empirically

8 certain overlap. And in cases, overlapping cases, we gave priority to the

9 report obtained for the PZAN variable, this means that those who were

10 report themselves as inactive and, at the same time, as helping the

11 self-employed, would be taken as help to self-employed in this study.

12 Q. And then reaching your conclusion that there seems to be a pattern

13 of targeting of that group, can you tell us -- I mean, we know that you

14 relied on these secondary sources, if you will, and you also relied on the

15 census exclusively to provide you with that information; correct?

16 A. This information for this part of the study, socio-economic study,

17 comes from the census and is based on the links of those missing, killed,

18 with that census.

19 Q. On page 4 of your report, you identify a term that I'm unfamiliar

20 with. I think it's in the last paragraph, "Quasi-independent sources."

21 Can you tell me what that means?

22 A. Quasi-independent sources. These are the sources that are not

23 fully independent. So there is some degree of dependency of the sources

24 that cannot be expressed quantitatively in terms of correlation measure or

25 something, but we expect that the sources are dependent, and accordingly,

Page 8457

1 we can modify the formulas used for capture-recapture method.

2 Q. And you would include in those sources what as being

3 quasi-independent?

4 A. Well, it is that a person reported missing, for example, in the

5 ICRC list. We think it's likely to be found in mass graves or a

6 declaration of death for such a person that is more likely to be obtained.

7 Q. Did you factor in Mr. Sebire's report the number of mass graves

8 versus the number of individual graves to make an assumption as to whether

9 or not those individuals may have died under any of the three items that

10 we discussed, namely, the accidental hump or the mortality within a

11 community or municipality of natural or unrelated causes?

12 A. So you are speaking about causes?

13 Q. Correct.

14 A. Causes of death?

15 Q. Correct.

16 A. I must say, I did not study causes of death. To a certain extent,

17 this information is available from the exhumation database, from an item

18 called circumstances of death. The reports, however, are very

19 unstandardised, so it is -- the circumstances tell us something about the

20 situation a person died. Sometimes it can be used to conclude that this

21 was a violent death. In most cases, it was violent deaths and the person

22 was killed, and someone, an eyewitness, saw it. But we didn't study

23 causes of death, due to the lack of standardisation, say, of the reports

24 we had at our disposal.

25 Q. If the exhumation reports have a specific line item which would

Page 8458

1 permit the person filling out the proof of death, such as circumstances

2 of death, and if you're giving testimony before this Tribunal as to the

3 number of deaths and to determine targeting, isn't cause of death a

4 significant issue for a demographer, such as yourself, to determine?

5 A. Of course. Cause of death is a very important item. And if I

6 would be able, I would certainly do so. But first of all, regarding

7 missing persons, it is highly likely that all these people reported as

8 missing are dead and died of violent causes of death. This is what ICRC

9 themselves say about the nature of the disappearances reported in their

10 lists. This is one thing. Well, the deaths coming from court

11 declarations, it is all violent deaths or circumstances unclear. But for

12 those with unclear circumstances, it is missing as the cause of death,

13 which is the same situation this person is missing and the body was never

14 found. But there were people who saw the person taken by some persons

15 against the will of the person, and since then no one has seen this

16 person. So it is -- even though I did not study the causes explicitly, I

17 assumed that the causes of death or disappearance were violent, were not

18 natural. And I am convinced that this is the case.

19 Q. Within the OTP demographic section, if there is such a section --

20 A. There is.

21 Q. Is that information readily available on a CD-ROM or disk, both

22 the exhumations as well as the ICRC report? Is that available --

23 A. Electronic format, you mean.

24 Q. Fair enough.

25 A. Of course. Yes.

Page 8459

1 Q. So if I were to ask the Court or someone to provide that to

2 another demographer if they were interested to look at it to determine the

3 cause of death, that would be something that you can provide or your

4 department can provide to me in a CD form correct? Or an electronic

5 version?

6 A. Yes. ICRC list is publicly available from the Internet, so you do

7 not have to request this list from anyone. The Knjiga Nestalih, I think,

8 is available electronically, and those who published it can share this

9 information with you. As to Exhumations and Proof of Death databases, I

10 think the question must be addressed to those who established the

11 databases.

12 Q. I will. I'm just asking if it's available in an electronic form

13 that's all, and it is right?

14 A. Yes, it is.

15 Q. And also --

16 A. But if I may add something. It is a complex database, so it is

17 not just one single list of names and date of death. That's it.

18 Q. Really what I'm actually -- let me just narrow it down if I may

19 with the Court's permission. The proof of death or the certificate of

20 death as it is called in numerous places is a page or two pages in length.

21 What happens in essence is that you scan that material and then it creates

22 a database of all the exhumations and proof of death from the Bosnian

23 Federation courts; correct?

24 A. Well, it is not necessarily scanned. It is -- the data is entered

25 by some entry personnel and it is just making electronic lists based on

Page 8460

1 the documentation that exists as hard copies, as reports.

2 Q. And also the 1991 census is available in electronic form; correct?

3 A. Yes, it is.

4 Q. And now let me ask you, based on your prior -- just now your

5 answer, if I recollect it correctly: You made an assumption that the

6 exhumations and the report of proof of death, that you're assuming that

7 those were of violent deaths; correct?

8 A. Yes.

9 Q. And in fact - there's no dispute about that - if you look, your

10 number on the exhumations for the data as a source was 960. That's what

11 we have exactly. However, you use 1.731 as the number of missing or dead

12 persons; correct?

13 A. One correction. The 960 was incorrect number. I corrected it

14 earlier today, and it is 638, 638. And the number 1.731 is the number

15 obtained not only on the basis of Exhumations and Proof of Death

16 databases, but also lists of missing persons.

17 Q. What I'm suggesting to you is this: If I were to ask you or

18 another demographer to take a specific item such as the exhumations and

19 do an analysis to determine whether there was targeting, the best possible

20 way to do that is to take the number of 638, which are the established

21 proof of deaths by virtue of violent means, and then make your calculation

22 and tables based upon that, instead of using what I consider,

23 respectfully, Dr. Tabeau, a grossly overinflated number of 1.731, which is

24 almost three times the number of those reported in the Federation courts.

25 Would that be a way that someone can do it?

Page 8461

1 A. No, I wouldn't do this in this way, because you restrict your

2 analysis to information obtained from exhumations and court rulings, which

3 we know is very much incomplete, and you should include as much

4 information as possible. And this is what we did. We analysed the other

5 total, because this is the way to do it.

6 Q. When you say "incomplete," you really mean it's unreliable and

7 inaccurate; correct?

8 A. No. Incomplete, I mean -- means that many records are excluded,

9 are not there, not in this list.

10 MR. OSTOJIC: That's all the questions we have, Your Honour.

11 Thank you very much.

12 Thank you, Dr. Tabeau.

13 JUDGE SCHOMBURG: Thank you. Just a final clarification. We had

14 already an extended discussion on this number 1.731, and I would ask when,

15 for future purposes, maybe in the case of Defence, you shouldn't take it

16 as given that the legal relevance of these numbers has such a high

17 importance, as it was stated in former decisions. Maybe this Chamber, as

18 already indicated, may come to other solutions and to other definitions.

19 And we have to be aware what we are discussing here. We are discussing

20 the alleged genocide, and I wonder if it's the real and concrete and most

21 appropriate approach to discuss this on the basis of figures only. The

22 term we have to discuss is whether or not there was an intent, and does it

23 really change whether the number is 1.700 or 2.300?

24 Questioned by the Court:

25 JUDGE SCHOMBURG: Having said this, only one question: It was

Page 8462

1 discussed whether or not there were other peoples included in this number

2 of 1.731. You stated this would be the absolute minimum; correct?

3 A. Yes.

4 JUDGE SCHOMBURG: It was discussed, from the point of view of the

5 Defence, no doubt correctly, what about the question: Is it an

6 overstatement? But can you please tell me: In how far are other persons

7 included in this number, and how does it affect this number? For example,

8 we heard on several occasions earlier from witnesses that we had visitors

9 coming to Prijedor, and no doubt not included in the former census, and

10 then allegedly later become witnesses in this tragedy in Prijedor. What

11 would be the impact on your numbers of persons, and did you take these

12 additional persons into account? Because it would maybe affect the

13 question of the percentage.

14 A. We only looked at the population who originally lived in Prijedor

15 in 1991 and was reported in that census, and only events related to this

16 population were analysed in this report. So if a visitor went missing,

17 for instance, or was killed, it is not included in this study.

18 JUDGE SCHOMBURG: Thank you for this. And one second -- just for

19 my own clarification, help me with this: Given the example a person was

20 killed in the municipality of Sanski Most and, for what reason whatsoever,

21 found his or her grave in Prijedor, was this included in the number of

22 persons you identified in your report?

23 A. If this person killed in Sanski Most would be reported as part of

24 Prijedor population in 1991 census --

25 JUDGE SCHOMBURG: Say this person -- yes. I don't know whether

Page 8463

1 there was such a census in 1991 also in, say, Sanski Most. And this

2 person, to give an example, was killed by whomsoever in the

3 municipality of Sanski Most but then later on buried in a grave or a mass

4 grave in Prijedor. In how far is this included in your report?

5 A. This is all about where the person lived in 1991. If this was

6 someone from the Prijedor municipality, killed in Sanski Most and buried

7 in a grave in Prijedor, the person would be reported, would be reported

8 in our study, if, of course, the person -- the record will be linked with

9 that census. So it is that we actually tried to trace individuals from

10 Prijedor municipality from 1991, their fate. We tried to estimate, to

11 calculate, how many of them just died or went missing. And we know that

12 most of them, almost all of them, died or went missing in the municipality

13 of Prijedor. This is what our study is reporting.

14 JUDGE SCHOMBURG: Thank you. We discussed already the question

15 how to define "elite," and I was to a certain extent alerted when you

16 mentioned this morning - and you do it in your report on page 11 - that

17 you cannot see an obvious pattern of targeting related to the educational

18 categories. Could you please give some more details on this, your

19 statement.

20 A. I suggest to look at figures 4A and 4B on page 12. This is next

21 page --


23 A. -- of the report. These two figures are made on the basis of the

24 numbers included in tables 5A and 5B. These are the same figures. It is

25 easier to see, to analyse numbers, by looking at the figures.

Page 8464

1 Figure 4A shows the risk of going missing or being killed for each

2 ethnic group separately by, at the same time, educational attainment, or

3 by education simply, by education, completed education. If we take as an

4 example the curve shown for Croats, the blue one, then we see a very clear

5 pattern, with low risks obtained for those with no education and

6 incomplete elementary education, and slowly increasing risks for those

7 with elementary education, secondary education, and highest risk for those

8 with post-secondary education. This is the blue curve for Croats.

9 If this type of pattern would be obtained for all other non-Serb

10 ethnic groups, we could speak about a pattern of selection in the missing

11 process and killing of the population. But if we look at the curve for

12 Muslims, the green one, the uppermost one, then we see that those with no

13 education or incomplete elementary indeed have lower risks of being killed

14 or missing. Then elementary, those with elementary and secondary

15 education, have elevated risks, clearly. But the risk for those with

16 post-secondary education is not higher than the risk for those with

17 elementary and secondary. It is actually a little bit lower.

18 If we then continue this type of observation with figure 4B, where

19 only men are shown, so the curve for Muslims is even more irregular. The

20 highest risk is only seen for those with elementary education. In my

21 view, it only confirms what we see also in other distributions of the

22 population, that the group who most frequently were killed and went

23 missing were men with relatively low level of education, elementary level

24 of education, and in eight groups from, say, 20 to 49, which are most

25 active military ages. So for men it would be a different mechanism, I

Page 8465

1 think, than educationally related targeting. But as I said, elite is

2 something that can be defined in very many different ways, and measuring

3 this is a very challenging task to do.

4 JUDGE SCHOMBURG: If not, even impossible, because no doubt it

5 could be related to high position, low position. You can discuss rich and

6 poor, occupation and so on. But I think this was covered.

7 Only one last question on the basis of a question by

8 Mr. Koumjian. Is there any hint or any information, reliable information,

9 you obtained that higher-educated-persons were "smart enough to leave

10 Prijedor before detained."

11 A. I don't have ready answer right now, but this can be

12 investigated. We could analyse the education level of those who left the

13 country.

14 JUDGE SCHOMBURG: And in case if it would be needed, you would

15 have the information who left the country between 1991 and then until

16 1992? Would this be possible?

17 A. Well, we don't have information for this particular period, but an

18 analysis could be presented quickly on the basis of the voters register,

19 which reports people living abroad by 1997, which is a distant period from

20 the period we are discussing here. But I believe we could use a sample

21 from a country like the Netherlands or a Scandinavian country like Sweden

22 and we could request lists of refugees from Bosnia and Herzegovina and

23 analyse educational level of those persons for this period.

24 JUDGE SCHOMBURG: No doubt insofar as the burden of proof is with

25 the Prosecutor, and from our side there will be no such request, taking

Page 8466

1 into account all the possibilities, difficulties with these numbers.

2 May I ask Judge Fassi Fihri, any additional questions? No

3 questions. There are no other questions emanating from my questions.

4 This is not the case. Then I have to thank you for your entire report and

5 the additional information you gave to us. And may I ask the usher to

6 escort the witness out of the courtroom.

7 In the meantime, unfortunately, I have to use two more minutes.

8 First, the Status Conference this afternoon, are there any objections by

9 the parties as regards the presence of Judge Vassylenko during the Status

10 Conference?

11 MR. KOUMJIAN: No, Your Honour.

12 MR. OSTOJIC: No, Your Honour.

13 JUDGE SCHOMBURG: Thank you. The Bench decided not to admit, at

14 least not today, the provisional marked Exhibit S338. No doubt there was

15 no possibility for the Defence to go into details of this document in the

16 framework of cross-examination. But most important is we should really

17 concentrate on Dr. Tabeau's report on Prijedor only, and therefore we

18 don't see the added value of this report, and therefore this is not

19 admitted into evidence.

20 [The witness withdrew]

21 JUDGE SCHOMBURG: Finally, I read out this morning my report to

22 the Chamber on the Rule 71(E). This document will be admitted -- or is

23 admitted into evidence as -- let me know CJ14, J14. And then this morning

24 I asked the parties whether there are any objections how to proceed on the

25 admission of evidence as suggested in this report. I have heard no

Page 8467

1 additional objections than those we can find in the transcripts, and

2 therefore all these documents listed in this report are hereby

3 admitted into evidence under the numbers you can read in this report.

4 This concludes today's hearing. The Court resumes tomorrow, 9.30,

5 in this courtroom. And a Status Conference in this case will be held as

6 from 2.30 this afternoon, also in this courtroom.

7 --- Whereupon the hearing adjourned at 1.13 p.m.,

8 to be followed by a Status Conference