Tribunal Criminal Tribunal for the Former Yugoslavia

Page 10334

1 Wednesday, 21 February 2007

2 [Open session]

3 [The accused entered court]

4 [The accused Pavkovic not present]

5 --- Upon commencing at 9.01 a.m.

6 [The witness entered court]

7 JUDGE BONOMY: Good morning, Dr. Ball.

8 THE WITNESS: Good morning, Your Honour.

9 JUDGE BONOMY: Your cross-examination by Mr. Ackerman will now

10 continue.

11 Mr. Ackerman.

12 MR. ACKERMAN: Thank you, Your Honour.


14 Cross-examination by Mr. Ackerman: [Continued]

15 Q. And good morning, Dr. Ball. How are you this morning? Did you

16 get some rest last night? Feel better?

17 A. Felt well last night. How did you rest?

18 Q. Not as well, probably.

19 A. I'll bet.

20 Q. I spent a lot of time last night and early this morning

21 considering what I might ask you about today, and I have significantly

22 restructured my questions in the hope that they are simple and can be

23 answered in many cases with just a yes or no answer. I think we might be

24 able to finish fairly quickly if you can cooperate in trying to answer yes

25 or no if possible. I have no objection to your giving explanations when

Page 10335

1 you think them necessary, however. Okay.

2 In the Milosevic trial at page 2178, you -- you told the Chamber

3 that 45 per cent of the material that you considered regarding KLA

4 activity was provided to you by the OTP and that 55 per cent of it came

5 from open sources. Can you tell this Chamber what were these open sources

6 that you got the 55 per cent of this information from.

7 A. We discussed this at some length yesterday, I believe. It was in

8 appendix 3.

9 Q. Okay. It's nothing beyond what's in appendix 3?

10 A. No.

11 Q. And did you engage in any kind of reliability checking regarding

12 material received from those open sources?

13 A. As we did with the death reports, we checked all of the sources

14 against each other so that the same events would not be counted in

15 multiple ways. Many sources report the same events and we strove to count

16 each event only once.

17 Q. Well, that would be more of an integrity check than a reliability

18 check, wouldn't it?

19 A. Well, I'm not sure how you're using those two terms. Perhaps you

20 could define them.

21 Q. All right. When I say an integrity check, to check that you're

22 not doubling up data. When I say reliability check, I'm talking about a

23 check on whether or not the data you were being provided is true?

24 A. No, sir. I'm not an investigator. I did not go to Kosovo and

25 investigate the cases.

Page 10336

1 Q. As to the OTP information, you were asked in the same part of the

2 Milosevic transcript if that was provided to you in the form of statements

3 or in statistical or summarised form, and your answer to that question by

4 Mr. Nice was: "It was provided in highly summarised form."

5 Now, what do you mean when you character the information as being

6 "highly summarised"?

7 A. I was given a spreadsheet. The spreadsheet contained fields which

8 described the very brief summary of the event in text as well as date and

9 place information. There were several other fields that were redacted.

10 Q. So like the material that came from open sources, with this

11 material from the OTP, you also had an opportunity to independently verify

12 the validity of that information?

13 A. No, sir. I'm not an investigator. I did not go to Kosovo and

14 investigate cases.

15 Q. Now, yesterday at page 89, line 21, in answer to similar

16 questioning you said that the ICTY provided you with a database of

17 materials. Is that this spreadsheet that you're talking about or was

18 there an additional database of materials that was provided to you by the

19 OTP?

20 A. With respect to the KLA activity, this is the only material that

21 was provided to me. There were also a number of other sources that were

22 also provided to me on other topics. We've discussed them.

23 Q. Okay. You do know at the time that the OTP provided you with this

24 information regarding KLA activity that the OTP had certain theories about

25 this case, didn't you?

Page 10337

1 A. I've read the indictments, if that's what you meant.

2 Q. Yeah. And engaging your assistance, what they were looking for

3 was help from you in supporting those theories as contained in the

4 indictment, as you appropriately mentioned.

5 A. That's not the way that Mr. Nice or any of his colleagues

6 characterised what they were looking for from me. They said they wanted

7 me to provide a statistical evaluation of the patterns of these four

8 phenomena.

9 Q. Well, I suspect you'll agree they wouldn't have found you very

10 useful if your analysis had destroyed their theories?

11 A. Apparently, my opinion of Mr. Nice is somewhat higher than yours.

12 Q. At page --

13 JUDGE BONOMY: Just one second.

14 What are the four phenomena?

15 THE WITNESS: KLA and Serb police interactions, so KLA activity is

16 the first of the phenomena; NATO air-strikes is the fourth -- second of

17 the phenomena, excuse me; killings of Kosovar Albanians is the third; and

18 migration of Kosovar Albanians, the fourth.

19 JUDGE BONOMY: Distinguishing between migration and killing.

20 THE WITNESS: Yes, sir.

21 JUDGE BONOMY: Thank you.


23 Q. At page 9624 yesterday you said this in how you weigh certain

24 evidence: "Evidence that goes against the interest is much more weighty

25 than evidence that goes in favour of interest."

Page 10338

1 Now, if that's your approach, since a great deal of your data came

2 from the OTP, did you consider that it might have less weight than other

3 data?

4 A. No, I did not.

5 Q. Okay. I want to talk now about your hypotheses and they're listed

6 in the report; there are three of them. The first one is action by the

7 KLA motivated Kosovars to leave their homes, either directly because the

8 KLA ordered people to leave, or indirectly because Kosovars fled fighting

9 between KLA and Yugoslav forces. Now, to analyse and deal with this

10 hypothesis, you needed to have the most reliable and complete information

11 about KLA activity that you could get your hands on; correct?

12 A. That I could get my hands on, that's correct.

13 Q. Mm-hmm. And you got no information directly from the KLA, did

14 you?

15 A. None.

16 Q. I think the only information that you had was information that

17 came either from the OTP or from these other open sources that you have

18 referred to.

19 A. I believe, principally, the information came from Yugoslav

20 sources -- both in the ICTY data and in my own research.

21 Q. Did the OTP ever give you a statement or a copy of testimony of

22 the Chief of the General Staff of the KLA, Mr. Bislim Zyrapi?

23 A. No, I was -- I never received any testimonies whatsoever.

24 Q. You told this Chamber yesterday that 60 per cent of the peaks in

25 killing and migration occur with no prior NATO or KLA activity. Do you

Page 10339

1 remember saying that?

2 A. I think that was derived from a table and we can consider it. I

3 can't confirm the statistic without referring to the table, but you can

4 continue with your question while we look it up.

5 Q. Wouldn't it have been fair to say that no prior NATO or KLA

6 activity that we know about?

7 A. Yes, that's fair.

8 Q. Okay.

9 If there was independent KLA activity or NATO activity that you

10 didn't know about that coincided with killing or migration, that could

11 have a profound effect on your results, couldn't it?

12 A. Probably not a profound effect. It would have to be systematic

13 and overwhelming for it to have a profound effect. Finding one or two

14 municipalities or even three or four municipalities in which that was the

15 case would have relatively little effect. Again, as we discussed

16 yesterday, the peaks and valleys method is a weak method for evaluating

17 the causality between these patterns. You and I discussed this at some

18 length yesterday and you indicated that you had difficulty finding the

19 points in my research where I made the reservations about that. I've

20 since revised the research and I can point you to several other places

21 where we make precise this reservation --

22 Q. Again --

23 A. Excuse, sir, if I may finish.

24 Q. If --

25 A. Your Honour, may I finish?

Page 10340

1 Q. I'm happy with where we got yesterday with that. I really don't

2 want to go into that anymore, unless the Judges want to hear from you

3 about it then you can of course talk about it.

4 JUDGE BONOMY: What is you want to say, Dr. Ball?

5 THE WITNESS: Well, yesterday Mr. Ackerman brought this up at some

6 length and now he seems to be bringing it up again notwithstanding his

7 demurral and he complained yesterday that he was unable to find our

8 reservations. I pointed him to one of those reservations and I would like

9 to point out another of those reservations in case he is not able to find

10 this one as well. It is on the second page of our 19 February 2002 report

11 in the discussion section, and it says: "The primary drawback of this

12 method is that it considers only the effect of the hypothesised cause on

13 the hypothesised effect at the time of the effect's peak or maximum

14 point."

15 We go on to discuss a subtler and finely ingrained analysis and

16 refer to the regression analysis as far more appropriate for a causal

17 analysis. Therefore, to get back to your earlier question, where you

18 implied that somehow finding additional KLA or NATO activity would have a

19 profound effect on our findings, I cannot agree. As we said in each of

20 these papers and our reservations about this technique, the technique is

21 not sufficiently finely grained to make a clear determination about

22 rejecting causes. So perhaps that helps clarify my disagreement with your

23 earlier statement.


25 Q. You just told us a moment ago that if the -- if you had found

Page 10341

1 conflicting data for as many as three or four municipalities it probably

2 wouldn't have changed your conclusion. How many municipalities did you

3 actually analyse?

4 A. All of them, 29.

5 Q. Okay. So four would be like one-sixth of 29, and if your data was

6 incorrect regarding one-sixth of those 29, isn't that a fairly profound

7 number, a high percentage?

8 A. Well, this is actually precisely why I asked the Bench's leave to

9 review the 19 February report and my conclusions I just read.

10 Mr. Ackerman is drawing unreasonable conclusions about the efficacy of

11 this method. Even if all 29 municipalities coincided with one specific

12 peak, with the peak in that municipality, if there were NATO and KLA

13 activity coinciding with that peak, that would be an inadequate basis for

14 finding that NATO or KLA activity were the cause of that activity. The

15 reason is explained in, I believe, fairly clear language in section 1.2 of

16 the 19 February 2002 report, which you have, in which we discuss - and I

17 used this metaphor yesterday - that coincidence does not imply causation.

18 Now, I understand that you're very adept at logical streams, so I assume

19 you understand that one or should we review it?

20 Q. No, don't review it.

21 A. The point is, Mr. Ackerman, you're giggling notwithstanding, that

22 you cannot find that simply because the sun rises and later that day there

23 are killings or refugee flow, that the sun's rising has caused them. The

24 coincidence of killing and migration is an inadequate basis on which to

25 make the conclusion that the NATO or KLA activity may have caused it;

Page 10342

1 rather, we need to see not only the coincidence but the common absence of

2 the two, so that if there is killing or migration without NATO or KLA

3 activity we must ask what has motivated that. If we look at the graphs,

4 we see that many times there are coincidences whereby there has been some

5 killing or migration on the same day as bombing or KLA activity but in

6 many more instances there is killing and migration without such activity.

7 So focusing, I believe, on this point with your proportions merely

8 distracts the Court from the broader point which we have made repeatedly

9 throughout these reports.

10 JUDGE BONOMY: Going back to what you said a moment ago about even

11 if all 29 municipalities coincided with one specific peak, if there were

12 NATO and KLA activity coinciding with that, would that not undermine your

13 conclusion that NATO and KLA activity was inconsistent with the movement

14 of people?

15 THE WITNESS: That is -- that's a good question and I think that

16 to answer it we might want to consider a particular graph and ask

17 ourselves about a specific pattern. So if we could look at the graph for

18 Djakovica on page 10 of my January 2007 report, we'll get a bit more

19 detail on that.


21 THE WITNESS: So we notice that there is a single KLA action and a

22 bombing that coincide with the peak of killing and precede a peak of

23 migration very early in the period late March/early April, but then there

24 is a series, and that is the kind of coincidence that we're talking about

25 here. However, then there's a series of KLA activities and NATO

Page 10343

1 air-strikes that are not -- that do not correspond -- oh, I'm sorry, this

2 is not the correct -- the graph that's shown on the screen is not correct.

3 I'm not sure if you would like me to pause. This is -- I'm looking at the

4 Decani graph from the 2007 --

5 JUDGE BONOMY: I thought you were looking at Djakovica.

6 THE WITNESS: Djakovica, yes. Decani is on the same page. It's

7 page 10.

8 JUDGE BONOMY: Yes, we have it in hard copy so you can carry on.

9 THE WITNESS: So there's a series of bombing and KLA activities

10 that follow those peaks, and what's striking is that the bombing and KLA

11 activity that follow the peaks are not -- do not coincide with any peaks.

12 A causal argument would suggest that continuing bombing or continuing KLA

13 activity would cause continuing killing or continuing migration if the

14 killing -- excuse me, if the bombing and KLA activity were causes.

15 So the question that Mr. Ackerman has raised is what if all 29 or

16 some high fraction of the 29 municipalities had the pattern we see for

17 Djakovica, where there is, in fact, a coincidence. My response is it's

18 not just the coincidence of the peaks that matters. We need to see both

19 the coincidence when there's a positive effect and the coincidence when

20 there's a negative effect. What we get here is the coincidence when

21 there's a positive effect we get killing and migration rising after a KLA

22 action and a bombing, but then a series of KLA and NATO actions that are

23 not followed by any significant killing or migration.

24 JUDGE BONOMY: Thank you.

25 THE WITNESS: Mm-hmm.

Page 10344

1 JUDGE BONOMY: Mr. Ackerman.


3 Q. Dr. Ball, with, you know, huge respect for your expertise and even

4 for your political views, it seems to me that your answer to the Judge was

5 just riddled with assumptions that may or may not be correct. Let me just

6 give you an example. You said you would expect to see continuing

7 migration if there were continuing bombing or activity, I think is what

8 you said. And that assumes that if a village is completely emptied

9 because of KLA activity or bombing, that you should see people leaving

10 that village later. Well, they can't if they're not there, so that

11 assumption, it seems to me, holds no water if that's the assumption you're

12 making but maybe I misunderstood.

13 A. I'm sorry, what's the question?

14 Q. Have I misunderstood?

15 JUDGE BONOMY: The position is -- the question is quite clear that

16 people have gone, as you've indicated by the peak, and therefore there

17 can't be another peak. That's the suggestion.

18 THE WITNESS: Well, we estimate there to have been about 875.000

19 plus or minus some uncertain figure of total refugees who left Kosovo

20 during the period between March and June. There were many more Kosovars

21 before that. Not everybody left. So it is not at all clear to me that

22 these municipalities were completely empty of people; to the contrary,

23 when people returned, I heard anecdotally from many people that there were

24 plenty of people who had never left. So there were people around who

25 could have been conceivably continuing migrants throughout the period if

Page 10345

1 the bombing in fact or the killing were their motivations for leaving.


3 Q. Well, for that to make sense you need to look at municipalities as

4 a whole instead of villages and I was talking about villages. Of course

5 within the entire municipality there were people that never left, but some

6 villages were completely emptied of people for one reason or another that

7 we really don't understand at this point I guess and that's the question

8 I'm asking you, not about municipality. So once you've got that peak for

9 a particular village, you're not going to get another one if the people

10 left. That's the point, and it's true, isn't it?

11 A. Well, no, it's unfortunately not true because the peak here is not

12 about villages. These peaks are at the municipal level.

13 Q. All right. I think we've beaten this horse to death. Let's go to

14 your second hypothesis. "Air attacks by NATO created local conditions

15 that led to Kosovars being killed and leaving their homes. The NATO

16 influence could either have been direct because people were killed in

17 air-strikes and others fled or indirect because local Yugoslav authorities

18 responded to air-strikes by killing Kosovars and forcing them from their

19 homes."

20 Now, with regard to this hypothesis, again the integrity of your

21 findings depends upon the integrity of your data or the reliability of

22 your data. For instance, if there were air-strikes that coincided with

23 killings or migration that you didn't know about, then your findings could

24 be in error; true?

25 A. If there were lots of air-strikes that spread across the period

Page 10346

1 and linked to periods, then that's possible. However, what we know from

2 adding additional information about air-strikes to this pattern is the

3 reverse and we discussed this at some length yesterday, that most of the

4 air-strikes that occurred occurred in late May and early June. So what's

5 much more likely about air-strikes is that adding data about air-strikes

6 will strengthen the finding here and in fact create the result which I

7 believe to be a coincidence that statistically the implication would be

8 that air-strikes ceased or had a negative effect on killing and migration.

9 Q. Well, that was that rather bizarre result that you called a

10 coincidence yesterday where your model found that NATO air-strikes in I

11 think April and May saved lives weeks earlier so you had to reject that.

12 A. No, sir, it was not a bizarre result, it was a fairly predictable

13 result which is why we excluded it from our original results. We did not

14 do that calculation in our original report. It's a fairly obvious result

15 if one thinks about the nature of what happens if you have lots of

16 air-strikes later than, in this case, weeks and weeks later than, the

17 killing and migration. The statistical effect of that will be to suggest

18 that there's somehow the killing -- I mean, the air-strikes are

19 suppressing the killing and migration. We don't believe that to be the

20 case, but perhaps -- perhaps someone could argue that. I wouldn't.

21 Q. Well, I'll accept that you don't think it's a bizarre result that

22 bombing in April saved lives in March. I think it is. I want to ask you

23 this: You got no data from NATO, did you?

24 A. We got no data from NATO, that's correct.

25 Q. You requested no data from NATO?

Page 10347

1 A. We did not request data from NATO, no.

2 Q. You did not ask the OTP to get data from NATO for you?

3 A. No, we didn't.

4 Q. You simply relied upon this secondary data that came primarily

5 from, I think you said, a Yugoslav web site and maybe a few other sources?

6 A. Principally Yugoslav web sites. I would think that that's

7 actually primary data since the Yugoslav web sites alleged that they were

8 eye-witness accounts. That seems to me primary data.

9 Q. If NATO had been willing to cooperate with you, if you had even

10 made an effort to get NATO to cooperate with you, and the OTP, they could

11 have told you exactly where they struck, what they struck with, what date,

12 what time, in great detail. They would have that information, and that

13 could be helpful, couldn't it?

14 A. It probably could be if it weren't classified.

15 Q. You never had that data to work with, and I know among other

16 talents you have, one of them is you are a pretty sophisticated computer

17 programmer. You really understand that process pretty well, and so you're

18 certainly well family with the phrase "garbage in and garbage out," aren't

19 you?

20 A. I am familiar with it as used inaccurately, yes. It's actually

21 not an appropriate understanding of how statistical analysis works.

22 Q. Well, statistical analysis doesn't work at all if you put garbage

23 into the computer you can't expect to get anything but garbage out. Your

24 data has to have some integrity, doesn't it.

25 A. Well, I'm sorry. Is your question about my integrity or are you

Page 10348

1 just asserting --

2 Q. No, not your integrity. The data, the data has to be accurate.

3 If you put a bunch of false data in there, then your result is going to be

4 false? Isn't it obvious?

5 A. Are you asserting or are you asking?

6 Q. I'm asking you.

7 A. It's not obvious. Let me explain to you why. Statistical

8 analysis is not about making sausage, as apparently Mr. Ackerman seems to

9 think. Statistical analysis instead is much more like listening to a

10 voice in a crowded, noisy room. You listen to -- you're trying to listen

11 to a voice, concentrate on what the voice is telling you while there are

12 many other voices speaking over you or shouting. Your job as a

13 statistical analyst is to make out the voice that you're trying to hear

14 and then make sense of it. The voice may be speaking softly while others

15 are shouting. It may have a strong accent.

16 I use this metaphor to explain that it is not -- a statistical

17 analysis is, in fact, to the contrary, garbage in garbage out; rather, all

18 statistical -- modern statistical techniques, sometimes called data mining

19 techniques, are about looking at large bodies of information and finding

20 patterns even when there is a lot of chaff mixed in with that data. This

21 notion of garbage in, garbage out, while apparently attractive to people

22 who are not computer programmers, is an inaccurate assessment of the way

23 this work is undertaken.

24 Q. You know what --

25 JUDGE BONOMY: Mr. Ackerman -- sorry. I don't want to disturb --

Page 10349

1 carry on. I'll ask the question when you finish.


3 Q. I think that answer is just gobbledygook. You have to agree that

4 if all the data you had was absolutely wrong, then there's no way that you

5 get a result that is meaningful?

6 A. If all the data were wrong, that's correct, there would be no

7 meaningful result.

8 Q. And there's a point between all and -- all wrong and all right,

9 where you cross a threshold?

10 A. Well, there's actually not a point. There's a continuum in which

11 at each point in the continuum where the data are noisier, is the term, as

12 we have more and more chaff, more and more confusing data in the input

13 string, we have a higher and higher error. In fact, that's precisely why

14 statisticians report the error in their results. We do a number of

15 tests - sensitivity analysis is the formal term for these tests - to try

16 to figure out how sensitive our results are to noise or lies or

17 inaccuracies or biases in the data.

18 That's the point of, for example, the entirety of the appendix in

19 the Policy or Panic report. There are a number of sensitivity analyses in

20 the January 2002 report and we spend an enormous amount of effort doing

21 this kind of analysis to ask precisely the question what fraction of the

22 data could be wrong before our results would change. And we have provided

23 extensive analysis to that effect in these reports.

24 MR. ACKERMAN: Judge Bonomy, you had some kind of --

25 JUDGE BONOMY: Yes. Mr. Ackerman, I have a vague recollection

Page 10350

1 that somewhere earlier in this case we had figures for NATO strikes,

2 air-strikes in -- used in cross-examination. Do you remember that or is

3 it fantasy on my part?

4 MR. ACKERMAN: I don't remember it. I know that there was some

5 data that came from the White Book regarding some NATO air-strikes. I'm

6 not sure that's what we're talking about.

7 JUDGE BONOMY: What I was going to ask you what was the source of

8 that data if you knew it, and it's the White Book, is it?

9 MR. ACKERMAN: I'm not able to precisely answer your question. I

10 know there was some data from that, but what apparently the Yugoslav

11 government was compiling there was NATO activity that they considered to

12 be war crimes or --

13 JUDGE BONOMY: Was General Naumann not asked some questions about

14 the extent of NATO air-strikes?

15 MR. ACKERMAN: I really would have to look at the transcript. I

16 don't recall, Judge. But I don't think there's any data that ever came

17 from NATO that I'm aware of other than what we can find on the NATO web

18 site. There's some data there.

19 JUDGE BONOMY: Thank you.

20 [Defence counsel confer]

21 MR. ACKERMAN: Judge, I was just reminded that at one point I

22 think with General Naumann I'd had some maps that showed NATO bombing,,

23 where the bombs fell and cluster bombs and things of that nature, that

24 were -- a copy was provided to each of you but then you determined that I

25 couldn't use it with General Naumann. Maybe that's what you're thinking

Page 10351

1 of.

2 JUDGE BONOMY: And were they a compilation from the White Book?

3 MR. ACKERMAN: I think they were a compilation based upon data

4 gathered by Yugoslav forces on the ground. That's what I think they are.

5 JUDGE BONOMY: I think Mr. Aleksic is perhaps going to give some

6 guidance.

7 [Defence counsel confer]

8 MR. ACKERMAN: Okay. I'm now told that data comes from KFOR.

9 MR. VISNJIC: Maybe I would help, Your Honour, I believe that

10 map -- [Interpretation] I believe that's the map where the uranium

11 ammunition used was shown. I think it's the map on which cluster bombs

12 and uranium ammunition were shown, and they were probably obtained from

13 KFOR, because otherwise, we wouldn't have had such data, and part of it

14 was contained in the UN report on uranium ammunition.

15 JUDGE BONOMY: So that wouldn't help on the dates of the bombing;

16 it's information about the results of the bombing, I think. Yeah. Please

17 carry on, Mr. Ackerman.


19 Q. Dr. Ball, we're not doing a very good job of moving as rapidly as

20 I hoped we could. I'm going to try to avoid some of these highly

21 controversial issues with you, if I can. I want to go to your third

22 hypothesis, and it was:

23 "A systematic campaign by Yugoslav forces drove Kosovar Albanians

24 from their homes and killings were used either to motivate the departures

25 or the killings were a result of the campaign."

Page 10352

1 And if I understand your position correctly, it is that you didn't

2 have sufficient data to draw any conclusions regarding this hypothesis?

3 A. That's correct.

4 Q. And since your initial report you haven't received sufficient

5 additional data to allow you to draw any conclusions regarding this

6 hypothesis?

7 A. No. I have not continued analysis on this report since 2002.

8 Q. At paragraph 5.1 of your initial report, then, and I can only give

9 you a K number -- page numbers -- 3831, since I haven't got page numbers

10 on mine for some reason, but it's paragraph 5.1 under the heading: "Kosovo

11 Liberation Army Activity." Are we all there?

12 A. Yes.

13 Q. It's also on the screen. You'll see your description of that

14 activity. "Information on KLA activity was obtained from interview

15 accounts and a variety of non-governmental reports summarised and provided

16 to this project by the ICTY.

17 "Using that information," and this is what I'm really interested

18 in, "the present study counted the number of reported battles between the

19 KLA and Yugoslav forces occurring in each municipality over time."

20 True?

21 A. Yes.

22 Q. And that has not changed -- you wouldn't change that, since this

23 report was issued?

24 A. No.

25 Q. Okay. The only other KLA activity to refer to in that paragraph

Page 10353

1 were isolated KLA attacks on Serbs, which didn't figure into your analysis

2 at all; correct?

3 A. No, that's not correct. Those were a second variable we used to

4 analyse KLA activity.

5 Q. Okay. But the primary KLA activity that you analysed was that

6 that involved reported battles between the KLA and Yugoslav forces?

7 A. There really wasn't a distinction between primary and secondary,

8 we used them both.

9 Q. The conclusion that you drew from this data then is over in the

10 last paragraph of 5.1. "There is no clear cause-and-effect relationship

11 between KLA activity and the pattern described here." Correct?

12 A. Yes, that's right.

13 Q. Now, what I'd like to do is look at this from a little bit

14 different perspective. If we're talking about data that counts the number

15 of reported battles between KLA and Yugoslav forces occurring in each

16 municipality over time, let's substitute Yugoslav force activity for KLA

17 activity so that when we get to your conclusion we can say: Thus there is

18 no cause-and-effect relationship between Yugoslav force activity and the

19 pattern described here. That would be a fair conclusion from the data,

20 wouldn't it?

21 A. No, it would be a conclusion if one assumed that the only Yugoslav

22 activity that occurred was that represented by the KLA interactions. It

23 is as if the Yugoslav activity -- the Yugoslavs were not, for example, the

24 authorities in the region, that they did not have police stations

25 throughout the region, it is as if they did not have any defence

Page 10354

1 whatsoever against NATO, it is as if their entire existence can be reduced

2 to this series of conflicts or disappearances committed by the KLA. That

3 seems to me quite a stretch.

4 Q. Well, you don't have any Yugoslav force activity beyond that which

5 is represented by reported battles between KLA and Yugoslav forces

6 occurring in each municipality over time?

7 A. That's correct. And therefore we draw no conclusions.

8 Q. Yeah, so you can substitute Yugoslav activity for KLA activity?

9 A. No.

10 Q. And you get the same place.

11 A. No, you can't, because doing so requires a massive assumption

12 which you've elided there.

13 Q. Well, I think doing so interferes with your assumption that

14 Yugoslav activity might have had some role, although you have no data to

15 indicate it at all. And I think it's extremely fair, since you are

16 relying on clashes between KLA and Yugoslav forces, that you can use

17 either term in the conclusion you get to the same place because you're

18 using the same data. How does that not take sense? You didn't take into

19 account that there were police stations in Kosovo --

20 MR. STAMP: He's getting into various statements without giving

21 the witness an opportunity to suggest each --

22 MR. ACKERMAN: Well, I haven't noticed that he has any hesitancy

23 to answer me, Judge.

24 THE WITNESS: I'd be happy to show hesitancy if you would like,

25 Judge.

Page 10355

1 JUDGE BONOMY: Yeah, we're getting into intense debate now, but it

2 seems fairly obvious that what Dr. Ball has been doing in relation to

3 Yugoslav forces is posing a hypothesis and the hypothesis relates to

4 illegal Yugoslav force activity. And as you know, Mr. Ackerman, there's

5 plenty of anecdotal material, on the strength of which I will not comment,

6 to justify starting off with that as one of the hypothesis here. Now, it

7 seems to me there's a clear distinction between looking at that question

8 and looking at the legitimate engagement of the Yugoslav forces with the

9 illegal activity of the KLA, which is the issue that we're addressing at

10 the moment. And I find it difficult to see that there is a way of

11 substituting the Yugoslav force activity of that nature for the KLA

12 activity that was being considered in 5.1.

13 MR. ACKERMAN: Well, Your Honour --

14 JUDGE BONOMY: If you want to explore it more, please do.

15 MR. ACKERMAN: Your Honour, I'm trying to ask questions within the

16 confines of the data that was used by Dr. Ball and what was furnished to

17 him by the Office of the Prosecutor. And that's what he used to draw his

18 conclusions. Now, it may be that there is data outside what he used that

19 maybe could have assisted, that maybe should have been supplied to him by

20 the OTP and wasn't, but it could also be that the OTP had no data that was

21 reliable of independent Yugoslav force activity beyond clashes with the

22 KLA. It seems to me they would have provided that had they had it, and if

23 they had provided it, it seems to me that he would have factored it in.

24 That's my point and I think it's a good one.

25 JUDGE BONOMY: It may be but it's a rather different point from

Page 10356

1 the one I think you're making at the moment in the question.

2 MR. ACKERMAN: I will move forward.

3 Q. Is there not a problem with your analysis in that the fact that

4 you limited it to only these three hypotheses which you pretty narrowly

5 constructed. For instance, you made certain assumptions about Kosovar

6 Albanians leaving their homes, when they did it, why they did it -- not

7 necessarily why they did it, but when they did it, how long it took them

8 to get from there to the border. Lots of assumptions were made about that

9 one.

10 A. Let's see. As you correctly note, we did not make any assumptions

11 about why they left their homes.

12 Q. Right.

13 A. Nor did we make assumptions about when they left their homes. We

14 measured when they crossed the border and we looked at the pattern of all

15 people who we could interview in camps who crossed the border on that day

16 and asked those people how long it took them to get to the border. That's

17 not an assumption, that's a model. Models are very useful because they

18 allow us to make a projection within known error. In fact, we devote

19 pages and pages of analysis in the book Policy or Panic to the effect of

20 the model of transit time on the results. Allow me for a moment to review

21 what I mean by transit time because I think it's crucial to the question

22 that's being asked here. This is, I want to emphasise, not an assumption.

23 We went out and asked people in camps, refugee camps: How long

24 did it take you to get to the border? And we knew that people from

25 different parts of Kosovo would take different amounts of time to get to

Page 10357

1 the border. We also knew that some people even leaving the same village

2 would be able to get to the border crossing point relatively quickly,

3 perhaps because they had a vehicle or perhaps because they did not

4 encounter any obstruction along the way, whereas other people trying to

5 get to the border would have a lot of trouble, they would be on foot or

6 they could be on buses that were detained or they could have been detained

7 for other reasons.

8 So we knew there would be a distribution of times that it took to

9 get from people's homes to the border and I'm calling this distribution

10 transit time. We measured the transit time among people in our surveys

11 and we found that for most people they were able to get out in the first

12 day. Over the course of the conflict, the amount of time people spent in

13 transit increased. So if you look at people crossing the border in early

14 March -- excuse me, late March or early April, they're in transit for a

15 relatively short time. If you look at people crossing the border in late

16 April, early May, they're in transit for substantially longer.

17 So what we did with the transit time structure is we said, okay,

18 given people crossing the border on a given day, we know that 50 per cent

19 of them, say, left home today. We know that 20 per cent of them or 30 per

20 cent of them left home yesterday, some smaller fraction left home the day

21 before that and so forth in a declining function. So we took all the

22 people from each village who crossed the border on a given day and we

23 pushed back their date of departure from their homes by the proportion

24 that was shown in the transit time distribution.

25 Now, that's a model and it's an estimate. Consequently, it has

Page 10358

1 error. And so we modelled the error in two different ways. The first way

2 we modelled the error, to see if it would impact our results, was to use

3 the simple sampling error which was the error that resulted because we had

4 taken probability samples of people in refugee camps in order to obtain

5 information about their transit time; actually some of the information was

6 probability sampled, others was not. About -- it was about half and half.

7 The second way, and actually the much more relevant way for this

8 purpose, for this question I think, that we analysed the impact of our

9 transit time analysis is through sensitivity analysis, which I alluded to

10 earlier. And in this case, what we did is think about all the factors

11 that shape the transit time distribution and we identified six different

12 factors that affect how long it would take people, statistical factors

13 that would affect the structure of how long it would take people to get to

14 the border. We varied all of those factors to what we found extreme

15 positive and negative values and recalculated the migration analysis, and

16 we found - and you will find if you look in appendix 1 of Policy or

17 Panic - that the transit time assumptions have almost no impact on the

18 conclusions that we draw about the process of migration, either looking

19 overall, looking at all of Kosovo, or looking at any of the four regions

20 that we identify, north, south, east, and west.

21 So let me be very clear with Mr. Ackerman that this is not an

22 assumption, but rather, a model and an analysis that we spent quite a few

23 weeks examining from many different angles.

24 JUDGE BONOMY: Mr. Ackerman, is Policy or Panic an exhibit in this

25 case?

Page 10359

1 MR. STAMP: A Defence exhibit.

2 MR. ACKERMAN: I think it is, yes.

3 JUDGE BONOMY: It is. Thank you.


5 Q. Dr. Ball, let me -- I didn't mean to get such a long answer from

6 you. Most of it you gave us -- most of it's an answer you gave us

7 yesterday, but let me see if I can talk about an assumption that you will

8 agree that you made. I think it's the case that you made an assumption

9 that if NATO bombing occurred after, significantly after, like four days I

10 think is your number, four days after people left the village, then the

11 NATO bombing was not a factor in their leaving?

12 A. Okay. We're going back into the peaks and troughs analysis here,

13 Mr. Ackerman, and I would hate to repeat my reservations with that

14 analysis again. Instead, let me say that our analysis rejecting the

15 hypotheses of NATO and KLA activity being plausible causes of killing and

16 migration rest on the residuals analysis which is in the original report

17 figures 10 and 11 of which, as far as I can tell, not a question has been

18 raised.

19 Now, I will -- I'm happy if it pleases the Bench to repeat my

20 reservations, but since you have just complained that I repeated my

21 correction of your inaccurate characterisation of our work as an

22 assumption, you then complained that I had to correct you at some length

23 because of your failure to understand what I said yesterday. I would hate

24 to do that again unless that would please the Bench.

25 Q. I really don't think you need to say it again if you said it

Page 10360

1 yesterday. That's up to the Chamber, if they want to hear it again or

2 not. I suspect they don't.

3 You will agree that people can't act on information if it's not

4 communicated to them?

5 A. Okay.

6 Q. And the converse of that is true. So if people live in a village,

7 for instance, just as a for instance, if people live in a village that has

8 a Yugoslavia army barracks in that village and the information that's

9 communicated to them is that NATO is bombing Yugoslav Army installations,

10 it seems to me they must consider that information and one of the things

11 they may decide to do based upon that information is leave the village.

12 If that installation -- and they did that, they left the village. If that

13 installation was then bombed, say, ten days later, your conclusion would

14 be that the NATO bombing had absolutely no effect on their leaving the

15 village, when in fact that was exactly the reason why they left the

16 village; correct?

17 A. My conclusion would be that the NATO bombing of the village had no

18 effect on their leaving. My work is not an analysis of the psychology of

19 rumour, Mr. Ackerman.

20 Q. Well, I'm not talking about rumour, I'm talking about a

21 hypothetical situation that I think probably has a lot of basis in reality

22 with regard to what happened on the ground in Kosovo. You need to

23 understand what we're not doing here is some kind of an academic

24 exploration of statistical methods; what we're doing here is deciding

25 whether or not there is sufficient information available to this Court

Page 10361

1 that would allow them to find beyond a reasonable doubt that certain

2 matters alleged in the indictment in this case is true. And I think one

3 of the things we're all having difficulty with is where -- you've made it

4 very clear, I think, that at best your conclusions don't prove anything.

5 I think you've used those exact words, that they don't prove anything.

6 A. My work disproves two hypotheses that may be relevant to the

7 Court; that's correct.

8 Q. Well, I don't think you want to go as far as disprove either

9 because disproving is proving?

10 A. That's sophistry, Mr. Ackerman. We are quite clear in our

11 analysis that we reject the hypotheses that either KLA or NATO activity

12 could have been the causes of killing or migration. I will not play with

13 the words that you've tossed out there. That's absurd.

14 Q. Did you, just for instance, did you look at the Kosovo media,

15 television, radio, newspapers, to see what kind of information was being

16 transmitted to the people in the villages regarding what might happen next

17 in their lives? Did you look at that to see if you could see if that had

18 any effect on whether or not they might have left their villages?

19 A. This analysis is neither an analysis of the psychology of rumour

20 or of propaganda, sir.

21 Q. Well, if you want to refer to it as rumour and propaganda, you

22 can. But what if NATO or what if KLA or what if Yugoslav forces were

23 actually warning villagers that there was going to be a battle in their

24 village and based on that warning, they appropriately left that village.

25 I mean, we have some information in this case that that was the case.

Page 10362

1 That's not rumour; that's reality. And you're ignoring reality, it seems

2 to me, in your report.

3 A. All right. Well, let's pop back to reality then, Mr. Ackerman.

4 It may be a refreshment. In figure 2 of my original report, page 6, we

5 show a comparison of people leaving their homes for whatever reason and

6 people being killed. If we then -- you'll notice that those two curves

7 are very similar and I can provide statistical descriptions of that

8 similarity that show that they are actually extremely similar patterns.

9 Similarly, figures 4, 5, 6, and 7 show the same pattern broken down by

10 region. If people were leaving their homes as a result of the rumours

11 that Mr. Ackerman alludes to, it fails to explain why they were being

12 killed at precisely the same times and same places, unless perhaps

13 Mr. Ackerman knows something about the potential fatality of newspapers

14 which has eluded us.

15 My explanation of these two patterns coinciding so very closely in

16 time and space is that, in fact, there is a common cause to both, and it

17 is highly unlikely to me that it is newspapers or any other -- a rumour or

18 anything else; rather, something that I believe we can agree that a

19 killing is a violent act, so there is something violent going on that is

20 creating the pattern of migration and killing, creating both, because it

21 is very difficult to explain how we would have such close patterns in time

22 and space without a common cause. Because we interpret the likely cause

23 as a violent one, we focussed our hypotheses on those parties able to

24 deploy violence in the conflict; hence, the three hypotheses.

25 JUDGE BONOMY: In relation to your findings, where you that - and

Page 10363

1 I'm looking at figure 8 and the analysis related to figure 8 in your

2 original report - where you say that in 11 of the 29 municipalities, KLA

3 activity coincided with the overall peak and we were to -- and we were to

4 be looking at only these municipalities, because that's -- the

5 indictment's divided into municipalities. I know that's not what you're

6 looking at at the moment, but let's assume that these were the 11

7 municipalities that we were looking at in this case.

8 THE WITNESS: Yes, sir.

9 JUDGE BONOMY: What would your conclusion be in relation to them?

10 THE WITNESS: It would depend on what happened at moments other

11 than the peaks, sir. So if the -- there might be a coincidence of KLA

12 activity, for example, and killing in a particular municipality at the

13 peak of that killing, but if there were continuing KLA activity throughout

14 the rest of the period without there being killing, we would have to ask

15 the question: Well, what's happening with the KLA activity then? This

16 gets back to this --

17 JUDGE BONOMY: We can find that in here in relation to each

18 individual municipality, can we?

19 THE WITNESS: Yes, sir. There are 29 graphs in the report that I

20 presented in the January 2007 --

21 JUDGE BONOMY: Yeah, and that will deal with it for every separate

22 municipality?

23 THE WITNESS: Yes, Your Honour. All four series are on those

24 graphs, and I'm happy to present them in other formats if it would please

25 the Bench.

Page 10364

1 JUDGE BONOMY: And one of the important ones, just as an example,

2 is Pristina --

3 A. Mm-hmm.

4 Q. -- and your conclusion about Pristina is that it's not consistent

5 with the overall view that you have of Kosovo in general. Is that

6 correct?

7 THE WITNESS: Pristina is a special case, because it is the site

8 of an overwhelming amount of NATO and KLA activity, being the capital.

9 Let me pull up that graph. Let's see. Page 16 --

10 JUDGE BONOMY: It's page 19.

11 THE WITNESS: Yeah. Pristina. Yeah. So you can see that

12 Pristina was the target of an almost daily bombing campaign throughout the

13 period of analysis here. But the peaks occur at the very beginning and

14 then again in mid-April. So, you know, there's bombing every single day;

15 yet there's migration only at specific moments. Why would -- you know,

16 why would there be migration on some days and not on others if there's

17 bombing every day? I mean, I think that would be my principal question

18 about this relationship.

19 The -- you know, in particular you can see that there's a

20 continuous bombing pattern toward the later period after the 7th of April

21 or so, or the 10th of April, it seems, whereas the peak of migration comes

22 right at the beginning of the conflict where there's also bombing. This

23 is Pristina. There's bombing there continuously. So the causality would

24 require, it seems to me, having both a positive correlation, which is what

25 this analysis of peaks would do, saying that there is bombing and

Page 10365

1 migration, or KLA attacks and killing, but there must also be the negative

2 correlation. There must also be the absence of the two that coincide

3 together, so that if one is present and the neither is not, neither is the

4 other present.

5 I think an interesting case that may bring this up is -- let me

6 scroll backward. An interesting municipality to this effect is -- yeah --

7 yeah, is back to Djakovica where we have a single air-strike and a single

8 KLA action and then a very extensive amount of migration and a fairly

9 large number of killings and then a nearly continuously pattern or a

10 fairly regular pattern, at minimum, of bombing and of KLA activity that is

11 not -- that does not correspond with subsequent killing and migration. So

12 I mean, if we contrast Djakovica and Pristina, we get at this sense of

13 things must vary in both directions, not simply in one. Does that assist?

14 JUDGE BONOMY: What I'm trying to understand at the moment is in

15 relation to a municipality where you cannot make the finding that what

16 happened is inconsistent with KLA activity --

17 THE WITNESS: Mm-hmm.

18 JUDGE BONOMY: -- what do we do about that? I mean, we can look at

19 that as a separate municipality and say: We've got no assistance there in

20 identifying support for perhaps oral evidence that we've heard about the

21 cause of people moving. I mean, we can't just look at this as a blanket

22 exercise for the whole of Kosovo when we have an indictment related to

23 individual parts of Kosovo.

24 THE WITNESS: I understand, sir. I think that the -- the purpose

25 of a statistical analysis is -- well, and, really, its limitation is this

Page 10366

1 big-picture approach. It is quite difficult to use a statistical analysis

2 of this kind to look at individual cases the way the indictment is

3 written. Instead, this may be understood as a kind of broad picture where

4 broad explanations might be tested. I'm afraid that -- I'm afraid that

5 the method does not easily permit very, very small-scale or focussed

6 findings of the kind that you've requested.

7 JUDGE BONOMY: Thank you.

8 Mr. Ackerman.


10 Q. Some time ago in answer to my last question about media, you said

11 something about newspapers and you used the term -- you used the phrase:

12 It seems highly unlikely to me. It was back a page or so ago when you

13 were answering that question about the possible effect of newspaper

14 coverage. The term "highly unlikely" is not a statistical term of art, is

15 it?

16 A. Not in this case, no.

17 Q. No. And so when you used the term "it's highly unlikely to me,"

18 you were not being a scientist but an advocate for a position?

19 A. No. I was being a scientist trying to explain something to a

20 non-scientist, sir.

21 JUDGE BONOMY: Now, Mr. Ackerman, that's about two hours you've

22 had, and we have a stream of counsel behind you. We have to be realistic

23 about how we approach this.

24 MR. ACKERMAN: Your Honour, I'm very aware that I'm basically out

25 of time. I was hoping to get through a lot more material in the time

Page 10367

1 available, and as you saw, we got bogged down in some pretty technical

2 things and --

3 JUDGE BONOMY: Well, I express the concern I had yesterday. Is

4 anyone going to be cross-examining on the basis of a competing expert

5 report, which is the sort of assistance, I think, we would particularly

6 welcome rather than the sort of common-sense question that might occur to

7 any of us but isn't terribly well-founded, perhaps, as a way of

8 challenging the exercise that's being conducted here. So you see if the

9 cross-examination was of the latter type, then it could go on forever. I

10 could think of questions to ask interminably about this from a

11 common-sense point of view, but they may be matters for argument rather

12 than matters on which Dr. Ball can comment. I'd much rather hear

13 cross-examination that focused on the expertise he's used.

14 MR. ACKERMAN: Well, Your Honour, I'm going to try to finish in

15 ten minutes. I may not be able to, because Dr. Ball may have to go into

16 further long explanations --

17 JUDGE BONOMY: I think what we should do is come back to you if we

18 have time, Mr. Ackerman. It think you've had a fair crack of the whip.

19 MR. ACKERMAN: I'm willing to do that.

20 JUDGE BONOMY: The only other thing I'm conscious of doing is

21 interrupting your question about the comparison between KLA in conflict

22 with Yugoslav forces and Yugoslav forces in conflict with KLA, and I may

23 have been premature in interrupting you on that, and if you do want to

24 explore that further, that would help, possibly.

25 MR. ACKERMAN: Well --

Page 10368

1 JUDGE BONOMY: You were suggesting that the same conclusion ought

2 to be drawn in relation to Yugoslav forces on the basis of the data that

3 were available.

4 MR. ACKERMAN: Your Honour, I think there's clarity in the report,

5 and I think anything further I would do with it would be, as you have

6 pointed out, more in the nature of argument than of a question.

7 JUDGE BONOMY: All right. Well, I think we should --

8 MR. ACKERMAN: I would be happy to yield to my colleagues, and if

9 there is some more time at the end, I actually do have some more questions

10 I would like to ask.

11 JUDGE BONOMY: Thank you.

12 Mr. Fila.

13 MR. FILA: [Interpretation] I will try and be very brief.

14 Cross-examination by Mr. Fila:

15 Q. [Interpretation] You may go ahead and laugh if I ask a layman's

16 question, because I've no idea about statistics whatsoever. What I wanted

17 to ask you about differs from the cross-examination so far. If I

18 understood properly, Mr. Ball -- but first of all, my name is Toma Fila

19 and I appear on behalf of Mr. Sainovic here.

20 In the report you compiled on the 3rd of January 2002, it is

21 entitled "Killings and Refugee Flow in Kosovo between March and June

22 1999." What I'm interested in is -- out of everything I've read is the

23 following: I saw that the report was not ordered or financed by the OTP,

24 but you nevertheless sent it to them. I'm interested in the title itself.

25 Who ascribed that title to the report?

Page 10369

1 A. I did.

2 Q. You did. I'm asking you this because when one looks at the title,

3 one is bound to conclude that this concerns movements of Kosovo

4 population; however, if you read through it - and you will agree, I

5 presume - one comes to the conclusion that it concerns itself only with

6 Albanians. Therefore, the question is this: Did you try to ignore -- or

7 perhaps to put it this way. We heard Ms. Sandra Mitchell on the 11th of

8 July before this Chamber. It was page 565 and 566, lines 20 to 25 and 1

9 to 7, if I'm not mistaken, since I don't have my glasses on. She

10 confirmed the findings of the UNHCR from Geneva of the 15th of October in

11 1999, whereby on that occasion during the NATO air campaign over 100.000

12 Serbs left Kosovo. Do you accept such data? If not, I can hand you the

13 report itself.

14 A. I'm sure that there's such a report. I have not studied it, so

15 I'm not competent to comment.

16 Q. No comments. I just wanted to see whether you are willing to

17 accept such information. Second of all, if you're familiar with the whole

18 thing, you will agree that apart from the Albanians, who were the most

19 numerous and the Serbs, there were other ethnicities whose members also

20 fled Kosovo and the war. My question, hence, is the following: Why

21 didn't you mention in the title that this concerns killings and refugee

22 flow of Albanian population in Kosovo? Otherwise, one is given the

23 impression that there were no others or that you found it unimportant for

24 them as to why they had fled Kosovo. It was not in dispute that they

25 indeed fled Kosovo. Wouldn't it be more scientific to analyse all 29

Page 10370

1 municipalities out of which some were purely Serb, some purely Albanian,

2 and others were mixed, to see what the attitude was of the population to

3 the bombs that were being dropped on their heads, but to focus also on the

4 population which was not Serb. And then to have an overall picture of the

5 impact of those bombs on the general population of Kosovo. That was my

6 question. And again, I apologise for putting it in lay terms.

7 A. No apology necessary. I think the essence of a scientific

8 analysis is to pick a sufficiently focused question so that we can address

9 it. I think it would be very interesting to conduct a study analysing the

10 patterns of migration and killing among the various other ethnicities in

11 Albania during this -- excuse me, in Kosovo during this time. We chose,

12 as you noted, to focus on the most numerous group, the Kosovar Albanians.

13 But that certainly would not exclude any study done of other groups.

14 Q. You will agree, I believe, that if you applied the three

15 hypotheses to the other peoples who also fled, you would probably come up

16 with different conclusions as to why they had fled. Probably their

17 reasons would be different to those of the Albanians.

18 A. I don't know. I haven't studied those groups.

19 Q. Very well. Another thing on the topic. You said that you went to

20 only one single border crossing, and perhaps this will be examined in more

21 detail by my colleagues. In the Milosevic trial where you testified on

22 the 28th of February 2005, there was a testimony by a doctor, a physician,

23 Dr. Aleksovski from Macedonia who received the refugees who crossed the

24 border over to Macedonia. He confirmed that the Albanian refugees stated

25 that they were fleeing from the NATO campaign, which is different from

Page 10371

1 what you have been telling us. And I hope that we will have that person

2 here as a witness but I wanted to ask you this and his transcript is in

3 evidence.

4 A. There were hundreds of thousands of Albanian refugees, perhaps

5 875.000 plus or minus a very large number. I'm not sure of the error

6 around that figure. Certainly some of them may have said some things,

7 such as what you suggest. However, my study is not concerned with what

8 people said they did, but rather with the sequence of events which may

9 have motivated their departures. This is not a study of attitude or of

10 people's reports about why they did something, but rather the timing of

11 the events in their area and then their subsequent killing or departure,

12 their being killed or departing, leaving for Albania or for Macedonia,

13 Bosnia, or for Serbia.

14 Q. Very well. I wanted to tell you something else. In your report

15 you mentioned a unilateral cease-fire that was put in force by Yugoslavia

16 on the 6th of April, 1999. You said that the cease-fire was not respected

17 by either NATO or the KLA. It was Easter, it was Orthodox Easter. Quite

18 the contrary, they stepped-up their activity. It can be read from your

19 report that the Serbs did not fire their weapons at that time. What I

20 will put to you is that they did because we refused to cease protecting

21 the Serbs because someone was opening fire at them. How would this fit in

22 with your report? If someone's opening fire at the Army of Yugoslavia, of

23 course they're not going to just sit around and wait.

24 A. Nowhere in my report do I say that the Serbs did not fire their

25 weapons. What I said is that the government of Yugoslavia announced on

Page 10372

1 television that there would be a cease-fire. Whether or not that

2 cease-fire was respected on the ground, I don't know. What I do know is

3 that immediately after the cease-fire, the number of reported killings and

4 migrations drops very low relative to its earlier period. That's the

5 observation that we make when we say that there is a strong coincidence

6 between the session of stated hostilities by the Yugoslav government and

7 killing and migration. It's whether or not the hostilities actually

8 ceased; that's beyond this report.

9 Q. This is not in the electronic system, therefore I cannot put this

10 before you in that way. But unless the OTP objects, I can show you this

11 and it shows the flow of refugees. You can even keep it if you wish to.

12 You can use it as evidence later on. This is one of OTP's documents.

13 This would actually conclude my cross-examination, since I wouldn't be

14 allowed to use this, I believe.

15 But just one more question of a more general nature. Of course

16 you need not know necessarily about it and I'll try and explain --

17 JUDGE BONOMY: Hold on, Mr. Fila, what's going on here? Why are

18 you handing this over?

19 MR. FILA: [Interpretation] Since I cannot use it, I just wanted to

20 hand it over. They can keep it and the witness can keep it. One can see

21 the migration patterns. It wasn't in the system, therefore I cannot use

22 it but the OTP may well use it if they want to.

23 Q. Last year there was a committee of the National Assembly formed

24 and some people were sent down to Kosovo. Together with Albanian

25 representatives they compiled a report, an overview, of all those missing

Page 10373

1 and killed. The figure is about 9.700, out of which some 4.000 are

2 Albanians and some 3.000 Serbs as well as others. I have their findings

3 here, although untranslated. I intended to use it later. In total it was

4 less than 10.000 of all groups. They counted every single head. Later on

5 when I will be allowed to use that, if I were able to show it to you now,

6 would it change your views that the total figure is not the one you

7 estimated? And there were only slightly over 4.000 Albanians killed or

8 missing. Even Ms. Natasa Kandic who cooperates with the Prosecutor's

9 office agrees with the figures. Would that have an impact on your

10 findings or thoughts on that? Of course, I'm not trying to enter into any

11 polemics about whether it should or not, but I'm simply asking whether it

12 would.

13 A. Let's separate the question. There's really two different

14 questions here. The first question is how many people -- how many Kosovar

15 Albanians were killed. And my experience over many years is that no many

16 how many times you try to --

17 JUDGE BONOMY: You've told us this yesterday. That's not the

18 question, Dr. Ball. The question is: Assume for the moment that slightly

19 over 4.000 Albanians were killed and that figure was accurate --


21 JUDGE BONOMY: -- let's assume that. What difference does that

22 make to your conclusions? That's Mr. Fila's question.

23 THE WITNESS: If those people are distributed in the same way that

24 we have shown in, for example, figure 1 of appendix 3 of my January 2002

25 report it will make no impact at all on our analysis. The analysis will

Page 10374

1 be the same. Our analysis does not rest on the total magnitude of deaths;

2 it rests on the distribution of those deaths across time and space.

3 MR. FILA: [Interpretation]

4 Q. And the figure wouldn't change your opinion? You dealt with the

5 topic extensively, even during the Milosevic trial, but I failed to

6 understand and I fail to understand now.

7 To move on to a third thing, to a third topic.

8 MR. FILA: [Interpretation] Could we please bring it up in e-court.

9 It is table 1, page 3 of the English.

10 Q. Your report of the 28th of January, 2007, the 29 municipalities.

11 MR. FILA: [Interpretation] I still don't have it. It is Exhibit

12 P2678. This is it.

13 Q. I tried to do some math with this, and if I was incorrect, I

14 apologise again. If we follow the columns, can you please count the

15 number of coincides to establish the number of municipalities in which

16 there was a coincidence of KLA and NATO actions with killing and

17 migration. You can use the electronic pen and I will give you my figure

18 so as to be able to compare that to yours. KLA activity coincided in nine

19 municipalities, whereas NATO activity in six.

20 A. I'll be happy to do this; however, I'm afraid that we're going to

21 go into the same area of discussion that Mr. Ackerman and I covered at

22 some length about this method. No?

23 Q. [In English] No.

24 A. Okay. So here we go. So you want me to --

25 Q. [Interpretation] Mr. Ball, in my country, we do not tend to

Page 10375

1 repeat. You only repeat if there was a lot of noise in the communication

2 channel, but I do not think we have that and I do not intend to follow

3 Mr. Ackerman's line of questioning. And since we're not deaf, either of

4 us, I guess we may pursue.

5 Therefore, in nine municipalities we had coincidences of NATO

6 activity -- sorry, KLA activity and in six municipalities NATO activity.

7 And in Djakovica, Pec, and Pristina we have both NATO and KLA activity.

8 Am I correct in thinking that in 12 municipalities the murders coincided

9 with KLA and/or NATO activity?

10 A. No.

11 Q. That's what you put in.

12 A. No, I'm sorry, sir. The -- what coincides is the single greatest

13 day, the peak of killing and migration. That is quite different than

14 saying that the entirety coincides. So again, as I said to Mr. Ackerman,

15 in order for us to make the conclusion that there is, as you just said,

16 that the murders coincide more broadly, we would have to find both the

17 positive and the negative correlation. We would have to find an absence

18 of KLA --

19 Q. I'm not asking you about that. I understand that. That's why I

20 said we shouldn't repeat twice. I'm counting your coincidences or the

21 number of the word and I'm trying to put it in figures so as to reflect

22 the table. When we are discussing migration - and you can count to see

23 whether I'm telling the truth - I'm telling you that in ten municipalities

24 migration coincided with KLA activity; that's what you put into the table.

25 And in nine municipalities, migration coincided with NATO bombardment.

Page 10376

1 Since in the municipalities of Orahovac, Pec, and Pristina, as well as

2 Prizren, migration coincided with both KLA and NATO activity, may we

3 conclude therefore that in 15 municipalities migration coincided with

4 either KLA or NATO activity or both? This is simply what I'm reading out

5 of the table. A sample yes or no would do.

6 A. I simply have to correct the identification of what you're

7 claiming coincides. What coincides again is not activity in a general

8 sense; what coincides is the peak of activity. Okay. So there's maybe

9 activity -- there may be killing and migration at other times -- but this

10 is quite an important distinction, sir, so I want to make sure that it's

11 included in the record.

12 Q. Very well. But the figures I stated are correct; isn't that so?

13 A. Without checking them, they seem roughly correct, yes. Go ahead.

14 Q. Do you agree with me that out of the total of 29 municipalities,

15 the word "coincidence," whether it reflects the peaks or totality,

16 killings and migration coincide to a large degree with - if I may have a

17 moment - the figure is quite a high one, having in mind the total of 29

18 municipalities, and we come up with the number of 15 more or less, 15 out

19 of 29. If we know - and I believe we do - that the municipalities of

20 Zvecan, Leposavic, Strpce and Novo Brdo are almost purely Serb ethnically

21 speaking and you said that the four Albanians -- the total of Albanians

22 that were there left and if we say that these are purely ethnically

23 Serbian places as well as the municipality of Gora, where there are

24 Albanians who speak Serbian, then the importance or the weight of the

25 conclusion is even greater because those municipalities need to be

Page 10377

1 extracted out of the group.

2 And you discussed this with Mr. Ackerman, and I just wanted to add

3 something and that is in the municipalities relevant to the indictment

4 there are two or more coincidences. And for Pristina there is a

5 coincidence in all four categories, which means that killings and

6 migration coincided with the activity of both KLA and NATO, and you

7 confirmed that in your text at page 4 and of the second paragraph in the

8 English. Is that correct? Do you stand by what you put in that text?

9 A. I stand by what I put in the text; however, I am concerned that

10 you are over-interpreting the coincidence. The coincidence must be

11 understood only as a coincidence because if we were --

12 Q. That was my understanding of it, yes, that's what you said to

13 Mr. Ackerman. Yes. But the coincidence is there --

14 JUDGE BONOMY: The importance of the coincidences is that they

15 remain in a minority of municipalities. Is that correct?

16 THE WITNESS: Not necessarily, sir. Even -- as I said before,

17 even if we had coincidences in every single one of these cells, the causal

18 argument would still be undermined if there were bombings when there was

19 no migration, if there were KLA attacks when there was no killing. You

20 see, the problem with this coincidence argument is that it only

21 observes --

22 JUDGE BONOMY: Where I'm struggling - I'm obviously not

23 understanding something here - but I understood that where there was a

24 minority of coincidence -- of municipalities where there was a

25 coincidence, that you could say that the activity you were investigating,

Page 10378

1 for example, bombing, was not consistent with the idea that that caused

2 the movement or the killings -- the movement of people. Now, are you

3 saying that that's a wrong understanding of your report, that the minority

4 is not significant in coming to that conclusion, that negative conclusion?

5 THE WITNESS: The minority would lead to that conclusion, but the

6 converse is not true --

7 JUDGE BONOMY: I'm not suggesting it is.


9 JUDGE BONOMY: But what Mr. Fila now is identifying - and all the

10 names have not come up on the transcript, Mr. Fila, so perhaps you can

11 tell us which of the municipalities in brackets are you saying are purely

12 or principally Serb populated? I think you said Gora was one --

13 MR. FILA: [Interpretation] No, Gora is a municipality where the

14 Gorani live. These are people who speak the Serbian language --

15 JUDGE BONOMY: [Previous translation continues]...

16 MR. FILA: [Interpretation] -- but they are Muslims. They are not

17 Albanians, they are Muslims.

18 JUDGE BONOMY: [Previous translation continues]... was one?

19 MR. FILA: [Interpretation] Zvecan, Leposavic, Strpce, Novo Brdo.

20 Let me do it again. Zvecan, Leposavic, Strpce, Novo Brdo, and Gora. Is

21 it wrong again? Oh, yes, and Zubin Potok.

22 JUDGE BONOMY: You see, there are six municipalities where it's

23 suggested that there's no significant Albanian population to be affected

24 by this. Now, is that of some importance in analysing your statistics?

25 THE WITNESS: I think that this is -- I would suggest that this --

Page 10379

1 while this table provides rich fodder for debate about specific places, I

2 think that the statistical analysis should be -- that the statistical

3 finding should be made on the basis of figures 10 and 11 in the original

4 report, where a much more thorough statistical test of these relationships

5 is made, Your Honour. That I really -- I must say that we've -- it is

6 unfortunate that we've focused so much on this because the real analysis

7 is in those two figures and that -- those are the two figures that show a

8 proper statistical test of the relationships among four simultaneous

9 variables. It's quite complex to look at a table like this and choose a

10 single day form each -- or single two-day period in fact from each of

11 these 29 municipalities and treat that single two-day period as though it

12 is determinant of the history of the conflict in that region. So I would

13 urge us to look at a more consistent and a more thorough test.

14 JUDGE BONOMY: Even though these -- when you look at table 1 --

15 THE WITNESS: Mm-hmm.

16 JUDGE BONOMY: -- and you look at these municipalities mentioned,

17 they are the ones where all along the row no prior -- no inconclusive -- I

18 think no prior in fact is the finding --

19 THE WITNESS: Mm-hmm. There are others, too, as well. Kosovo

20 Polje, for example.

21 JUDGE BONOMY: If there aren't people there who are likely to be

22 moved by Serb forces, does that not tend to have an impact on the overall

23 significance of looking at 29 municipalities --

24 THE WITNESS: Well, it --

25 JUDGE BONOMY: -- when six of them are populated by people to whom

Page 10380

1 your third hypothesis would not apply?

2 THE WITNESS: No, but an inverse-thought hypothesis might apply

3 which is that if we were to assume - which my study does not - if we were

4 to assume that the cause of killing and migration were the Yugoslav

5 authorities, and if we were to assume that Yugoslav authorities were not

6 killing or forcing Serb populations to move, then we would in fact find

7 precisely the patterns in the Serb-dominant municipalities that we find

8 here. So although it had not occurred to us and it is outside the realm

9 of our report, the observations that have been made by counsel confirm or

10 would tend to confirm the claim that --

11 MR. FILA: [Interpretation]

12 Q. I'm sorry for interrupting. First of all, I wasn't the person who

13 chose this table; it's your table. I'm not the one who made it.

14 Secondly, could you please do what it was that I asked you to do. First

15 answer a question that I have for you.

16 Kosovo is a relatively small territory. You come from America.

17 It is so small for you. It's small from the point of view of my country

18 let alone yours. So let's assume that these municipalities are not

19 islands. It's not Indonesia is it? These are villages that are nearby to

20 each other. So isn't it logical to assume that panic and fear spread and

21 that migrants spread this and that it has a hallow effect, you know, it's

22 when libel starts, you don't know who even started it. So that is how it

23 spreads. And if we were to look at this spiral, now how should I put

24 this, rumours spread and people move out as a matter of inertia especially

25 from mixed communities. Albanians see that Serbs are fleeing, Serbs see

Page 10381

1 that Albanians are fleeing. All of them together see that the Gorani are

2 fleeing.

3 A. I'm sorry, what's the question?

4 Q. This panic spreads among these municipalities. Now, did you take

5 into account that effect? You know, the bombing is an ongoing process.

6 The document that the Prosecutor did not give you, I am so sorry about

7 that, because there is this review of when who left but they didn't want

8 to give you the document so there. You see people are leaving all the

9 time at different points at time. You only looked at Albanians; we looked

10 at others as well. So in the question of intrigue, this is called the

11 hallow effect, but I mean here it is panic that spreads especially in such

12 a community like Kosovo is. Now what am I trying to say. It is not that

13 the three hypotheses are the only causes, the one that you mentioned, it's

14 panic, too, fear.

15 A. If the question is, did we take into account panic, I repeat my

16 earlier answer that this is not a study of psychology of war or rumour.

17 We took into account simply the movements of people and the actions of the

18 KLA and NATO and killings.

19 Q. All right. Now, in order to check my conclusions, could you use

20 your own pen on this graph that is right in front of you and mark the

21 numbers that I mentioned to you so that this could become an exhibit as

22 well.

23 JUDGE BONOMY: I don't think that's useful. We've got the

24 figures, Mr. Fila.

25 MR. FILA: [Interpretation] All right. Fine. No problem. Right.

Page 10382

1 Thank you. That would be all, Your Honour.

2 JUDGE BONOMY: Thank you.

3 Do you have questions, Mr. Cepic?

4 MR. CEPIC: No, Your Honour.

5 JUDGE BONOMY: Thank you.

6 MR. CEPIC: Thank you.

7 JUDGE BONOMY: Mr. Lukic, can you guide us on the timing?

8 MR. LUKIC: I hope it would be very short.

9 JUDGE BONOMY: Well, I think what we'll do is come back at 10

10 past -- in fact, we'll come back at quarter past and you can start at that

11 time.

12 MR. LUKIC: Thank you, Your Honour.

13 JUDGE BONOMY: We'll have to have a break now, Dr. Ball. The

14 usher will take you again and you will leave the courtroom and we'll see

15 you at quarter past 11.00.

16 THE WITNESS: Thank you, Your Honour.

17 [The witness stands down]

18 --- Recess taken at 10.41 a.m.

19 --- On resuming at 11.19 a.m.

20 [The witness takes the stand]

21 JUDGE BONOMY: Mr. Lukic.

22 MR. LUKIC: Thank you, Your Honour.

23 Cross-examination by Mr. Lukic:

24 Q. [Interpretation] Good morning, Dr. Ball. I am Branko Lukic, and I

25 have only a few questions for you. In your introductory section in

Page 10383

1 paragraph 2, you say: "Nothing contained in this publication should be

2 considered legal advice for concrete cases. This publication like all

3 formulations and positions contained herein are only meant for information

4 purposes."

5 Nevertheless, we believe that in your findings you do give a

6 opinion and you draw some conclusions and you even give legal opinions,

7 albeit unintentionally perhaps. For example, you say that in the

8 questions that you dealt with there were killings involved. I would like

9 to ask you the following. In your findings, did you take into -- a single

10 category all fatalities, civilian deaths, killings of KLA fighters,

11 Albanians who were killed by the KLA, as well as people who died of

12 natural causes?

13 A. We included in our analysis only those deaths that occurred by

14 violence, that occurred during the period of the study. We do not know

15 who killed the people, and in fact the inferences about who killed the

16 people are the basis for the hypotheses.

17 Q. Thank you. Is it correct that all the information you published

18 was published in the following way. You protected the privacy of the

19 persons involved and that all the data you published actually do not

20 contain any personal identification of the victims or the witnesses of

21 these events?

22 A. That's correct.

23 Q. So the Defence has no possibility of checking this information.

24 Is that right?

25 MR. STAMP: That's a matter which is outside of the competence of

Page 10384

1 the witness. The Prosecution did disclose to the Defence over a period of

2 years the supporting material that was used or the source material that

3 was used.

4 JUDGE BONOMY: Including the identity of all the victims?

5 MR. STAMP: No, no, Your Honour.

6 JUDGE BONOMY: Well, that's the question he's asking. Which is a

7 perfectly legitimate question for him to ask --

8 MR. STAMP: Of this witness -- very well.

9 JUDGE BONOMY: If the witness can't answer it, he'll tell us he

10 can't answer it. He's a perfectly sensible person when it comes to

11 knowing his limitations.

12 Mr. Lukic.


14 Q. Mr. Ball, can you answer this question, please?

15 A. I'm limited by not knowing precisely what the OTP has disclosed to

16 the Defence. If the OTP disclosed to the Defence the original statements

17 and interviews that were conducted by the groups that provided data to us

18 for our analysis, then you should have all the victims' names unless those

19 names have been redacted.

20 MR. STAMP: I'm sorry about that. I was incorrect. The

21 disclosure did include the names of the victims.

22 THE WITNESS: So the answer is: Yes, you do have all those names.

23 JUDGE BONOMY: Thank you.

24 MR. LUKIC: [Interpretation]

25 Q. Mr. Ball, in terms of evidence supporting your report, we have

Page 10385

1 officially been provided with one document, P3193. That is one of the 690

2 reports you collected at the Morina border crossing and we did not receive

3 anything else officially from the OTP?

4 JUDGE BONOMY: I think, Mr. Lukic, that must be wrong. You're

5 distinguishing between what's an exhibit in the case, I think, and the

6 disclosure. The Prosecution position is that they have disclosed it all

7 to you but it's not exhibited. Now, are you disputing that?

8 MR. LUKIC: Actually, I'm not aware of -- we had the discussions

9 during the cross-examination of Mr. Abrahams and Ms. Mitchell, and I'm not

10 aware that we received all those statements disclosed. So -- but I

11 wouldn't bother this witness with this. I'll move on.

12 JUDGE BONOMY: It is something you obviously can take up with the

13 OTP and raise with us also if, in fact, our understanding is inaccurate.

14 I would just like to ask one question, though.

15 Mr. Lukic touched initially there on something which may be of

16 some importance. The KLA are said on occasions to have killed Albanians

17 who were perhaps collaborators or seen to have official positions or

18 whatever. In your hypothesis in relating to activity of the KLA, you

19 confine it to the KLA ordering people to leave or because people fled

20 fighting between KLA and Yugoslav forces, which seems to leave out of

21 account discrete acts of murder that may have been committed against

22 individuals.

23 THE WITNESS: That's correct; however, you are also correct that

24 if the KLA were to have conducted executions in the style you suggest,

25 they would also have been included in that hypothesis. The same logic

Page 10386

1 would be obtained. Any KLA killing directly or indirectly would have been

2 cover by that hypothesis.

3 JUDGE BONOMY: Thank you.

4 Mr. Lukic.

5 THE WITNESS: Your Honour.


7 THE WITNESS: Mr. Lukic mentioned a number that I think may have

8 caused -- flags for me that there may be a little bit of confusion I can

9 clear up. He said he has only one of 690 pages. I think the 690 pages

10 refer to the Albanian border documents that were disclosed to the

11 Milosevic court - I don't know if they have been disclosed here - but they

12 are distinct from the Human Rights Watch and OSCE materials which were the

13 subject of the exchange with Mr. Stamp.

14 JUDGE BONOMY: I think you may have -- oh, yes, it is. It's P1393

15 and it is one of the documents taken from the border post, yeah.


17 JUDGE BONOMY: Mr. Lukic.

18 MR. LUKIC: Thank you.

19 Q. [Interpretation] Today on page 37, line 12, in response to a

20 question you said that you did not care about what people say. It had to

21 do with what Albanians crossing the border said. However, was that not

22 precisely your own source of information, what people were saying as they

23 were crossing the border, in relation to when it was that they had left

24 their homes and when they were crossing the border?

25 A. Yes. I am interested in when people said they left their homes.

Page 10387

1 I'm interested in when they crossed the border. I am not interested - and

2 we did not ask - why people left their homes, nor what their fears were or

3 any other of the psychological discussions that surrounded the

4 conversation to which you allude.

5 Q. We'll get to that soon. Do you know, since foreigners did not

6 speak Albanian, that the interpreters who were interpreting for the

7 foreigners there were primarily from the Council for the Protection of

8 Human Rights and Freedoms, who were Albanians themselves?

9 A. The Council for the Protection of Human Rights and Freedoms, I

10 believe, conducted some translations, but certainly not for all four of

11 the projects. I believe they did for two of the four original projects,

12 and it is not my understanding they had anything to do with the three

13 recent projects we have just used in the January 2007 report. So perhaps

14 two of the seven data sources used for the analysis of killings were

15 related to the Council for Human Rights and Freedoms.

16 Q. Very well. Thank you. Do you know that --

17 JUDGE BONOMY: Which two, I think we should be clear about.

18 THE WITNESS: I believe I believe for the ABA/CEELI data, and I

19 think there may have been - and here I'm remembering perhaps an error -

20 that there were some translators for the OSCE that may have been from the

21 Council For the Defence of Human Rights and Freedoms. That can be

22 confirmed with OSCE.

23 JUDGE BONOMY: Thank you.

24 Mr. Lukic.

25 MR. LUKIC: [Interpretation]

Page 10388

1 Q. Do you know that members of this council were at the same time

2 members of the KLA, often very highly positioned in the ranks of the KLA?

3 A. I have no such knowledge.

4 Q. Thank you. Several times today you pointed out that you were not

5 interested in psychology but that you were dealing with statistics;

6 however, I would like to ask you precisely about the psychological part of

7 your findings. Actually, you took two days at a time, that is to say, two

8 days prior to the peaks of killings or of departures that people feared

9 NATO bombings only two days prior to -- or rather, after the bombing

10 itself.

11 So could you kindly explain to the honourable Trial Chamber and to

12 me on the basis of what do you reach these findings? How come you took

13 only these two-day periods?

14 A. First, we used two-day periods throughout the analysis because the

15 amount of data available to us was inadequate to do day-by-day analysis.

16 So we cumulated each pair of days together in order to do the analysis.

17 There simply wasn't enough information to do consistent analysis in single

18 days.

19 Second, I think what you refer to in your question is that in the

20 analysis of peaks and troughs, particularly peaks, which we have been over

21 at some length this morning and is captured most clearly in the table

22 we've been debating, which I don't have at hand but which is the one that

23 discusses coincidences no prior or inconclusiveness, that to define a peak

24 of killing or migration as coinciding with KLA activity or bombing, we

25 used four days prior to the bombing or the -- or the KLA activity in order

Page 10389

1 to determine whether or not the event were to coincide with the peak. So

2 it's not a two-day period, it's a four-day period. We chose four days

3 because that seemed to us a reasonable period of time for people to react

4 to a specific event. If the event were longer, it would be classified as

5 inconclusive, and in fact we provide in the new report that the January 27

6 report -- January 28th, sorry, 2007 report, we provide a fairly extensive

7 analysis of these inconclusives. And inconclusive in this context is a

8 municipality in which the air-strike or the KLA activity occurred more

9 than four days prior to the peak of killing or migration.

10 So if you prefer a longer period, if you wish to make an argument

11 about a longer period, then the graphs -- the 29 graphs in that report and

12 the analysis of inconclusives can provide you with material to make such

13 an argument.

14 Q. Can we conclude that in documentation or literature from

15 psychiatry or psychology you did not find a base for such calculations.

16 Is that right?

17 A. That is correct.

18 Q. Thank you.

19 On the 20th and 21st of March, 1999, there were convoys of

20 refugees, and we heard about them from an Albanian witness who testified

21 here, that was prior to the bombing. How would they fit into your

22 analyses before the fighting and during the stay of the verifiers in

23 Kosovo? Can you somehow fit that into your own analysis?

24 A. Our analysis starts later than that period, precisely because

25 that's before the Albanian border guards began systematically recording

Page 10390

1 people crossing the border, neither did OSCE or UNHCR have systematic and

2 rigorous counting going on at that time. However, let's keep in mind that

3 our argument is that it is unlikely that the -- in fact, we've rejected

4 the hypothesis that the bombing is the cause of refugee flow. So the fact

5 that there were -- that there was extensive refugee flow before the

6 bombing is perfectly consistent with our broader claim. If there were

7 convoys of refugees before the bombing, clearly the bombing did not cause

8 those convoys of refugees to be on the move. That's similar to what we

9 see throughout the region.

10 Q. But can we conclude that that is not in accordance with your own

11 conclusion, that people were leaving only because their particular region

12 was under attack? Is this not along the lines of what my learned friend

13 Mr. Fila said, that fear cannot be confined so narrowly? Fear can exist

14 without direct attacks and that one has to have a broader view than you

15 had in your own tables.

16 A. Well, this is a somewhat complicated statement. Let me see if I

17 can unpack it. The claim is that the convoy of people that's moving

18 before the bombing are moving from fear. I have no idea why they're

19 moving. This is not an analysis of psychology. What is clear, and what I

20 think we can all agree on, is that these people are certainly not moving

21 because their region has been bombed. That's the point of our analysis.

22 If the bombing has not started, then clearly the bombing has not motivated

23 them to start. If we want to speculate about the psychological state of

24 these people, then -- well, I suppose that speculation is possible but I

25 do not engage in it.

Page 10391

1 Q. Can we agree that at that time there weren't any attacks by Serb

2 forces either because verifiers were in Kosovo and they never reported

3 about any such attacks?

4 A. I don't know whether that's true or not.

5 JUDGE BONOMY: That's beyond the competence of the witness, that

6 question, Mr. Lukic.

7 MR. LUKIC: Thank you, Your Honour. I'll move on.

8 Q. [Interpretation] At page 15 in the English, paragraph 2, you

9 say: "The statistical analysis of correlation cannot prove that the

10 Yugoslav forces were the external factor responsible for the patterns

11 observed." It is item 5.4, Yugoslav forces, the first paragraph. Do you

12 abide by that?

13 A. Page 15, paragraph 2 --

14 Q. [In English] 5.4, Yugoslav forces, first paragraph.

15 A. Mm-hmm. That's correct. We are clear throughout the report that

16 we do not make affirmative proofs in this report or in any of the various

17 addenda.

18 Q. [Interpretation] Right below you state: "However, the findings of

19 this study are in keeping with the hypothesis that the activity of the

20 Yugoslav forces was the cause of murders and migration of refugees."

21 My question, hence, is this: If you weren't able to conclude that

22 based on statistical analysis of correlation, based on what did you make

23 your conclusions as a statistician?

24 A. If you follow in the next paragraph after the one you've read, we

25 observe a circumstantial link between the cease-fire of the 6th and 7th of

Page 10392

1 April and the decline of killings and migration. That is the basis for

2 our observation that the evidence is consistent with the hypothesis that

3 Yugoslav forces were responsible for killing and migration. This is not

4 proof, but it is an interesting circumstantial link.

5 Q. In the last sentence of the very -- paragraph you say: "Further

6 connection or links can be established if it can be proven that the

7 movement of Yugoslav troops followed the same pattern as the cases of

8 killings and refugee flow; however, such analysis would lie outside the

9 scope of the study."

10 Therefore, you did not investigate or study that any further since

11 it went out of the scope of your study?

12 A. That's correct.

13 Q. Thank you. At page 45, paragraph 5 in the English, item 3, 3.3,

14 aggregation of cross-checked tables to explain, or rather, aggregation of

15 the cross-classification tables to account for sparseness.

16 A. Mm-hmm.

17 Q. Then you say: "The next goal for this analysis is the estimation

18 of the number of killings for each of 192 space time points representing

19 48 two-day time-periods and four geographical regions."

20 Who determined this to be a goal, you or the OTP?

21 A. My colleagues and I did, my co-authors and I did this.

22 Q. Thank you. At page 47, second paragraph, figure 7.

23 "Counts of zero cells for the four-way tables."

24 You say: "Collapsing to the 24 four-day periods over the four

25 regions yields 96 cross-classification tables. Of these, 48 or 50 per

Page 10393

1 cent contain more than ten zeros. Collapsing to the 16 six-day periods

2 over the four regions yields 64 cross-classification tables. Of these, 26

3 or 41 per cent contain more than ten zeros. Collapsing further will

4 impede the analysis desired."

5 However, we heard that you collapsed that to two days. Did that

6 impede the analysis desired?

7 A. When we are doing the analysis over the entire region, we do it by

8 two-day periods. The migration analysis -- the stuff we're talking about

9 here in this appendix that you've just referred to exclusively treats the

10 killing analysis. The migration analysis, first to set aside, was done

11 exclusively in two-day periods by municipality that were then aggregated

12 to regions. This analysis on killings was done in a slightly more complex

13 way because the data is -- there's less of it, it's thinner. So we did

14 the analysis for the entire period -- excuse me, for the entire region by

15 two-day period and then we -- if you'll follow into the next paragraph, we

16 found that we could do, instead of four-way cross-classification tables,

17 we could do three-way cross-classification tables and then do each

18 combination, each possible combination.

19 If we have four systems, you can create four possible combinations

20 of three systems. So we have four data sets, now we have seven, but then

21 we had four, and we could create them by saying we'll take A, B, and C or

22 A, B, and D or any possible combination you can draw, four possible

23 combinations of three sets from four possible sets. And what we did is

24 find that by doing so, as we explain in the subsequent narrative, we could

25 make much more finally grained estimates than we could by approaching it

Page 10394

1 at the top level using four-way classification systems. So this is an

2 explanation of why we chose to use the three-at-a-time method.

3 Q. Therefore, your explanation is that this did not impede the

4 analysis desired or it did?

5 A. No, that's right --

6 Q. [In English] My mistake.

7 A. It did not impede. I'm sorry.

8 Q. [Interpretation] At page 57, paragraph 4, you state: "The

9 regression models using migrations as the dependent variable, however,

10 yield a different interpretation. In this case there appears to be an

11 association between KLA activities and migrations, specifically the

12 association between KLA battles within the previous time-period and

13 migrations in the current time-period appears to be significant. The R

14 square for both of these models are high, further confirming an

15 association."

16 Can we say that the KLA, or rather, their activities had any

17 impact on migration?

18 A. What we're finding here is that there's some association. What we

19 continue -- and if you continue looking through on to page 59 where we

20 make our final conclusions with that, we are observing that there are some

21 statistical associations but that those statistical associations are

22 insufficient to change the underlying patterns of the migration and

23 killing. Figures 20 and 21 give you the detail on that. In particular -

24 and I appreciate you raising this, because this is, in fact, the core of

25 our causal argument about how this -- how these relationships work.

Page 10395

1 The statistical analysis finds that there is some relationship

2 between KLA patterns and NATO patterns; however, once one controls for

3 those relationships, once we remove the statistical effect of the

4 relationships between the KLA and NATO activity and killings and

5 migration, we find that the patterns of killing and migration remain

6 substantively unchanged and that the -- the lack of change is shown quite

7 clearly in the eight graphs shown in figures 21 and -- 20 and 21, pages 59

8 and 60. If I may explain one, perhaps, they're a bit confusing. If --

9 would it help the Bench if I explained one?

10 JUDGE BONOMY: If you think so.


12 So if we consider figure 20 on page 59. Figure 20 is titled:

13 "Estimated Total Killings and Residuals by Region Over Time," and these

14 are the graphs that result from the statistical coefficients shown in

15 figure 19 in the second column. The effect of this graph, to start out

16 is -- well, the meaning of this graph is, first, that the dark black

17 lines in each of the four graphs show the estimated total killings for

18 that region, the total number of killings that we believe to have

19 occurred, given our four sources and given the adjustment that we've made

20 for the under-reporting bias which we believe to exist.

21 Underneath the dark grey line -- the dark black line there is a

22 light grey line, and the light grey line shows the same pattern with the

23 statistical effect of the NATO and KLA activity removed from the original

24 line. Okay. We have extracted the statistical relationship of the KLA

25 and NATO on total killings, and we have then re-shown, re-plotted total

Page 10396

1 killings. Okay. The statistical term for this is the residual. The

2 underlying light grey line is the residual, that which remains after the

3 effect of NATO and KLA activity has been removed.

4 The point of -- the interpretation of these graphs and the point

5 that I think is so useful from these is that it allows us simultaneously

6 to consider both the effect of NATO and the KLA all at the same time over

7 the entire period, not on a single day, not on a single peak, but over the

8 entire period. And what we see is that removing the effect of NATO and

9 the KLA has no substantive impact on the patterns that we see, and that's

10 the fundamental conclusion.

11 So again, returning to figure 20 for the north region, the upper

12 left graph, we see that the light grey line tracks the dark black line in

13 every substantive particular. It rises during the early phase, the first

14 phase, drops dramatically on the night of the 6/7 April cease-fire, stays

15 at a low point, and then rises again. You can see that it is a very close

16 transition. There are throughout these eight graphs a few anomalies.

17 Some of them are discussed in the text. Some of them, I would be happy to

18 discuss at more length now, if you prefer, but these two graphs in

19 particular are the heart of our argument that we should reject the claim

20 that either NATO or the KLA are plausible causes of killing and migration.


22 Q. Thank you, Doctor. Therefore, the essence or the gist of your

23 analysis is precisely this part and the conclusion that the KLA did not

24 have any impact on migration. You probably are unfamiliar with some

25 testimony by Albanians before this Chamber inter alia the chief of the

Page 10397

1 general -- the main staff of the KLA, who stated something to the

2 contrary. He stated that the KLA ordered civilians to withdraw alongside

3 troops.

4 Had you had those testimonies before you and had you been able to

5 see what was the extent of the KLA's orders for civilians to move together

6 with them, would that have had any impact on the results of your findings?

7 A. Well, let's refer to figure 21. First off, I would have -- I

8 would make it a practice to ignore any claim by a politician about what

9 people do or don't do in response to his orders. But I'm interested in

10 the patterns, I'm interested in what people actually did rather than what

11 people were told to do --

12 Q. If I may interrupt. We had witnesses which obeyed such orders.

13 In combination with everything aforementioned, would that have had an

14 impact on your standpoint?

15 A. I'm sure there were some people who did obey those orders. The

16 question is: Did the hundreds and hundreds of thousands of people obey

17 those orders or some other pattern? That's my question. That's what is

18 interesting me and what is the core of this report. No doubt, one can

19 find someone who will say they followed such and such an order, but if we

20 observe the graphs on figure 21, it seems to me that where we look at

21 refugee flow we see that removing the effect of the KLA and NATO does not

22 leave as clean a line as figure 20, and we discussed this at some length

23 in the text, but nonetheless, in the substantive conclusions, particularly

24 for the major peaks in each of the regions where we make findings, the

25 patterns are similar.

Page 10398

1 To review, our argument is that in the south and west there is a

2 major peak in the early period which declines dramatically at the 6/7

3 April point, whereas in the north and east the major peak comes in the

4 middle period between the 7th and 24th of April. Here we find that the

5 residual patterns track closely the original lines. There is an exception

6 in the east which is dominated by Pristina, and we have some discussion of

7 that in the text. That is the place where there seems to be some question

8 about this relationship, but the other seven graphs are all -- all four

9 killing graphs and the other three migration graphs are extremely close.

10 JUDGE BONOMY: Did you as part of that answer refer to the middle

11 period?

12 THE WITNESS: Yes, sir. The middle period is of interest to us

13 particularly in the north and east where we see more action in the middle

14 period in both -- in both figures 20 and 21.

15 JUDGE BONOMY: Which you described as the 27th of April; is that

16 right?

17 THE WITNESS: From the 6th to the 24th.

18 JUDGE BONOMY: 6th to the --

19 THE WITNESS: 24th, sir.

20 JUDGE BONOMY: Yeah. The transcript hasn't picked this up, but

21 that clarifies it. Thank you.

22 THE WITNESS: Yeah. All of the graphs have a pattern of dark

23 banding and then light banding, and the intention of those bands is to

24 distinguish the three phases.

25 JUDGE BONOMY: I follow that. It's just I'm trying to get your

Page 10399

1 answer expressed clearly in the transcript. Thank you.

2 MR. LUKIC: [Interpretation]

3 Q. We as attorneys at law love to talk, but I think you far exceed

4 us. All I wanted to hear from you was this. The data you received was

5 from your interviews with people. Had you had data derived from such

6 interviews about these issues, would that have had any impact on your

7 report?

8 A. Well, some of our data is from interviews but not all of it. Some

9 of our data is from exhumation records, other data is from administrative

10 records. It's not all interview data, so that's an important

11 clarification. I'm not sure what issues you're referring to that would

12 have changed it. I've already answered your question that I'm not

13 interested in people's personal attribution about why they did something

14 or am I interested in a given politician's claim that he has such and such

15 influence.

16 Q. We are talking about the reason why people fled. I suppose that

17 you weren't able to establish that based on exhumation data. We're

18 talking about the data you collected at the border post or the border

19 crossing. Had you had such data at the time, would that have had any

20 impact on the findings of your report, yes or no?

21 A. At the border post we asked people only where they came from, when

22 they left, when they crossed the border. We did not ask them why they

23 crossed.

24 Q. Thank you. That explains it all.

25 I received this document from my learned friend Mr. Fila. He says

Page 10400

1 he discussed it with our learned friend Mr. Stamp, who did not object to

2 showing this document to Dr. Ball.

3 MR. LUKIC: [Interpretation] With Court's permission, if you find

4 that necessary, we can do that in order to clarify whether there were any

5 interruptions between the 6th of April -- on the 6th of April concerning

6 migration flow of refugees. If you deem it inappropriate, we will

7 certainly accept that.

8 JUDGE BONOMY: Well, it's impossible to tell because we don't know

9 what your document says, but since there's no objection, perhaps you could

10 have it displayed on the ELMO and ask whatever question you wish to ask.

11 MR. LUKIC: [Interpretation] Well, I wasn't the one who spoke to

12 Mr. Stamp, so I personally am not acquainted with his position.

13 MR. STAMP: [Previous translation continues]... the total of the

14 document is that the witness is being shown.

15 JUDGE BONOMY: I'm sorry, Mr. Stamp.

16 MR. STAMP: I thought counsel would explain what the document is

17 that the witness is being shown. But have a look at the document. We

18 have no objection --

19 JUDGE BONOMY: Let's have it on the ELMO and ask whatever question

20 you wish to ask.

21 MR. LUKIC: On my document something is written by one of our

22 clients and if possible if Mr. Stamp can provide a document which is a

23 clean version of the same. If the usher can take it from Mr. Stamp.

24 If we can go just bit down so it's -- now it can be seen the

25 source. It says "UNHCR Geneva October the 15th, 1999."

Page 10401



3 Q. And, Dr. Ball, you can see on this document that there is no

4 interruption of the refugee flow from the 6th of April onwards until the

5 10th, as you claimed. So did you have this document in front of you

6 before you composed your report or this is the first time you see it and

7 how you can comment on this?

8 A. Well, actually this is -- we've already discussed this document,

9 counsel. I discussed it in direct, and I consider -- and I discussed why

10 I consider it to be unreliable. It is graphed along with the other

11 sources in figure 3 of my 28 January 2007 report, people leaving Kosovo,

12 four sources. And I've already discussed that I consider it unreliable

13 because it shows a batching effect whereby numbers are averaged or totals

14 are averaged across and you see curiously flat lines, weird plateaus

15 through the data that shows that data was batched up and reported by

16 Geneva at specific points rather than when the data occurred. I can

17 repeat my critique if that's useful.

18 Q. That's fine. I remember your answer. So you still maintain your

19 position that this document is unreliable and you do not accept it?

20 A. I think that it is unreliable because it was collected in Geneva.

21 It was too far from the events.

22 Q. Okay. Thank you, Dr. Ball. That's all the questions I have had

23 for you today.

24 JUDGE BONOMY: Since the document doesn't exist in any other form,

25 I think it should be given an IC number.

Page 10402

1 THE REGISTRAR: That will be IC122, Your Honours.

2 JUDGE BONOMY: Thank you.

3 MR. STAMP: Could I just indicate to the Court the document is

4 substantially a reproduction of page 1. It's a reproduction of a page in

5 Exhibit P737 which is already in evidence.

6 JUDGE BONOMY: Well, we have it in its current form and I would

7 like actually to see it again, please. Thank you.

8 MR. STAMP: I'm just doing it just so that later on there could be

9 ease of cross-reference. It is substantially a reduction of page 1 of

10 Exhibit P737.

11 JUDGE BONOMY: Thank you.

12 Mr. Zecevic, do you have any questions?

13 MR. ZECEVIC: No questions, Your Honour, for this witness.

14 JUDGE BONOMY: Mr. Ackerman, have we exhausted you?

15 Further cross-examination by Mr. Ackerman:

16 Q. Dr. Ball, Judge Bonomy has invited some discussion with you about

17 what we may have learned from consultation with other experts in this

18 field. I think I'll accept that invitation just very briefly. I

19 mentioned the name yesterday of a statistician from I believe the

20 University of Wisconsin by the name of John Steele. I want to read you

21 something and just get your comment on it if I can. Commenting on your

22 reports, your testimony, in Milosevic.

23 "There are two levels to address reliability. First, is the

24 analysis technically correct and secondly, how reliable is the inference

25 of the causal relationship based on the analysis. The problem I see here

Page 10403

1 is that there is little or no precedent with regard to how reliably one

2 can make the inference in this and similar cases. It's not been done

3 often enough in cases where the cause is, in fact, known. The whole

4 concept suffers from a lack of ground truth, as it were. Perhaps the

5 expert can address that issue, but I suspect not. What he has done really

6 seems to be called heuristic statistics, seeking to understand something

7 or to generate testable hypotheses by a data analysis. Such studies can

8 be very useful, but they are not themselves conclusive. They can point

9 the way for more studies, but do not themselves confirm by hypotheses

10 coming from them."

11 What do you have to say about that?

12 A. That's an impressively careful statement. I do not agree with the

13 statement "heuristic statistic," but I do agree that more studies would be

14 helpful. Let me go back, though, point by point, because it's actually --

15 there's terms of art in nearly every line.

16 "Is there little or no precedent with regard to how reliably one

17 can make the inference in this and similar cases?"

18 In human rights, certainly not; however, the methods in historical

19 demography with respect to migration and the methods and the adjustment of

20 censuses with respect to multiple systems estimation are fairly

21 well-known, in fact, we have hundreds of references that we could provide.

22 We've provided several dozen in the original report. It's easy enough to

23 generate many, many more.

24 JUDGE BONOMY: Let me interrupt you there.


Page 10404

1 JUDGE BONOMY: Why do you identify human rights and what is it

2 about human rights that makes it any different from carrying out the

3 analysis in any other field?

4 THE WITNESS: To be crass about it, Your Honour, money. Nobody

5 cares.

6 JUDGE BONOMY: Yeah, but --

7 THE WITNESS: People care a lot about censuses --

8 JUDGE BONOMY: I didn't mean to ask you why perhaps it hadn't been

9 done. What is it about the field of human rights that makes it any

10 different in the sense of making the work that you do any different?

11 THE WITNESS: I don't believe it does, Your Honour, which is why

12 I've applied the methods in this way.

13 JUDGE BONOMY: Indeed, I have to question the description of the

14 field that you're actually in as the human rights field, but --


16 JUDGE BONOMY: -- there's little point in us debating that here so

17 please carry on with your analysis of the question.

18 THE WITNESS: Is it possible to scroll back the transcript so that

19 I can see the question because it was a very, very precise statement or

20 put it on the other screen. It was an enormously carefully crafted --

21 [Trial Chamber and registrar confer]

22 JUDGE BONOMY: I think you can now see it, can you?

23 THE WITNESS: No -- oh, here we go, yes. Now I can. It's very

24 fuzzy. A little bit higher. Can we scroll up another half-dozen lines.

25 There we go. Stop. Thank you. Thank you. That's very helpful.

Page 10405

1 The claim that Professor Steele is making is that we have used the

2 statistics to generate the testable hypotheses, and to the contrary, we

3 drew the hypotheses from the political debate swirling about the conflict

4 in Kosovo. So we certainly did not use this sort of -- this sense of

5 heuristic statistics where we generated the hypotheses from the data

6 analysis. We generated the hypotheses from the -- from the political

7 debate and tried to quantify those hypotheses in ways that would be

8 testable.

9 Similarly, I noticed that Professor Steele talks about the

10 confirmation of hypotheses. Again, he's quite right. The confirmation of

11 hypotheses would depend on a case-control study, not an observational one.

12 This is purely an observational study and so we have confined ourselves

13 very strictly to the disconfirmation of hypotheses, which is, in fact, a

14 much simpler intellectual exercise. It is much easier to reject

15 hypotheses which are clearly contradicted by data than it is to confirm

16 hypotheses which requires a much more elaborate experimental design, which

17 in that -- oops. The screen just went blank. But confirming hypothesis

18 requires a fairly elaborate experimental design and is far beyond the

19 capacity of a statistician to conduct in this kind of -- in a real

20 political situation.


22 Q. Okay. Let's try one more, and to some extent, I think you have

23 already answered this at least a little bit today. So understand that you

24 won't need to repeat things you've already commented on.

25 Here's the statement: "The causal reasoning in the expert's

Page 10406

1 report and testimony has a possibly significant flaw. He is relying on

2 temporal relationships and making the assumption that a cause must precede

3 its effect, thus he rules out, say, NATO strikes as being causal of

4 refugee flow because they took place after the flow began. However, it

5 may be that one cannot apply this kind of logic to human beings, since we

6 are somewhat capable of anticipating future events. In that situation,

7 the usual causal logic may not be applicable. If you see rain clouds

8 gathering overhead and move your party indoors, is the cause the clouds or

9 the anticipated rain storm? How can one really assign causation in such a

10 situation? Can one really say that the rain did not cause you to move

11 indoors?"

12 A. That strikes me as interesting psychological speculation, but I'm

13 willing to stand on several hundred years' of western logic in which we

14 assume cause to precede effect.

15 Q. Yeah, I guess that's what you'd say but, you know, I think the

16 point there is that if you go indoors because you see rain clouds, then

17 the cause of your going indoors may be to avoid rain, but it doesn't rain

18 until sometime after you've gone indoors, and that's the problem I guess

19 Dr. Steele and I both see in your analysis, saying that the bombing

20 preceded the movement in a situation where it was pretty well clear to

21 everyone in Albania, I think, that there was the imminence of NATO bombing

22 was there and -- something one had to get away from. I think that's the

23 point that Dr. Steele was trying to make and the one that I have been

24 trying to make. Maybe that was more of a comment than a question, but if

25 you have something to say about it go ahead.

Page 10407

1 A. Thank you. Well, I think this is an like an excellent -- I mean,

2 I am greatly hopeful that this is near the close. It seems like an

3 excellent place to come full circle to the first major figure in the

4 original January 2002 report because that addresses precisely the point

5 you're raising. If it were the case that people were to leave their homes

6 because they believed that they were going to be bombed in your rain cloud

7 analogy, that might at some level be plausible. But how did they get

8 killed? As we show in figure after figure, figures starting with figure 2

9 in our original -- I'm sorry, I'm not sure which figure it is. You guys

10 had me moving back and forth and it will take a while for me to find the

11 figures. But in the original report, when we show the correlation between

12 killings and migration, the killings match the migration perfectly.

13 Now, I'm willing to agree that we might move a picnic indoors if

14 we see rain clouds because we anticipate the cause of the rain. I'm

15 unwilling to believe that we will then massacre everyone at the picnic in

16 anticipation of the rain. And what we find here, sir, is that the

17 killings and migration match each other over and over again. They match

18 each other at the global level, at the top level across Kosovo. They

19 match each other at the regional level. They match each other at the

20 municipal level. While your claims about people's anticipation might make

21 sense for bombing, they make no sense whatsoever with respect to killings.

22 Q. I think we might be able to get into some issues of coincidence

23 here but I don't think I want to do that. I think we're close enough to

24 the end that maybe I can go into one more issue with you, and it's one

25 that certainly troubles me. I don't know whether it troubles anyone

Page 10408

1 else. And that is that you only used data from this one border point, the

2 Morina border point. And in Milosevic at page 2157 when you were asked

3 about this you said: "That one border point was sufficient because,

4 according to the available data, I believe that more than half of all

5 Kosovar Albanians who left Kosovo did so through this point."

6 So when you say "I believe that more than half left through this

7 point," are we saying that that's simply some belief you have or is that

8 based on some kind of hard data that you had available that would prove

9 that?

10 A. I'm sorry, you're so troubled, Mr. Ackerman. You've used that

11 term many times. You must not sleep well. Perhaps I can help.

12 If you examine the appendices to the Policy or Panic report you'll

13 find an elaborate sensitivity analysis in which we compare refugee flow to

14 a number of countries, principally Macedonia in addition to Albania, but

15 other crossing points in Albania as well as Bosnia, and we do not have

16 extensive and detailed information about the people in Macedonia or in

17 Bosnia or from the other crossing points in Albania. Macedonia, Bosnia,

18 or other crossing points in Albania.

19 But we have some information about them and we used that

20 information to, again, create simulations to say how different would those

21 people's patterns have to have been in order to change our results based

22 principally on the data from the Morina border point. And we found that

23 we would have to make radical assumptions about the people entering

24 Macedonia and Bosnia, that they would have to be completely different in

25 their points of origin and in their patterns of movement. And we found

Page 10409

1 that unlikely given the limited survey data we have from both places,

2 which indicate that people were behaving in roughly very similar ways to

3 the people entering Albania. So it was not a belief based on some sort of

4 instinct but rather on a fairly detailed sensitivity analysis which is

5 available to you should you choose to look at it.

6 Q. Well, in Milosevic you went on to say this about that: "All the

7 other evidence available from other border points in Albania as well as

8 the border points in Macedonia and data from Bosnia suggest," emphasise

9 'suggest,' "that their movement patterns and departure patterns were

10 substantially similar," emphasis 'substantially similar,' "to the patterns

11 found among refugees who entered Albania."

12 Now, first of all, the terms "suggest" and "substantially similar"

13 are not terms of art in your field, are they?

14 A. Actually, they are.

15 Q. Both of them?

16 A. We used "suggest" because we do not say "prove" nor do we

17 say "confirm." We say "suggest" because this is the kind of evidence

18 which alone would leave us speculating, but we would present in a process

19 of fact-finding so that it can contribute with other forms of evidence to

20 make a broader conclusion.

21 Q. Well, did you actually receive data from these other border points

22 that you talk about here that was like the data that you received at

23 Morina?

24 A. No.

25 Q. Do you know how the data from those other border points was

Page 10410

1 compiled? Was it someone using clickers or did you have data that showed

2 when people left their homes and where they came from?

3 A. We had -- no, we did not have evidence from the border points

4 about when people left their -- where -- excuse me -- where people came

5 from. We had evidence on the gross numbers of people crossing rather

6 crudely aggregated across time that was available from UNHCR. We then

7 conducted surveys -- well, I had a group of colleagues that -- "we" is in

8 a very broad sense here. I had a group of colleagues from the University

9 of California Berkeley who conducted almost 2.000 surveys among refugees

10 in Bosnia and -- from the Human Rights Centre at Berkeley, and we used the

11 surveys conducted by Physicians for Human Rights were conducted -- which

12 were conducted on a probability basis in refugees camps in Macedonia. It

13 was on the basis of those two surveys that we concluded that people's

14 movement patterns were substantially the same as the people in -- as the

15 refugees in Albania.

16 Q. Okay. Since you told us that the word "suggest," for instance, is

17 a term of art, then can you tell us with regard to this data a percentage

18 probability figure that is accurate, that it suggests it but to what level

19 of accuracy? Are you at 95 per cent? Are you at 70 per cent? Where are

20 you?

21 A. There's not a probability associated with it; instead, it is

22 analysed via a sensitivity analysis. I don't have that in front of me.

23 If we could pull it up, I'd be happy to walk you through it. It's in the

24 appendix of "Policy or Panic" --

25 Q. You've answered my question. I really don't want to go through

Page 10411

1 your sensitivity now. So you --

2 A. Hang on. That --

3 Q. Can you --

4 A. Your Honour, I'm sorry. May I finish?

5 JUDGE BONOMY: I want to hear the next question.


7 Q. Can you derive a probability? Is there a way you can do it? I

8 know you didn't do it, but is there a way you can?

9 A. There is not a way to derive a probability in a formal sense.

10 There is, however, a way to - which is what we did - derive the range of

11 plausible outcomes based on a wide variation in assumptions about things

12 that may have biased the underlying surveys. There is not a strict

13 probability measure for this.

14 Q. Okay. You went on then again --


16 MR. ACKERMAN: I'm sorry, Your Honour.

17 JUDGE BONOMY: You are going to refer, then, to the appendix to

18 "Policy or Panic" --


20 JUDGE BONOMY -- to do what?

21 THE WITNESS: Yes, Your Honour. I'm sorry?

22 JUDGE BONOMY: You were going to do that for what purpose?

23 THE WITNESS: To show how the sensitivity analysis had been

24 conducted.

25 JUDGE BONOMY: Do you need to do that specifically or to direct

Page 10412

1 our attention to it?

2 THE WITNESS: I would direct your attention to it. If there are

3 questions about the validity of our broader analysis, I would be happy to

4 clarify them at this time.

5 JUDGE BONOMY: Mr. Ackerman.


7 Q. You went on in Milosevic at page 2310 on this same subject: "I

8 examined survey information from other places and considered what the

9 likely pattern from those places would be," the likely pattern. Now, that

10 means you don't have enough data to construct a pattern; you just consider

11 a likely pattern and guessed at what the likely pattern might be?

12 A. No, Mr. Ackerman, as I suspect you know. What a likely pattern is

13 is a pattern that you look at from sample data. When you have sample

14 data, you do not know what the necessary pattern is. You don't know,

15 because you only have a little bit of the data.

16 So what you do with the sample is you make an estimate of what the

17 pattern would be and that estimate is subject to the error that -- or the

18 margin of error that governs the sampling process. So I hope that is

19 sufficient, but it is not a -- it's not a guess.

20 Q. You went on --

21 A. Yes.

22 Q. This carried on in Milosevic with some more --

23 A. Mm-hmm.

24 Q. -- answers. And you say this: "We know something," talking about

25 what went on at these other border points, "we know something from the

Page 10413

1 counts kept by the Albanian border guards at two other primary crossing

2 points. We know something -- although, in my opinion, with quite a bit of

3 error, we know something about refugees who crossed into Macedonia, and we

4 do know something about refugees who crossed into Bosnia, and we know a

5 little bit with less certainty about those who crossed into Montenegro."

6 A. That's right.

7 Q. So what you then finally say in Milosevic that you can -- you

8 think you can take it from assumption to reasonable estimate.

9 A. No, sir. That's your term. What I mean by something is again an

10 attempt to explain to lay people the difference between the quality of our

11 scientific understanding of the cross -- of the people who crossed at

12 Morina relative to the people who crossed at other places. We know

13 something. We have a considerable error around that. Knowing something

14 with error is not an assumption, sir, and I'm afraid to have to have

15 repeated this. I kept -- started keeping track of these corrections this

16 morning. This is now the seventh time I've corrected you on this point.

17 I hope that we can become clearer as we continue.

18 Q. Well, you know, I wasn't using my words. I was using yours

19 words.

20 A. No, sir. You are introducing the term "assumption" repeatedly.

21 Q. Here are your exact words: "So knowing those other things, I

22 think we can narrow the claim I've made from an assumption to, I think, a

23 reasonable estimate."

24 Now, those are your words. I didn't make up that word

25 "assumption." You used it in Milosevic.

Page 10414

1 A. By refuting it, Mr. Ackerman, by saying that we have made an

2 estimate. Okay. Finding the word in a sentence is not the same as me

3 asserting it, as you with I believe your elaborate training in logic,

4 should understand.

5 Q. Well -- okay. I thought I was through, but I've got to do one

6 more thing, if I can find it. At page 56, line 9, when Mr. Lukic was

7 asking you questions about people leaving before there was ever any

8 bombing, first you said this: "If there were convoys before the bombing,

9 clearly, the bombing did not cause those convoys of refugees to be on the

10 move." And then very shortly after that you said: "I have no idea why

11 they're moving."

12 So wouldn't it be fair to say that what you said in that first

13 statement, that "clearly, the bombing did not cause those convoys of

14 refugees to be on the move" was just an assumption on your part?

15 A. No, sir, it was not. I don't know why they were moving, but I

16 think I can be clear why they were not moving. I know what their -- the

17 reason was not; that does not mean that I know what the reason was.

18 Q. Well, you said these people are certainly not moving because their

19 region is being bombed which is different from what you said in your first

20 statement with regard to it. Now, if you have more to say it --

21 A. I'm sorry, Your Honour, this strikes me as sophistry. I'm not

22 sure that I can clarify further.

23 Q. If you have more to say about that you can, but otherwise I'm

24 finished with my questions and I thank you.

25 JUDGE BONOMY: The recent exchanges are matters that we can

Page 10415

1 reflect upon at the end of our -- when we come to deliberate; they're not

2 matters that need any further explanation from you, Dr. Ball. I have at

3 this stage two questions. One is for Mr. Stamp. There's been some

4 reference in the recent questions to this border crossing point at Morina

5 and on the second occasion that Mr. -- that Dr. Ball gave evidence in the

6 trial of Slobodan Milosevic, he was presented with a number of documents

7 to try to identify the locus. Now, are they exhibits in this case because

8 they don't appear on your list?

9 MR. STAMP: No, Your Honour.

10 JUDGE BONOMY: Are we all satisfied that we know where this

11 crossing point was and that the issue that appeared to be such a big one

12 for Milosevic is no longer an issue.

13 MR. STAMP: I thought so that it was clarified clearly in

14 Milosevic where he marked a map on the border.

15 JUDGE BONOMY: I'm sorry, he?

16 MR. STAMP: He marked a map in Milosevic.

17 JUDGE BONOMY: Do we have it?

18 MR. STAMP: I can produce it but it wasn't raised. I thought that

19 was a settled issue.

20 JUDGE BONOMY: If we haven't seen the map, we don't know where it

21 is. Please don't keep secrets from us, Mr. Stamp. If you want to prove

22 the case you may have to produce that material so that we can be

23 satisfied. You're right that it hasn't been challenged in the

24 cross-examination, but I still think that you should produce the material

25 that was used to establish the crossing point.

Page 10416

1 MR. STAMP: Very well.

2 JUDGE BONOMY: The other question is one I have for you, Dr. Ball,

3 that may display a basic misunderstanding of something but I hope it

4 doesn't. Mr. Milosevic tried to ask you about the notion that NATO and

5 the KLA were working together and tried to suggest that that was a

6 separate hypothesis from the three that you were working on. First of

7 all, is that a different hypothesis from any of the three you were working

8 on?

9 THE WITNESS: It could be a separate hypothesis in order --

10 JUDGE BONOMY: And second -- well, I think that answers that

11 question.

12 My second question then is have you somehow or other in the 2007

13 report dealt with that?

14 THE WITNESS: No, Your Honour, we have not.


16 THE WITNESS: And the reason is the same that I explained to

17 Milosevic, that in order for a combined or interaction effect to have --

18 to have meaning in a statistical analysis, both of the underlining

19 components must themselves have been significant. They both themselves

20 must have had values that were significantly different from zero by --

21 they could not have occurred by chance. That did not occur in any of the

22 four models until consequently we were unable to calculate any meaningful

23 test of an interaction between the two. I think I may have said it in

24 less formal language then, but that's the --

25 JUDGE BONOMY: You seem to have been prevented from dealing with

Page 10417

1 the question in that trial.

2 THE WITNESS: Well --

3 JUDGE BONOMY: But I don't think I have understood your answer at

4 the moment.


6 JUDGE BONOMY: Can you help me a little. Can you clarify that?

7 THE WITNESS: Okay. Let me -- I'm --

8 JUDGE BONOMY: I don't understand what you mean by both of the

9 underlying events must themselves have been significant.

10 THE WITNESS: Right.

11 JUDGE BONOMY: And they both must have had values that was

12 significantly different from zero.

13 THE WITNESS: Right. So when we construct a statistical model,

14 the logic is that we are predicting some value, we are trying to make an

15 estimate of something. In the case of the analyses that we got to this

16 morning with the estimate of the total number of killings and migration,

17 we're estimating the total numbers of killings and migration and we're

18 using information -- even though we know those values, we're using

19 information about KLA and NATO activities to try and make those

20 predictions. Okay. So the idea is to determine how good a prediction of

21 killings and migration we can make based on our knowledge of NATO and KLA

22 activity.

23 The core test about whether or not we have managed to make a

24 reasonable -- whether or not one of the effects, in this case KLA or NATO

25 activity, whether that effect has some impact is whether we can look at

Page 10418

1 the value that we calculate for the level of its impact and distinguish

2 that value from zero. And the reason we do that is because in any

3 statistical model some values will occur by chance and we have to be able

4 to rule out that the value we have discovered has occurred by chance. If

5 we fail to find the values of, in this case, the KLA actions and the NATO

6 actions, both as significantly different from zero, that we can reject the

7 claim that these both could have occurred by chance, then we can construct

8 a combined measure of the two and test whether the combined measure of an

9 interaction between the KLA and NATO, which would test the idea that they

10 were working together, could itself be a further higher-order cause of

11 killing and migration -- killing and/or migration. However, in none of

12 the four models we examined were we able to find both KLA and NATO effects

13 that were significantly different from zero. So it's as if we had to

14 multiple two numbers, one of which could have been zero, and so that would

15 have been a meaningless exercise.

16 JUDGE BONOMY: I'm still struggling, I'm afraid.

17 You've been able, obviously in relation to each of them,

18 separately, to come to the conclusion that their effects were

19 significantly different from zero, otherwise you would never have been

20 able to deal with the hypothesis in relation to each of these two

21 separately or have I misunderstood something totally here?

22 THE WITNESS: In some models we had some values that were

23 significantly different from zero, in others not. In fact, the strictest

24 possible rejection of the claim that NATO, for example, was the cause of

25 the killings, would be to find that its effect could have occurred by

Page 10419

1 chance. That would be a sort of slam-dunk rejection of the NATO

2 hypothesis. However, we found that in some of our models NATO had a

3 significant effect and in some of the models KLA had a significant effect

4 but not in the same model. Okay. So it was not that we could pick a

5 model and then test the interaction of the two. That's the limitation,

6 sir.

7 JUDGE BONOMY: Now, are you saying the job couldn't be done on the

8 material that you had, or that that actually does demonstrate something

9 negative?

10 THE WITNESS: It could not be done given the material. I would

11 hesitate to speculate further.

12 JUDGE BONOMY: All right. Thank you.

13 Do you have re-examination, Mr. Stamp?

14 MR. STAMP: Yes, Your Honour.

15 JUDGE BONOMY: Do you know how long, roughly, you'll be?

16 MR. STAMP: 20 minutes.

17 JUDGE BONOMY: Let me check with the interpretation because of

18 the --

19 MR. STAMP: Perhaps, Your Honour, before we do, it might well be

20 wise to take the adjournment here --

21 JUDGE BONOMY: No, the reason I don't want to do that is because

22 it takes time to make arrangements for the next witness, and therefore I'm

23 anxious that these arrangements should be made during the break and not

24 cause us to be interrupted again later in the day. So I'd like to find

25 out if the interpreters could bear with us until 1.00.

Page 10420

1 THE INTERPRETER: Until 1.00, yes.

2 JUDGE BONOMY: So you can --

3 THE INTERPRETER: Yes, Your Honour --

4 JUDGE BONOMY: Continue to 1.00. If we don't make it, we don't

5 make it, but let's see if we can make it.

6 MR. STAMP: What I wanted to do was to find Exhibits 70 and 71 in

7 the Milosevic case which I referred to at pages 2320 of the transcript in

8 which the witness clarified where Morina was. There were two Morinas; one

9 on the Albanian side of the border where he and his team collected

10 material and there was another Morina somewhere within Kosovo which I had

11 thought settled that issue. But I would think that I should bring the map

12 because I have a copy of the map.

13 JUDGE BONOMY: I don't think it's necessary. I think you can

14 produce that as an exhibit and we can interpret the evidence in the light

15 of it.

16 MR. STAMP: Very well.

17 JUDGE BONOMY: I don't think it's necessary to go back over it

18 with the witness.

19 MR. STAMP: Maybe I could just ask him one question to confirm.

20 JUDGE BONOMY: Very well.

21 Re-examination by Mr. Stamp:

22 Q. You recall in the Milosevic case you were shown a map of southern

23 Kosovo and you identified the village of Morina near to which was a border

24 crossing?

25 A. Yes.

Page 10421

1 Q. Was this village of Morina on the Kosovo side or the Albanian side

2 of the border?

3 A. On the Albanian side.

4 Q. Now, about how far from the border was this village?

5 A. A few dozen metres. It's just a scattered bunch of houses and a

6 border post. There's not much there.

7 Q. Thanks.

8 You said in cross-examination that you received United States

9 government support for your work, including the principal report, that is,

10 support from the United States State Department and United States AID.

11 Can you just tell us briefly the nature of the support and whether or not

12 the support had any role in your statistical work.

13 A. The money was given to our partners at the ABA/CEELI they

14 subgranted that money to us at the American Association for the

15 Advancement of Science. I had no reporting relationship with CEELI on

16 that money other than to conduct the research and I had no relationship

17 whatsoever with the State Department. I made one presentation to the

18 State Department. They asked no questions of any substantive nature, nor

19 did they offer any suggestions.

20 Q. Thank you. I think you explained this quite adequately, but since

21 you were asked a couple of times in different forms, the system that you

22 used to determine the time at which people left their homes during the

23 period when there was a report of this unilateral cease-fire by the

24 Yugoslav government. Now, was the system used to measure the time the

25 people left their home based on a calculation for a model or was it based

Page 10422

1 on the actual report of the people from the individual interviews, or

2 maybe I should ask you just to explain how you worked out the period when

3 they left their home.

4 A. The short answer is that both are true. Models are built from

5 data, so we built the model of transit time from the information given to

6 us in a series of interviews that we collected and that -- that Human

7 Rights Watch collected and that Physicians For Human Rights collected.

8 Q. You were -- it was read to you a passage from your article

9 entitled "Statistics and Slobodan," in which it was stated - and I am

10 paraphrasing now, because I don't have the exact quotation right here -

11 that it was not possible in which -- the assertion was made in which you

12 said it was not possible to separate the actions of the KLA and Yugoslav

13 forces.

14 Do you recall that or would you like me to find it? And I think

15 you wanted to explain the context in which you made that statement in that

16 article. Would you like to look at the article before you answer or can

17 you --

18 A. I think that's unnecessary. The context was that given the data

19 we had, it was not possible to separate those actions. It would be

20 possible if one had extensive evidence on the movements of Yugoslav

21 forces. That would be a quite different study and could be done, but we

22 didn't have that data.

23 Q. Thanks. An issue was raised as to whether or not it was valid or

24 accurate for you to be estimating exactly 10.356 people killed, and I

25 think you were trying to explain that it was an estimate and that there

Page 10423

1 was a margin or a confidence interval. Can you look at page 6 of your

2 report, the principal report, the second paragraph beginning "second."

3 A. Yes. It's the third line from the bottom, I believe.

4 Q. You gave there the confidence interval.

5 A. That's correct.

6 Q. And prior to that, you said that the estimate was generated by

7 means of a widely used demographic statistical technique known as multiple

8 systems estimation.

9 A. Correct.

10 Q. Can you -- I know it's elaborated upon in your appendix, but can

11 you in two sentences in layman's terms explain how does a multiple systems

12 estimation work, especially in light of the fact that you have had recent

13 figures in respect to death -- deaths which confirm the older figures that

14 you used.

15 A. First, I think the new data confirmed the pattern of the data that

16 we presented earlier. We have no -- we've neither confirmed nor

17 disconfirmed the total estimate using the new data. As I explained

18 before, that would require substantially greater resources.

19 Multiple systems estimation, there are a couple of metaphors which

20 might be possible, but I think that the easiest to understand is to

21 imagine that we approach a dark room and we have a handful of small rubber

22 balls that bounce very vigorously. And the idea that we have is that we

23 would like to have some sense of how big the room is. We can't go in and

24 measure it for some reason, but we throw the balls into the room as hard

25 as we can and we listen to the balls bouncing about, and we can hear the

Page 10424

1 balls striking each other on occasion and we hear the balls click, click,

2 striking each other.

3 We then gather the balls, go to another room, and throw the balls

4 again with the same force into the second room and we again listen for the

5 balls to spread through the room and then occasionally hit each other, and

6 we hear them striking click, click, click, click, click. We would draw

7 the inference that the second room is considerably smaller than the first

8 room, because using the same number of balls and the same number of force

9 to throw into the room, the balls encounter each other more frequently

10 because there's less room for them to spread into the space.

11 Similarly multiple systems estimation is a process of taking the

12 enumerations of some kind of phenomenon, in this case human deaths, human

13 killings, and saying that each of the projects, the seven projects we now

14 have that have enumerated deaths have in some sense been thrown into the

15 space of deaths that occurred in Kosovo during this period. And what we

16 then do is measure how often these projects encounter each other, how

17 often do they bump into each other in that space. The more frequently

18 they bump into each other, then the smaller the space we infer to be. The

19 less frequently they bump into each other, the more they spread out in

20 that space, the larger we infer that space to be. Another approach,

21 another intuitive approach to the same problem is the following --

22 JUDGE BONOMY: I think we understand this well enough not to have

23 such a long answer, Dr. Ball. I thought I understood it before the

24 question was asked, and I don't think it really justifies this extensive

25 further investigation.

Page 10425

1 Please carry on with something else, Mr. Stamp.


3 Q. Thank you very much, Doctor.

4 Apart from your estimate, are you aware of other teams, for

5 example, Physicians for Human Rights or the demographers Speigel and

6 Solana and other persons that might have done estimates of the total death

7 in Kosovo and as a ball-park, if you're aware do these estimates -- are

8 they consistent with yours, in a ball-park manner?

9 A. Yes. Speigel and Solana are epidemiologists. Physicians for

10 Human Rights conducted a quite different epidemiological study that was

11 conducted early in the conflict, mid to -- mid-April. Both estimates are

12 consistent with ours. Neither estimate measures precisely the same thing,

13 the same period or the same deaths exactly that ours do, but our broad

14 confidence intervals all agree with each other.

15 Q. Thanks. You were asked a lot of questions about addendum 2, and

16 it might well be that you're -- you have explained all there is to explain

17 about it. When I say addendum 2, I mean your last addendum. May I just

18 try to summarise what I think you're saying just to confirm, if possible,

19 or to clear up any ambiguities that might exist in respect to what was

20 said about it.

21 MR. ACKERMAN: Excuse me. Your Honour, none of us remember asking

22 any questions about addendum 2. We just had a little brief conference

23 here. I don't think any of us did.

24 MR. STAMP: Sorry. I meant the addendum of the 28th of January,

25 2007, which is the second addendum that he produced.

Page 10426

1 JUDGE BONOMY: That was certainly explored. Please carry on.


3 Q. You produced this addendum at our request to incorporate new data

4 that is available to the OTP. This is what you said in chief. However,

5 you also produced table 1 because when you were previously here you had

6 been requested by the -- you had been requested in the course of your

7 testimony to produce it.

8 A. That's correct.

9 Q. However, apart from that, it is your view that it is because of

10 its limitation not a useful tool to do the sort of analysis that you did

11 in the principal work.

12 A. That's correct.

13 Q. Because -- and if I may give you three reasons and ask you to

14 comment. One, it measures only one peak not a flow over a period of time?

15 A. Right.

16 Q. And two, for a variety of reasons, you made it ultimate

17 conservative which you have explained in that report.

18 A. Yes.

19 Q. And what you're saying is that even if all of the boxes had

20 inconclusive or coincides, it would not be a tool that could be used for

21 the analysis that we are concerned about here or that is the subject of

22 your principal report.

23 A. There would still be substantial questions to be asked about the

24 moments during which they were not peaks, yes.

25 Q. In fact, in your substantive report, you have in the appendices

Page 10427

1 done -- breakdowns of municipalities and you also preceded, generally

2 speaking, with a breakdown of the country into four regions and that is

3 reflected in figure 3 of the substantive report?

4 A. Yes.

5 Q. And also if you look at the charts at page 59 and page 60, you

6 have analysed the regions, the flows in the regions --

7 MR. LUKIC: Your Honour, I really apologise, but we cannot follow

8 who asked what in the cross-examination. So if our learned friend could

9 point out in the transcript which part is he --

10 JUDGE BONOMY: Yeah, it's getting too complicated for me. We're

11 going to have to adjourn now and resume after lunchtime if this has to be

12 done in this sort of detail. I'm now confused, where I thought I was

13 clear, and I need to ask some questions in any event. So we're going to

14 go on for longer than we anticipated. I'm sorry about that, Dr. Ball. I

15 thought we could finish this, but we can't. So could you please leave and

16 we'll be resuming at 2.00.

17 THE WITNESS: Yes, Your Honour.

18 JUDGE BONOMY: You should bear in mind, Mr. Stamp, that generally

19 speaking, leading questions in re-examination are not helpful to the

20 Bench.

21 MR. STAMP: Your Honour, I was just referring him to the evidence

22 and summarising it before the question. This is evidence that's --

23 JUDGE BONOMY: That's precisely what you should not be doing in

24 re-examination. You should be letting the witness answer the questions

25 himself.

Page 10428

1 [The witness stands down]

2 --- Luncheon recess taken at 12.56 p.m.

3 --- On resuming at 1.59 p.m.

4 [The witness takes the stand]

5 JUDGE BONOMY: Mr. Stamp.

6 MR. STAMP: Thank you, Your Honour.

7 Q. So back to where we broke off. We were going back to questions

8 you were asked about the graphs on page 59 and 60 which explain the flows

9 and your analysis in respect to the regions and which showed the flow --

10 correlations of KLA activity and NATO activity.

11 MR. STAMP: May I just find the precise quotation from the

12 transcript for that question.

13 Q. Right. You said in -- these graphs took us to the heart of the

14 matter in explaining these flows, and you said that there are some

15 anomalies, some of which are discussed in the text, and that what I want

16 to ask you about.

17 The question is this: Having regard to that, to what I just

18 showed you and those graphs, do these anomalies impact on the results and

19 could you explain without making it complicated very briefly why they do

20 not.

21 A. No, they do not impact the results. As we discussed in the

22 original report, the single most interesting anomaly is in the refugee

23 flow in the eastern region where the residual pattern crosses. It

24 reverses the pattern of the original data, indicating that there is

25 perhaps some kind of influence of KLA or NATO activity at that point in

Page 10429

1 that region. That is interesting, because it's throughout the rest of the

2 data unique. None of the other graphs show any such relationship, so it's

3 an interesting point. It was discussed at some length in the Milosevic

4 trial and it is also discussed here in the text of this appendix.

5 We did not consider it an especially important point because it is

6 so focused. It is mentioned in the original -- I mean, in the body of the

7 material along with the figures which do not break down by region but

8 rather look at Kosovo as a whole, and then there's a reference to these

9 two more detailed graphs.

10 Q. Thank you. And that's an example of the anomalies that you were

11 discussing?

12 A. It's not only an example; but it is the anomaly. I mean, it is

13 the only one.

14 Q. Thank you. With the leave of the Court, I would like to show you

15 a map that you -- well, a copy of the maps that you marked in the

16 Milosevic case and ask if you could point out where the town or the

17 village of Morina is. Could that be placed on the ELMO? And that's the

18 one that's adjacent to the border post. Can you think you could do that?

19 Do you think you could recall it?

20 A. Sure.

21 [Prosecution counsel confer]

22 THE WITNESS: I'll need a magnifying glass.

23 MR. STAMP: I think we might have to zoom it somewhat.

24 THE WITNESS: I can't see it here.


Page 10430

1 Q. I need you to show the bottom half of the -- we need to show the

2 bottom half of the map.

3 A. There we go. Yeah, we're going to have to zoom it a lot. It's

4 not very focussed. It's actually more like that. Okay. Can I get up and

5 look at it more closely? It's very finely grained.

6 MR. STAMP: No, no. Can we just show the -- can we zoom out,

7 please.

8 Q. Can you mark it, can you circle the area where it is?

9 A. Yeah. I have to look at it very closely. I'm sorry, my eyes

10 are -- the letters are very small. Comes down through here. That's the

11 town right there.

12 Q. Right on the Albanian side of the border, that is?

13 A. That's correct. And you'll see the road comes down through --

14 from Prizren down to Srb and then the long debate that was held with

15 former President Milosevic was about this town here, which I apologise for

16 being unable to pronounce Srbica on the Kosovar side of the border.

17 Morina is noted here as where my pen is marking. Morina is shown a bit

18 off the border, and that's perhaps where the houses are, but the locals in

19 this area refer to the entire border crossing point in this area as

20 Morina. So when they talk about the border point, they talk about it

21 being the border point at Morina. It's right there.

22 Q. Can you just circle "Morina," please.

23 A. [Marks].

24 Q. And just to be sure. I think I asked you. You said that there

25 was another Morina within Kosovo itself.

Page 10431

1 A. Well, former President Milosevic said so, and he pointed it out.

2 I'm not sure where it is.

3 Q. Okay. Thank you.

4 A. There are actually many towns, probably, by that name.

5 MR. STAMP: Thank you very much, Your Honours. Nothing further.

6 Could this -- before I sit, I'm reminded that -- could this be

7 given an IC number?


9 THE REGISTRAR: That would be IC123, Your Honours.

10 Questioned by the Court:

11 JUDGE BONOMY: Dr. Ball, I wonder if you could answer the

12 following questions, please, as precisely as you can. We were told that

13 the January 28th, 2007, report was written on the instructions of the

14 Office of the Prosecutor. Can you tell me, please, what these

15 instructions were.

16 A. Yes. They asked me to do two broad things. First, they asked me

17 to follow-up the questioning that I believe originated from Judge Kwon in

18 the Milosevic trial, which led to -- in which he asked specifically for us

19 to break down by municipality what were then, I believe, figures 8 and 9

20 in the original report. Let me refer to them to make sure I've got them

21 just right. Those figures showed in summary form the discussions that

22 we've been having about this issue of peaks and valleys.

23 JUDGE BONOMY: That's sufficient for --

24 A. For the question.

25 JUDGE BONOMY: For that point.

Page 10432

1 A. Right.

2 JUDGE BONOMY: There must have been more to the instruction than

3 that.

4 A. Yes, sir. The second point that I was asked to investigate would

5 be the impact of potential additional data that was not available when we

6 did the 2002 report, but has become available since. There were four data

7 sets I was asked to look at --

8 JUDGE BONOMY: We've got that. So these were the two things you

9 were instructed to do?

10 A. That's correct.

11 JUDGE BONOMY: Now, I said I think when you were giving evidence

12 in chief that I hadn't grasped that part of the message in this new report

13 was that it had inadequacies, and I'm still having difficulty accepting

14 that that point is being made to us. If you could look at it, please.

15 You have a hard copy there, do you?

16 A. Yes, sir. If I may have a moment to find it. Here we go. Which

17 in specific do you want to look at?

18 JUDGE BONOMY: Look first at the beginning, the executive summary,

19 the first paragraph, the second line you'll see the words: "We confirm

20 and expand upon the results presented in 2002."

21 A. That's right.

22 JUDGE BONOMY: And then you go on to say: "We describe a revised

23 and expanded analysis of the potential causal relationship between KLA or

24 NATO activity and the subject we are dealing with."

25 A. Mm-hmm.

Page 10433

1 JUDGE BONOMY: You then say that: "We discuss how our coding

2 scheme was intentionally conservative to provide the strongest possible

3 test of the claims," so presumably that's a reference back to 2002 also.

4 You then in the next section, in the third line, you say that in

5 both the 2002 and the present report you consider three hypotheses and you

6 say what these are. And then in the next paragraph you say: "If KLA

7 activity or NATO air-strikes occur immediately before or during periods,"

8 plural, "of high levels of killing and migration, these events may

9 plausibly be the cause of the rise and fall pattern." So that's a

10 reference to the pattern which emerged in the original report.

11 And then if you go over the page you talk in the first full

12 paragraph over the page that -- about the current analysis showing no

13 discernible causal relationship. And then you say: "Because the data

14 coding process was automated for the 2007 analysis, the analysis reached

15 slightly different conclusions about a few peaks; however, none of these

16 recategorisations alters the conclusions."

17 The next paragraph: "The procedure used in the current report was

18 analogous to the procedure followed in 2002; however, to further increase

19 precision the analysis was automated."

20 And then skipping down to the last paragraph on that page: "As

21 noted above, the 2007 analysis is consistent with the conclusions of the

22 original 2002 analysis."

23 We then have table 1 and then in the next page after that, third

24 line in the graphs presented -- well, you're referring to the graphs

25 presented and you state what emerges from some of these. And then in the

Page 10434

1 next section, revaluating the effect of the coding rules you say: "Both

2 the 2002 and 2007 analysis used conservative coding rules."

3 If you go to the next paragraph, the third line: "Peaks towards

4 the end of the period are thus more likely to be labelled as

5 inconclusive."

6 A. Right.

7 JUDGE BONOMY: And when I look at the various graphs there, they

8 don't just have one peak; they have a number of peaks. And it was being

9 suggested that this analysis that was carried out in this report was

10 confined to a single peak and thus less inadequate, quite contradicting, I

11 would suggest, the way in which the reader is likely to understand the

12 import of this report. Now, would you like to comment on that, please.

13 A. Yes, I would first like to say that I apologise for any confusion

14 that may have been caused by that. Our comments here -- unfortunately

15 we've got a bit of blinders because we're focussed very, very tightly in

16 this discussion on the few paragraphs that surround figure 8 and 9 in the

17 original report. So all of our discussion about confirming earlier

18 hypotheses have to do with the conclusions we made with respect to those

19 particular figures which are being expanded upon here. That said, and as

20 I've said throughout the last two days, our preference very strongly is

21 for any statistical conclusions to be made on other grounds not

22 necessarily -- not on these peaks and valleys analysis but rather on the

23 graphs and residuals analysis -- the regressions and residuals analysis

24 that we discussed earlier this morning.

25 JUDGE BONOMY: Does it not ultimately boil down to peaks and

Page 10435

1 valleys, too?

2 A. It does but it boils down to peaks and valleys in a much more

3 finally grained way that takes into account both the presence of the

4 putative causal phenomenon and the putative results together and their

5 absences. And what this -- I mean -- excuse me, what this method lacks is

6 the ability to consider what happens then there are no putative causes or

7 putative results.

8 JUDGE BONOMY: Well, do you agree with me that it was unfortunate

9 that these limitations were not highlighted in this report?

10 A. I agree with you, sir. Yes.

11 JUDGE BONOMY: Thank you very much.

12 Mr. Visnjic.

13 MR. VISNJIC: [Interpretation] Your Honour, I do know that this is

14 rather unusual, but by your leave, I just wanted to pose one question to

15 the witness concerning the map shown by Mr. Stamp.

16 JUDGE BONOMY: Certainly you may pose a question about the map.

17 MR. VISNJIC: [Interpretation] Thank you.

18 Further cross-examination by Mr. Visnjic:

19 Q. Mr. Ball, am I correct in saying that in one of your earlier

20 reports or statements, although I do not have the time reference, you

21 stated that the road going through the village of Morina where you were

22 and which eventually leads one to a border crossing is an unpaved road?

23 A. No, sir, it's a two-lane blacktop.

24 Q. Thank you.

25 A. There's -- if I may. There's a photograph of that road, actually,

Page 10436

1 on the cover of Policy or Panic if you'd like to see it.

2 JUDGE BONOMY: Mr. Ackerman.

3 MR. ACKERMAN: Your Honour, in Mr. Lukic's examination, the

4 witness himself said that it would help explain if we looked at page 60 in

5 the graphs there. As Mr. Stamp pointed out, what the witness talked about

6 there were anomalies and Mr. Stamp read that testimony. And then when he

7 was asked about that by Mr. Stamp, he pointed out that one anomaly there

8 in the graph of the east area and then said that was the only anomaly.

9 And I feel a need to try and clarify that because when I look at it, I see

10 a number of anomalies and I wonder why they're not. If the one in the

11 east is an anomaly, why isn't the one in the north and two in the south

12 and two in the west -- three in the west, actually, maybe even four.

13 JUDGE BONOMY: I always have difficulty finding this because the

14 pages aren't numbered --

15 MR. ACKERMAN: It's 3880, page 60.


17 MR. ACKERMAN: 3880 on the top there.

18 JUDGE BONOMY: Yeah. Can you direct me to the reference to

19 anomalies, please?

20 MR. ACKERMAN: It's -- Mr. Stamp was reading from the transcript

21 at page 93, line 15.

22 JUDGE BONOMY: So it's not something that's said in the report?

23 MR. ACKERMAN: No, it was in his testimony. When Mr. Lukic was

24 examining him, he referred to anomalies and then Mr. Stamp read it, as I

25 understand it, from the transcript at 93, 15, to be precise in his

Page 10437

1 question and then we get down to 94, 12, and he says that the one in the

2 east graph was the only anomaly. And I, in some ways, hesitate to ask for

3 leave to ask any more questions. Sometimes when you get what you want,

4 you wish you hadn't.

5 [Trial Chamber confers]

6 JUDGE BONOMY: Yes, you can ask that, Mr. Ackerman.

7 Further cross-examination by Mr. Ackerman:

8 Q. Dr. Ball, like some other graphs we were looking at yesterday,

9 because of the numbers from each of these regions that you were dealing

10 with, the scales of each of these four graphs differ in some cases

11 significantly. So if we were to simply try to compare them one with the

12 other regarding height and depth of peaks and troughs, that wouldn't be a

13 fair comparison. Is that a fair statement to start with?

14 A. Well, that's the point -- that would not be the point of the

15 graphs, no.

16 Q. Right.

17 A. It's not a matter of fairness, sir; it's a matter of what the

18 objective is in analysing the graphs.

19 Q. Fairness may have been a real bad choice of words.

20 A. Yes.

21 Q. If we look at the west graph.

22 A. Mm-hmm.

23 Q. I see around the 24th of March in the lower line, the grey line, a

24 dip that doesn't correspond to the darker line.

25 A. Mm-hmm.

Page 10438

1 Q. I see around maybe the 8th or 9th or 10th of April a dip that

2 doesn't correspond. Right after that a kind of a trough that doesn't

3 correspond.

4 A. Mm-hmm.

5 Q. If you go and you see these and -- is it your position that these

6 are not anomalies?

7 A. The anomaly that is of interest is the point at which the lines

8 cross each other in opposite directions. They go apart from each other,

9 that they diverge from each other, is interesting, but only interesting

10 insofar as the error would support that they are truly different. Again,

11 these are estimates and we must be concerned about spurious differences.

12 The major difference that we see, the big spike that's different, the

13 place where the two lines truly do cross each other and cross each other

14 with some vigour is the point at the 6th of April or really from about the

15 3rd to about the 7th or 8th of April in the eastern region. That's the

16 only place in these four graphs where there is a difference that is

17 significance enough to be considered something that we would interpret.

18 Q. That clarifies it because when I -- what I understood as the

19 anomaly was that large dip after where they cross each other. So if we go

20 over to the north graph, there is a small place where the lines cross. I

21 assume that would be a anomaly but maybe one that you wouldn't think is

22 significant?

23 A. They cross going in the same direction.

24 Q. Right.

25 A. If you remember the definition of anomaly, they cross going in

Page 10439

1 opposite directions. That's what's of interest.

2 MR. ACKERMAN: That's all I have, Your Honour. Thank you for

3 leave to do that.

4 JUDGE BONOMY: Thank you.

5 [Trial Chamber confers]

6 Further questions by the Court:

7 JUDGE CHOWHAN: Well, I acknowledge and appreciate your expertise

8 and skills on the subject. Obviously, as I understand, that you were

9 given data and those you used empirically in the process of statistical

10 evaluation and we got the end result, and on which you had a command to

11 tell us all what you have been doing. But the question is: Did you

12 verify the data or it appeared truthful? Whatever. I mean, just try

13 to -- that's all.

14 A. I think that's a great question. I -- as I said to some similar

15 questions, yours is a subtler question, but to some similar questions

16 earlier I said I did not conduct field investigations so I have not

17 verified the data in the sense of going to the field and confirming

18 individual stories; however, at a statistical level, we have confirmed the

19 data or verified the data or done sanity checks on the data is the term

20 that statisticians would use. And a sanity check would have the following

21 feeling. We would say: For example, when we're looking at lots of data

22 sets that describe the same events, do we see a distribution of repeated

23 discussions of the same events. Do we see the kind of clustering around

24 events that we would expect.

25 If people were, for example, committing some sort of wholesale

Page 10440

1 perjury in these stories, we would expect them to not converge at all.

2 They would tell stories. It would be very, very hard. It's an immensely

3 complex task to coordinate them so that two people told one story and a

4 third person told a separate story and then four, five, and six persons

5 told a combined story. That starts to require a tremendous amount of

6 coordination. Instead, when we think about wholesale fabrication, we

7 think of people going in and making things up, and we did not see that.

8 And that's one way that we would, at a statistical level, look for

9 large-scale fabrications.

10 A second way we concerned ourselves about the veracity of the data

11 was simply to ask the counter-factual question which is: What would

12 happen to our results if some fraction of the data were, in fact, false.

13 And So what we did was systematically perturb the underlying data.

14 We just switched it all around. We made it lies and then we reran the

15 analysis repeatedly as a kind of sensitivity analysis to falsification.

16 Again - and this is found most in "Policy or Panic." There's a little bit

17 of it in what the Prosecutors called the Principal Report. We found very

18 little change in the results. We find the results to be very robust to

19 what was told to us. This is what I was trying to explain earlier when I

20 said that data analysis is like listening to a voice in a crowded room,

21 that if you can make out the voice, external noise in the room does not

22 necessarily change your ability to hear the voice.

23 JUDGE CHOWHAN: Thank you.

24 JUDGE BONOMY: Well, Dr. Ball, that completes your evidence.

25 Thank you for coming again to the Tribunal to give it. You're now free to

Page 10441

1 leave.

2 THE WITNESS: Thank you, Your Honour.

3 [The witness withdrew]

4 JUDGE BONOMY: The next witness, Mr. Stamp.

5 MR. STAMP: Your Honour, that witness is K88, for whom I think

6 special arrangements would have to be made, and my friend Ms. Daniela

7 Kravetz would be taking that witness.

8 While we wait for those arrangements to be made, there's an

9 application I'd like to make.

10 JUDGE BONOMY: Well, we can't, because we have to be out of the

11 courtroom for these arrangements to be made, I'm afraid.

12 MR. STAMP: It's a brief one.

13 JUDGE BONOMY: All right.

14 MR. STAMP: The Court had directed myself and Mr. Ivetic to

15 produce by close of business today --


17 MR. STAMP: -- some -- a document. Could that period be extended

18 until tomorrow? It's a joint application, even though he's not here.

19 [Trial Chamber confers]


21 MR. STAMP: Thank you very much, Your Honour.

22 JUDGE BONOMY: Well, we'll adjourn now until 3.00 while the

23 arrangements are put in place for the next witness.

24 --- Recess taken at 2.30 p.m.

25 --- On resuming at 3.01 p.m.

Page 10442

1 JUDGE BONOMY: Please bring the witness into court.

2 [Closed session]

3 (redacted)

4 (redacted)

5 (redacted)

6 (redacted)

7 (redacted)

8 (redacted)

9 (redacted)

10 (redacted)

11 (redacted)

12 (redacted)

13 (redacted)

14 (redacted)

15 (redacted)

16 (redacted)

17 (redacted)

18 (redacted)

19 (redacted)

20 [Open session]

21 THE REGISTRAR: We are in open session, Your Honours.

22 JUDGE BONOMY: Thank you.

23 I think there's one blind not functioning there. In fact, can we

24 just leave it because it's ...

25 Ms. Kravetz.

Page 10443

1 MS. KRAVETZ: Thank you, Your Honour.


3 [Witness answered through interpreter]

4 Examination by Ms. Kravetz:

5 Q. Good afternoon, Witness. I would like to start today by handing

6 you a document and I would ask you to read this document to yourself, not

7 out loud, and if you could please confirm whether the information in this

8 document is correct.

9 A. Yes.

10 MS. KRAVETZ: Your Honours, this is P2691, and I would like to

11 tender that under seal.

12 JUDGE BONOMY: Thank you.


14 Q. Witness K88, did you provide a statement to the Office of the

15 Prosecution on 16th and 17th April 2005 and on 21st January 2006?

16 A. Yes.

17 MS. KRAVETZ: Could the witness please be handed a copy.

18 Q. Could you look at the statement and please confirm whether this

19 was the statement you provided to the Office of the Prosecution.

20 A. Yes, that's the statement.

21 Q. Thank you. Have you recently had the opportunity to review your

22 statement?

23 A. Yes, I did.

24 Q. And in the course of reviewing your statement, did you make two

25 minor corrections which you pointed out to me during our proofing session?

Page 10444

1 A. Yes.

2 Q. One correction that you pointed out in proofing was that the last

3 phrase of paragraph 3 should be deleted, starting with the words "my

4 official title was" until the end of that sentence. Is that correct?

5 A. Yes, that's correct.

6 Q. The second correction you indicated in proofing --

7 JUDGE BONOMY: Sorry. I've -- the last -- this is paragraph 3?

8 MS. KRAVETZ: This is paragraph 3 and it's the last sentence. The

9 phrase that starts with the words: "My official title was."

10 JUDGE BONOMY: Oh, yes, thank you. The last line.

11 MS. KRAVETZ: The last line.

12 Q. The second correction you made in proofing was at paragraph 14,

13 line 6 you indicated that the following phrase should be deleted: "And I

14 had modified and made it usable again." This is in the middle of

15 paragraph 14, line 6. Is that correct?

16 A. Yes.

17 Q. Thank you, K88. Other than those two minor corrections, after

18 having reviewed your statement, are you satisfied that this statement

19 accurately reflects to the best of your knowledge and recollection the

20 truth about the events that are described in the statement?

21 A. Yes.

22 Q. Thank you.

23 MS. KRAVETZ: Your Honours, I would like to tender the statement

24 under seal. This is P2691.

25 JUDGE BONOMY: Thank you.

Page 10445

1 MS. KRAVETZ: I would like to inquire with the Court whether in

2 this case we would need to provide a redacted statement. This has not yet

3 been uploaded in e-court basically because -- basically -- I'm sorry.

4 Because in the process of carrying out these redactions we noticed that

5 substantive portions would have to be redacted in order to protect this

6 witness's identity, and I'm not sure that the information that will remain

7 if we carry out these redactions will be very meaningful or useful to the

8 Court.

9 JUDGE BONOMY: Well, first of all, I think it's P2681. Is that

10 correct?

11 MS. KRAVETZ: Yes. I'm sorry.

12 JUDGE BONOMY: And I think what you should do is let the Trial

13 Chamber see the redacted version so that we can decide whether it's

14 appropriate to publish it in that form.

15 MS. KRAVETZ: Okay. Well, I'll advise the Court when this has

16 been uploaded in e-court.

17 JUDGE BONOMY: Thank you.

18 MS. KRAVETZ: I think it will be later today.

19 I would like to show the witness the attachment to his statement,

20 this is the B/C/S version, and would ask that this not be broadcasted

21 outside the courtroom. This is page 11 of his B/C/S statement. There's

22 no equivalent in the English. I see the English is being opened.

23 And when this comes up could we please rotate it to the right.

24 Q. Witness K88, I understand from your statement that this is a map

25 you drew for OTP investigators to explain the evidence that you describe

Page 10446

1 in your statement. Is that correct?

2 A. Yes.

3 Q. In paragraph 14 of your statement you refer to a truck, to a FAP

4 truck which you say that you observed at the base in early April 1999 when

5 you went there. I would like you to point out where this truck is -- has

6 been depicted on this map that you have drawn. If you could draw a circle

7 around it just for clarity.

8 A. [Marks].

9 Q. Thank you.

10 MS. KRAVETZ: Your Honours, in the English version of this

11 statement, this has incorrectly marked as 1V. It should be number 4 on

12 the map.

13 JUDGE BONOMY: On mine it's number 6 in paragraph 14. Is that

14 right? Or is it a later -- are we talking about some --

15 MS. KRAVETZ: Number --

16 JUDGE BONOMY: Oh, yes, I see.

17 MS. KRAVETZ: It's later down towards the middle of the paragraph.

18 JUDGE BONOMY: Sorry, IV just means 4.

19 MS. KRAVETZ: It just means 4.

20 Q. On this map you have drawn an arrow, a long arrow, towards the

21 right of the circle that you just made, a number 4. Could you please

22 explain what is depicted by that arrow that you have drawn?

23 A. I didn't quite understand.

24 Q. There is a line that goes to the right of number 4, which you have

25 drawn on this map, and I am just asking if you could indicate what that

Page 10447

1 line indicates.

2 A. That line shows the direction as to where the corpses from that

3 truck were buried.

4 MS. KRAVETZ: Could we go to the next page, page 12. We have to

5 give it an IC number under seal.

6 THE REGISTRAR: That will be IC124 under seal, Your Honours.

7 MS. KRAVETZ: Thank you.

8 Q. Witness K88, what is depicted here by the number 1 that you have

9 drawn towards the left of this map?

10 A. That's grave number 1, that first one from the mentioned truck.

11 Q. Thank you. Is this grave that we're looking at located outside

12 the area that you had drawn on your map, the previous page of the map?

13 A. Yes, the previous drawing was a drawing of the firing range and

14 between there is a parapet about 12 metres tall I think.

15 Q. Would this first grave then be located outside the firing range

16 area?

17 A. Yes, outside the firing range area.

18 Q. And can you explain who, if you know -- under whose control fell

19 this area where the first grave was dug?

20 A. Well, roughly we used that area, too, although - how should I put

21 this? - it was afforested at the time, or rather, it was covered with some

22 kind of bush. It wasn't really used for anything, but it was under our

23 control because after that there is a fence and then the PKB, the

24 agricultural combine farm of Belgrade; it's owned by them, that land.

25 Q. Thank you.

Page 10448

1 MS. KRAVETZ: Could we return to page 11 of the statement.

2 Q. Now, at paragraph 14 you also mentioned seeing two further trucks

3 which I understand are depicted here by 2V and 3V right under the number 4

4 you had drawn before. Is that correct?

5 A. Yes.

6 Q. At paragraph 16 you referred to these two trucks and you say that

7 you saw them and then they left the base, and you say: "I later heard

8 they also contained bodies but never saw the trucks again."

9 Could you explain from whom you heard that these trucks numbered 2

10 and 3 also contained bodies.

11 A. Well, a colleague of ours, since we had been dislocated, he toured

12 this area, this centre, from time to time and he looked a bit more closely

13 and he probably came to this conclusion on the basis of the smell, and I

14 don't know what else, that there were corpses inside.

15 Q. Did this person tell you anything about what happened to these

16 trucks after they left the base?

17 A. No. Just one moment. He probably just stopped by. I can't even

18 remember who it was from among the men who were in charge of security at

19 the time.

20 Q. Thank you. At paragraph 22 you referred to a fourth truck that

21 you saw on the base -- that are different base some days later. Would

22 this be what you have depicted as 4B to the bottom left-hand side of --

23 left-hand corner of your map?

24 A. No.

25 Q. Where would this be depicted on your map, the fourth truck?

Page 10449

1 A. The fourth truck, 4V, in the lower left-hand corner?

2 Q. Yes, I'm asking if that is where you indicated that you had seen

3 this truck, what you have depicted there in the lower left-hand corner of

4 your map.

5 A. Yes, but that was after all of this had happened.

6 Q. Okay. Would you describe the drivers of these truck -- of this

7 fourth truck as having a Montenegrin accent and you said they were dressed

8 in blue camouflage uniform, probably re service. And at paragraph 24 you

9 said that after this truck had been emptied, the contents of this truck

10 had been emptied, they had to go back to Kosovo. Could you explain how

11 did you know that these truck drivers had to return to Kosovo. Is this

12 something they told you themselves?

13 A. Well, probably they said so. I don't remember exactly now. It's

14 been a long time and a lot of things have happened, but I think they were

15 the ones who told me that they were going back.

16 Q. When you say they had to return to Kosovo, did they tell you from

17 where they had come? Had they originated -- that their journey from

18 Kosovo? Is that what they had said?

19 A. No, they didn't tell us where they had come from or where they

20 were returning. They did not refer to a specific place.

21 Q. But you say in paragraph 24 that they had to go back to Kosovo.

22 What do you mean by that statement?

23 A. Well, probably, that is where they had come from, from down there,

24 since most of the war events were taking place in Kosovo.

25 Q. And what made you assume that they had come from Kosovo?

Page 10450

1 A. Well, probably the way in which they spoke or I don't know exactly

2 what to say now.

3 Q. Okay. Thank you. At paragraph 29 of your statement you talk

4 about two further trucks, truck 5 and 6, and you describe the truck

5 drivers as being an old man in civilian clothes and two others in police

6 uniform. These two policemen that came with the fifth and sixth truck,

7 are these different than the ones you saw with the fourth truck that you

8 depicted on your map?

9 A. Yes, they were different ones.

10 Q. And you say in the same paragraph that you didn't know where they

11 were from, maybe Kosovo, they spoke Serbian. What made you think that

12 these policemen and this man with this fifth and sixth truck came also

13 from Kosovo?

14 A. Well, I don't know why I thought that. They probably said so.

15 Q. Thank you. At paragraph 30 of your statement you speak about the

16 drivers on the second time that told you that while they were on their way

17 to Belgrade, they had an accident and around 50 bodies had fallen out of

18 the truck. When you say the drivers on the second time, are you referring

19 to the drivers of truck number 5 and 6?

20 A. Well, I don't know exactly now which truck was which, roughly, but

21 it was their story, that is what they said, that that is what happened to

22 them on the way, that the truck door had opened and that they started

23 falling out, and ultimately they had a problem with the truck itself. The

24 truck broke down once it had reached our base.

25 Q. Did they explain to you why they were transporting bodies on

Page 10451

1 their -- in their truck?

2 A. No.

3 MS. KRAVETZ: If we can look now at page 11 of this map.

4 Q. Just for the sake of clarity, you have drawn a number 2, 3, 4, and

5 5 on the right-hand corner of the map. Is it correct that these depict

6 the different locations of the graves that were dug at the base, at the

7 firing range?

8 A. Yes.

9 MS. KRAVETZ: Your Honour, could we briefly go into private

10 session for a couple of questions?

11 JUDGE BONOMY: Is this to protect the security of the witness?

12 MS. KRAVETZ: Yes.

13 JUDGE BONOMY: Yes, we can.

14 MS. KRAVETZ: We'll no longer be referring to page 11.

15 [Private session]

16 (redacted)

17 (redacted)

18 (redacted)

19 (redacted)

20 (redacted)

21 (redacted)

22 (redacted)

23 (redacted)

24 (redacted)

25 (redacted)

Page 10452











11 Page 10452-10454 Private session.















Page 10455

1 (redacted)

2 (redacted)

3 [Closed session]

4 (redacted)

5 (redacted)

6 (redacted)

7 (redacted)

8 (redacted)

9 (redacted)

10 (redacted)

11 (redacted)

12 (redacted)

13 (redacted)

14 (redacted)

15 (redacted)

16 (redacted)

17 (redacted)

18 (redacted)

19 (redacted)

20 (redacted)

21 (redacted)

22 (redacted)

23 (redacted)

24 (redacted)

25 (redacted)

Page 10456

1 (redacted)

2 (redacted)

3 (redacted)

4 (redacted)

5 (redacted)

6 (redacted)

7 (redacted)

8 (redacted)

9 (redacted)

10 (redacted)

11 (redacted)

12 (redacted)

13 (redacted)

14 (redacted)

15 (redacted)

16 --- Whereupon the hearing adjourned at 3.38 p.m.,

17 to be reconvened on Thursday, the 22nd day of

18 February, 2007, at 9.00 a.m.