Tribunal Criminal Tribunal for the Former Yugoslavia

Page 25969

1 Thursday, 24 April 2008

2 [Open session]

3 [The accused entered court]

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

5 JUDGE BONOMY: Good morning, everyone. We shall continue now to

6 hear the evidence of Dr. Fruits.

7 [The witness entered court]

8 JUDGE BONOMY: Good morning, Dr. Fruits.

9 THE WITNESS: Good morning.

10 JUDGE BONOMY: We'll now continue with your examination by

11 Mr. Sepenuk.

12 Mr. Sepenuk.

13 MR. SEPENUK: Thank you, Your Honour.

14 WITNESS: ERIC FRUITS [Resumed]

15 Examination by Mr. Sepenuk: [Continued]

16 Q. Dr. Fruits, you'll see we're in a new courtroom now.

17 A. It's very nice.

18 Q. Thank you. I'm going to try, Dr. Fruits, to emulate the mastery

19 of my colleague Mr. Stamp in his deliberate and measured cadence this

20 morning. If I can do half as well as Mr. Stamp, I will have regarded

21 myself as successful.

22 A. You're doing very good so far.

23 Q. Thank you. Let's try to keep it quite measured, sir.

24 A. I will attempt that.

25 Q. Now, I do want to go back very briefly yesterday to an answer

Page 25970

1 that I don't think was as clear for the Trial Chamber as it should be,

2 and that concerned P1506, which was the comment by Dr. Ball, at page 13,

3 that there were relatively few NATO air-strikes during the period 2 to 4

4 April due to bad weather. Do you remember that?

5 A. Yes.

6 Q. All right. E said that this -- he said there were relatively few

7 NATO air-strikes and he said it included substantial peaks in Kosovo-wide

8 killings and refugee flows, and you disagreed with that observation;

9 correct, sir?

10 A. Correct. In particular, I disagreed with the statement regarding

11 the relatively few NATO air-strikes.

12 Q. Yes.

13 A. I apologise. That is reflected in, I believe, it is Exhibit 8 to

14 my report.

15 Q. And in your answer yesterday, you said there were some ten

16 air-strikes during that period; correct?

17 A. Yes.

18 Q. But what we didn't explain yesterday was how you arrived at that

19 number, and I want to go into that now very briefly. Just tell the

20 Trial Chamber, if you will, how you arrived at that number, and I believe

21 that you're going to need Exhibit 8 to your report.

22 MR. SEPENUK: The report is 3D893, Exhibit 8.

23 Is there a problem? We had it up yesterday, I believe. It's

24 3D24-0054, if that would help you. Yes, there it is.

25 Q. So now, Dr. Fruits, please explain how you arrived at the figure

Page 25971

1 of ten air-strikes in the period 2-4 April 1999.

2 A. Okay. This exhibit may be a little confusing at first. Each

3 date has two panels in it; the bottom panel, for example. On April 3rd,

4 the bottom panel is -- are the NATO air-strikes that were reported by

5 NATO on April 3rd. I also included the NATO air-strikes for the day

6 before just for comparison; and so, for example, the panel above that one

7 on April 3 - so, in other words, the top panel - is the April 2nd

8 air-strikes.

9 Q. And how many air-strikes does that show?

10 A. Well, I don't think this pointer works on the screen, but there

11 are -- if you can zoom in on that, you would see that there are six NATO

12 air-strikes on April 2nd in Kosovo.

13 If you start on the left-hand side, you will see that there is

14 three, clustered together. One is a security forces staging area; the

15 one below that says, "security staging area"; the one next to that says,

16 "security force staging area"; the over on to the right, you will see

17 there is one that says, "army garrison"; another one below that, I

18 believe, says, "airfield"; and then one at about 5.00 says, "army

19 garrison."

20 Q. So that's how you arrived at your six targets?

21 A. That would be six on April 2nd.

22 Q. And tell us how you arrived at three targets on April 3rd?

23 A. If you look at the bottom panel, which is the April 3rd -- from

24 April 3rd release by NATO, you'll see that there are three little

25 explosion pictures. Those are what NATO describes as targets, and they

Page 25972

1 are labelled "staging areas," and there are three in Kosovo there.

2 Then on the 4th, if you move over to the --

3 Q. You mean on April 4th?

4 A. On April 4th, yes. On April 4th, if you move to the next panel

5 to the right, you will see that there is one air-strike that NATO

6 reported, and that one is the triangle that's right above Skopje.

7 Q. And that, sir, is how you arrived at your ten strikes; correct?

8 A. Correct. So that would be six on April 2nd, three on April 3rd,

9 and one on April 4th.

10 Q. And Dr. Ball's data showed how many air-strikes?

11 A. If I recall correctly, I believe his data showed that there were

12 two NATO air-strikes on April 2nd and none on any of the other days.

13 Q. And I would just like you to explain to the Trial Chamber how you

14 arrived at the two air-strikes on April 2nd according to Dr. Ball?

15 A. I used data that Dr. Ball provided to the Tribunal, and I believe

16 it's listed appendix A.2 of my report, which if I recall correctly might

17 be on page 38.

18 Q. And if we can be more specific, was it on page 39, item 8 of your

19 report?

20 A. That sounds correct. If they can bring it up, I can confirm

21 that.

22 MR. SEPENUK: I'm not sure it's necessary, Your Honour, but we

23 can do it. For the record, Your Honour, it's 0324-9511. It's referred

24 to on page 39 of item 8 of the report, and that's also for the benefit,

25 Your Honour, of Mr. Stamp who might want more specific information.

Page 25973

1 JUDGE BONOMY: Thank you.

2 MR. SEPENUK:

3 Q. Let's move on, Dr. Fruits. You devoted approximately six pages

4 of your report in describing what you call the flaws in Dr. Ball's

5 migration estimates. Would you briefly summarize your opinions regarding

6 the migration data.

7 A. Yes. A very brief summary would say that Dr. Ball's migration

8 estimates do not accurately reflect where people came from or the dates

9 that they left their homes. Dr. Ball only has - well, not only has - but

10 he has information regarding when they crossed the Morina border. For

11 everyone who crossed the Morina border, Dr. Ball does not have any

12 information regarding when they left their homes. For approximately half

13 the people who crossed the Morina border, Dr. Ball doesn't have any

14 information on where they came from.

15 So, Dr. Ball must guess or estimate or what he calls impute where

16 they came from and when they left their homes. This imputation process

17 introduces substantial errors that creates some very unusual results, and

18 those unusual results are highlighted in Exhibit 4, 5, and 6 of my

19 report.

20 Q. What are those unusual results?

21 A. Well, Exhibit 4 shows the -- shows the municipality of Leposavic;

22 and according to the Morina border --

23 Q. Excuse me for interrupting you. Would it be helpful if the

24 Trial Chamber has Exhibit 4 in front of it?

25 A. Yes. It would be helpful if I had it in front of me.

Page 25974

1 MR. SEPENUK: It's 3D893, Exhibit 4, please.

2 Q. As I understand it, you went into a bit more detail on pages 16

3 and 17 of your report concerning these matters, correct?

4 A. Yes.

5 MR. SEPENUK: Okay. Now, can we turn that around, please.

6 MR. VISNJIC: Page 15 in e-court.

7 MR. SEPENUK: Can that be turned around.

8 Q. You want it blown up a bit?

9 A. Yes. Thank you.

10 The picture on the right-hand side is taken from Dr. Ball's 2007

11 report. I believe it's page 15. That's less important for the point I'm

12 trying to make here, which is, if you look over on the left-hand side

13 where the numbers are, you'll see the far left column is the date and

14 that represents Dr. Ball's two-day periods. The next column is the

15 number recorded crossing the Morina border. In other words, this is from

16 Dr. Ball's data the Morina border crossing, and it shows how many people

17 crossed the Morina border according to that data -- I apologise, were

18 recorded crossing the Morina border on that date.

19 You will see there were actually no one recorded crossing the

20 border. The next column, the third column over, says, Ball estimate of

21 number leaving their homes. This is after Dr. Ball's imputation process.

22 It indicates how many people Dr. Ball estimates left their home on a

23 given day from Leposavic to cross the Morina border, and you'll see that

24 that number totals to 514.

25 So even though no one is actually recorded crossing the Morina

Page 25975

1 border, Dr. Ball's imputation process reaches the conclusion that 514

2 people crossed the border, and he's able to somehow to identify precisely

3 what date they left their homes.

4 Q. A flaw in his data in your estimation?

5 A. I would consider that a flaw.

6 Q. And do you have, generally without going into detail or without

7 looking -- perhaps we don't need to look at the exhibit, do you have the

8 same comment, roughly speaking, with respect to Novo Brdo?

9 A. Yes, that is true. Then Exhibit 6 shows the differences between

10 the number crossing the border in Dr. Ball's estimate for every

11 municipality.

12 Q. And what effect do these flaws you've cited have on Dr. Ball's

13 statistical analysis in your opinion?

14 A. Well, they suggest that his imputation process generates

15 unreliable estimates of migration, and those unreliable estimates really

16 can't be used to generate any sort of meaningful conclusions regarding

17 the effects of anything on migration.

18 Q. Okay, Dr. Fruits. I want to turn now to the information that a

19 statistician would need to properly analyse deaths during the Kosovo war,

20 and you've devoted approximately seven pages of your report to describing

21 the alleged flaws in Dr. Ball's death estimates, and that's pages 18

22 to 21 and 32 to 33. Would you please summarize your conclusions.

23 A. Yes. Again, as with the migration, Dr. Ball's death estimates do

24 not accurately reflect or likely do not accurately reflect the number of

25 people who were killed or where they were killed. He has several wildly

Page 25976

1 different estimates of the number of deaths. One of them, the largest

2 number, is -- is statistically significantly larger than two other

3 independent studies of the number of deaths. Then he also -- he creates,

4 he generates a methodological mess, where he mixes up the number of

5 deaths from two different estimates and puts them together into a single

6 series. I can go into that in more detail, if you like.

7 Q. Just give us another couple of sentences, please.

8 A. Sure. Dr. Ball has a number of recorded deaths which total

9 4.400, 4.400 deaths. Out of those 4.400 deaths, 200 of them do not have

10 a definite date of death; in other words, it's not known when they died.

11 For those people, Dr. Ball randomly assigns them a date of death. The

12 result of that is that in some cases, there are some municipalities that

13 have no recorded deaths; but after Dr. Ball's what he calls hotdecking or

14 I would call a shuffling procedure, he assigns deaths to municipalities

15 that may not have had a death on a certain date.

16 From that 4.400 number, he uses another technique that inflates

17 that number to 10.356 deaths. He indicates in his report that that

18 number is consistent with three other studies. I reviewed those studies.

19 Two of them actually covered a much longer time-period than Dr. Ball's

20 study; and if you adjust for those -- for the differences in

21 time-periods, I found that Dr. Ball's estimates were actually

22 significantly higher than those two other independent studies. The third

23 study was conducted by Dr. Ball himself, so it's unsurprising that the

24 results would be consistent with the study that he presented here.

25 The last issue is that Dr. Ball takes some of the numbers --

Page 25977

1 creates a third series of deaths that he uses for his linear regression

2 analysis, and that third series is created out of some of the deaths on a

3 given day from the 10.356 series and some deaths from the 4.400 series.

4 He puts the two together and then creates what I call a mish-mash series,

5 which really is a meaningless count. It's impossible to determine from

6 that any sort of impact on -- of any other explanatory variables on the

7 number of deaths.

8 Q. So, in your opinion, Dr. Fruits, do the migration or do the death

9 data provides the kind of reliable information necessary to identify the

10 causes of migration and deaths?

11 A. No.

12 Q. And with respect to all of the variables that have been

13 mentioned, whether it's the action of Serb forces, NATO action,

14 KLA action, migration figures, death figures, do any of these variables,

15 any of them, provide the kind of reliable data necessary, in your

16 opinion, for an adequate statistical analysis?

17 A. No.

18 MR. SEPENUK: That's all I have, Your Honour.

19 JUDGE BONOMY: Thank you, Mr. Sepenuk.

20 Dr. Fruits, you'll now be cross-examined by the Prosecutor,

21 Mr. Stamp.

22 MR. STAMP: Thank you very much, Your Honour.

23 Cross-examination by Mr. Stamp:

24 Q. Good morning, Dr. Fruits.

25 A. Good morning.

Page 25978

1 Q. Now, this case concerns demographic techniques applicable to

2 making estimates -- or not this case. Dr. Ball et al report concerns

3 applying certain demographic techniques to make estimates in a complex

4 humanitarian crisis. Have you ever published anything in that field?

5 A. In what field are you speaking?

6 Q. In the field of demographic? Let's start with that.

7 A. I've written several reports that involve using demographic data.

8 Q. Have you ever published any research or study that has demography

9 as its specific area of research?

10 A. I've published one article that uses demographic information to

11 estimate government spending.

12 Q. And that is in your --

13 A. That is in my CV, I believe.

14 Q. A part of your -- would you call yourself a demographer?

15 A. I would not call myself a demographer.

16 Q. Certainly not a demographic expert?

17 A. Well, I use demographic data. I don't know what constitutes an

18 expert.

19 Q. I do, too. You are aware that Dr. Ball has published a lot of

20 work involving demographic estimates relating to complex humanitarian

21 crises or disasters. Are you aware of those publications?

22 A. I know he's published a few. I'm not exactly sure of -- I can't

23 give you a list of what it is.

24 Q. And Fritz Scheuren, who is also an author, and as a matter of

25 fact, he wrote the statistical part of Dr. Ball's report. You're aware

Page 25979

1 of his publications, as well?

2 A. I am not.

3 Q. Are you aware that he's a past president of the American

4 Statistical Association?

5 A. I'm not aware of that.

6 Q. Are you a member of that?

7 A. No.

8 Q. Well, you use some terms, in your report here, critiquing their

9 work, terms like the use of figures concoct. You refer to some of their

10 methodologies as nonsensical. At one stage, I see you say they had ceded

11 common sense. So, you set that their presentation aside. That type of

12 language, is that what you consider to be the appropriate language in

13 expressing your disagreement or disputing matters in your field or in

14 your discipline?

15 A. Yes.

16 Q. Very well. You say that Dr. Ball uses an estimation approach to

17 inflate the number of KLA deaths -- sorry, of Kosovar Albanian deaths.

18 Dr. Ball, in his report - and I think it was Dr. Scheuren who wrote this

19 part of the report - indicated that they had data on Kosovar Albanian

20 deaths from various sources. Now, would you agree with me that not every

21 death or person who migrates out of the country would be recorded by

22 persons working in the field during this humanitarian crisis?

23 A. I think that would be a fair statement.

24 Q. And would you agree with me that there is a large body of work by

25 persons who specialise in the area of demographic studies of complex

Page 25980

1 humanitarian crises as to how you can make accurate estimates of deaths

2 or migration from what are certainly incomplete surveys taken in the

3 field?

4 A. I would agree that there's a large body of work, but my

5 understanding of that large body of work is that there's still a large

6 degree of uncertainty about the appropriate ways to go about estimating.

7 Q. The estimate that Dr. Ball or the methodology that Dr. Ball used,

8 Dr. Ball and the others used, to make an estimate of the number of

9 deaths, that is called the multiple systems estimation procedure. Are

10 you familiar with it?

11 A. I'm familiar with the basics of it.

12 Q. Have you ever written about it?

13 A. No.

14 MR. ACKERMAN: Excuse me, Your Honour.

15 JUDGE BONOMY: Mr. Ackerman.

16 MR. ACKERMAN: Your Honour, I'm wondering about the relevance of

17 this whole area of inquiry. This is not a meeting of humanitarian

18 demographers, where we're trying to figure out ways to estimate deaths.

19 This is a court of law, and the job of the Prosecutor is to prove these

20 deaths beyond a reasonable doubt. I seriously doubt that Your Honours

21 are going to be impressed with some estimate made by some demographer and

22 say that's how many people were killed and the Prosecution has,

23 therefore, proved it. So I think it's completely irrelevant.

24 MR. STAMP: Your Honour --

25 JUDGE BONOMY: With respect, Mr. Ackerman, I think that misses

Page 25981

1 the mark. This evidence is not about proving how many people were killed

2 or how many moved. This evidence is about the cause of death and

3 movement, and it seems to us that it is relevant to pursue the line that

4 Mr. Stamp is pursuing.

5 MR. ACKERMAN: Well, I would have no problem if that were the

6 line, but what he's talking about is the estimation of the number of

7 deaths. He's not talking about movements at all.

8 JUDGE BONOMY: But it's crucial to the ultimate conclusions that

9 are reached by Ball.

10 So please continue, Mr. Stamp.

11 MR. STAMP: Thank you, Your Honours.

12 Q. So you have a basic knowledge of multiple system estimation, the

13 capture/recapture method?

14 A. Yes.

15 Q. Isn't it, at least your basic knowledge, is that this is a

16 methodology that has been around for over a hundred years in statistics?

17 A. I know it's been around for a while. I'm not sure exactly how

18 many years.

19 Q. Now, where two separate estimates have overlapping confidence

20 intervals, does that indicate that they are statistically consistent?

21 A. Not necessarily.

22 Q. If somebody estimates a number at 45, three more or less, and

23 somebody else estimates a number at 47, isn't that statistically

24 consistent in your view? The margin of error overlaps with the other

25 number?

Page 25982

1 A. Oh, in that case, yes, that's true.

2 Q. In your report at section 8.2, you used a proportion, 60 per

3 cent, to calculate the period or the proportion of deaths in Spiegel's

4 report that would cover the period March to June.

5 A. Is that a question?

6 Q. Yes. 60 per cent is the number you used?

7 A. I don't have my document in front of me right now, but I will

8 accept that as correct.

9 Q. Would you like a copy of it? You said that Dr. Ball's estimate

10 is not consistent with Dr. Spiegel and Dr. Salama's estimate, and you

11 said that Dr. Salama's -- Dr. Spiegel's estimate is for a longer period.

12 But you calculated that for the period March to June, it was 60 per cent

13 of his numbers?

14 A. Yes. I think there may be some confusion in how that 60 per cent

15 was arrived at. If I could see the next page of the -- of my report.

16 MR. SEPENUK: Excuse me, Your Honour. As I understand it,

17 Dr. Fruits certainly can have his report with him under these

18 circumstances.

19 You can retrieve your report, Dr. Fruits.

20 MR. STAMP: Yes. I was looking for one. Thank you very much,

21 Counsel.

22 MR. SEPENUK: Do you have a page number you're referring to,

23 Mr. Stamp?

24 MR. STAMP: 8.2 is the section and it's page 20.

25 THE WITNESS: I don't think I understand your question regarding

Page 25983

1 the 60 per cent, sir.

2 MR. STAMP:

3 Q. You said the proportion -- you said: "Approximately 60 per cent

4 of Spiegel and Salama's estimated deaths occurred in March to June of

5 1999."

6 A. That is correct.

7 Q. And the footnote indicates where you got the figure from.

8 A. Yes.

9 Q. You calculated this from a graph?

10 A. Yes.

11 Q. If I tell you that the number was more approximately 65 per cent,

12 would you agree with me?

13 A. I don't think it makes much of a difference.

14 Q. So would you be able to agree with that, that it's possible?

15 A. It's possible.

16 Q. Very well. But let's work with your 60 per cent for the time

17 being. Spiegel and Salama's estimate had a wide margin -- confidence

18 interval, let's call it that. Do you recall it was 5.500 to 18.300.

19 A. I'll accept that.

20 Q. Well, did you bring --

21 A. I did not bring that article with me.

22 Q. Okay. I have it here.

23 A. Oh, actually I might. Yes, I do have it here. I apologise.

24 Q. Findings at the first page of it, would you agree with me that

25 their findings were that: "The survey included 11.000 -- sorry, 1.197

Page 25984

1 households comprising 8.605 people. From February 1998 through June

2 1999, 67 persons of the 105 deaths in the sample population were

3 attributed to war-related trauma, corresponding to 12.000 (95 percent

4 confidence intervals 5.500 to 18.300) deaths in the total population."

5 So their confidence interval was 5.500, 18.300?

6 MR. SEPENUK: I just can't resist this, Your Honour. Could

7 Mr. Stamp speak just a little more slowly, please.

8 JUDGE BONOMY: That's also repelled.

9 Please continue, Mr. Stamp.

10 THE WITNESS: I can't debate that's what it says; however, that's

11 not relevant to the time-period which we're discussing.

12 MR. STAMP:

13 Q. Just answer my questions, if you don't mind. You have counsel --

14 well, counsel is there who will ask later --

15 A. Okay. Then, yes, I agree, that's what the findings say.

16 Q. Yes. So if you apply 60 per cent to that confidence interval,

17 you get 3.300 to 10.980. Would you agree with that?

18 A. I will accept that. I can't do the math without--

19 Q. Would like a calculator?

20 A. Only if I really need one.

21 Q. Well --

22 A. I'll accept your number, rather than spend time calculating.

23 Q. Now, the 10.980 being the upper level of his margin of error of

24 confidence interval covers Dr. Ball's figure of 10.356, does it not?

25 A. I'm sorry. What was that number you said again? Oh, yeah, I can

Page 25985

1 see it. 10.980.

2 Q. Yes.

3 A. Yes, that would.

4 Q. In fact, Dr. Ball's margin of error and Spiegel and Salama's

5 margin of error overlap quite significantly. Would you like me to remind

6 you of what the margins of error were?

7 A. No. I don't understand what you mean by quite significantly. I

8 will agree that they overlap.

9 Q. So I suggest to you that using your number, 60 per cent, applied

10 to Spiegel and Salama's figure demonstrates that Dr. Ball's estimate is

11 statistically consistent. His estimate of 10.356 is statistically

12 consistent with Spiegel and Salama's --

13 A. I disagree with that.

14 Q. -- figure. Well, you tell us that Dr. Ball's figure falls within

15 their margin of error, so on what basis would you disagree with that?

16 A. I think you're misrepresenting what I said.

17 Q. I asked you the 10.980, being the upper level of his margin of

18 error or confidence interval, covers Dr. Ball's figure of 10.356, does it

19 not?

20 A. Yes, I agree with that.

21 Q. Now, having regard to what you said then, wouldn't you agree with

22 me that Dr. Ball's figure is statistically consistent with Dr. Spiegel's

23 finding?

24 A. No, I do not.

25 Q. On what basis?

Page 25986

1 A. I don't believe that your calculation of the 10.980 is actually a

2 statistically correct calculation.

3 Q. Well, I'm just using your methodology.

4 A. No, you weren't.

5 Q. Well, you said -- well, tell me what's the difference.

6 A. I applied my methodology to the estimate of the mean; you applied

7 it to the estimate of the variants or the variation.

8 Q. Well, you're telling me that there is a statistically significant

9 difference in that?

10 A. I'm telling you there is a methodological difference in that.

11 Q. Well, let me ask you this: Does Dr. Spiegel's confidence

12 interval have any meaning; and if so, what it is?

13 A. The confidence interval that's reported here in the findings

14 don't have any meaning with respect to the conclusions that I reached

15 here.

16 Q. Well, does it mean that he's saying that the possible figure or a

17 possible estimate falls within that range?

18 A. I don't understand the question.

19 Q. Is he saying that his estimation is there were 12.000, but he's

20 95 per cent certain that the number, the proper estimate for the number

21 of people killed, falls within the range that he has given?

22 A. No. That's an incorrect characterization of confidence interval.

23 Q. Well, let me ask you this: What is survivor bias in

24 demographics?

25 A. Survivor bias can have several different meanings depending on

Page 25987

1 what the context --

2 Q. What is survivor bias in respect to Dr. Spiegel's study?

3 A. Survivor bias, in this case, probably refers to -- actually, I

4 don't know what it means in this case.

5 Q. Okay. You're not really an expert in this field, are you?

6 A. I'm not sure what field you refer to.

7 Q. Applying statistics to demographical issues involving complex

8 humanitarian crises.

9 A. I would consider myself an expert now. I've spent two years

10 doing this.

11 Q. Your preparation for the case -- well, Dr. Spiegel and Dr. Salama

12 tells us, and I'll read from the report.

13 "Villages with fewer than 100 individuals were not available for

14 selection for a survey, and there is evidence that the more rural

15 isolated areas were targeted by Serbian forces. Furthermore, only

16 households present on the day of the survey were sampled. These factors

17 could have biassed the study since households in which all members had

18 died during the war would not have been selected resulting in an

19 underreporting of deaths, survivor bias."

20 It's really a standard methodological approach in this area, in

21 this subdiscipline of demographics. I go on.

22 "Both these facts would have led to an underestimation of

23 mortality. And, lastly, the small number of deaths and missing people in

24 our study has resulted in a wide confidence interval."

25 Firstly, where you make adverse comments using the strong

Page 25988

1 language that you do in comparing Dr. Ball's et al.'s report and the

2 estimate to Spiegel and saying that his estimate inflates the number,

3 don't you think that, as an expert, you should have made the Court aware

4 that Spiegel and Salama said that for methodological reasons their number

5 is likely to be an underestimate? Simple question: Don't you think you

6 should have told the Court that?

7 A. I cited the document in several places.

8 Q. Don't you think you should have told the Court, when you were

9 making the comparison and criticising Dr. Ball in strong language about

10 inflating numbers, that Spiegel and Salama said that their number is

11 likely to be an underestimate?

12 A. No. I think that underestimate is covered by the confidence

13 interval that they provide in the findings.

14 Q. But don't you think, in that case, you should have told the Court

15 what the confidence interval was?

16 A. No.

17 Q. Very well. And I ask you this, this is a question, because later

18 on people will review what you say here and might make comments about it.

19 You weren't trying to hoodwink the Court, were you?

20 A. Hoodwink? No.

21 Q. Very well.

22 MR. SEPENUK: Well, what, Mr. Stamp? Well, what?

23 MR. STAMP: I said very well. I accept your answer. I hear it.

24 MR. SEPENUK: Thank you, sir. I think what you're saying borders

25 on the improper. That's all I'll say at this point.

Page 25989

1 MR. STAMP: I won't engage in an exchange with counsel.

2 JUDGE BONOMY: Now, Mr. Sepenuk, you know that these comments

3 should be directed through the Court, not across the courtroom.

4 MR. SEPENUK: I apologise, Your Honour.

5 JUDGE BONOMY: Let's calm down and continue, please.

6 Mr. Stamp.

7 MR. STAMP: Very well.

8 Q. You didn't think it was important to tell us that.

9 A. Is that a question?

10 Q. No. I'm coming to the question.

11 So you have a basic knowledge of the capture/recapture method.

12 MR. STAMP: Can I show a document, P3154. I apologise. This one

13 was given to counsel. The notification was a little bit late.

14 MR. SEPENUK: Excuse me. Is this the article on forced

15 migration --

16 MR. STAMP: And mortality.

17 MR. SEPENUK: Yeah, three-page article. Is that correct?

18 MR. STAMP: No, not a three-page article. We just have the front

19 page, the first page of that.

20 MR. SEPENUK: I have, Your Honour, P3774. Is that the wrong

21 number?

22 MR. STAMP: I'm sorry. May I just explain. There is a number to

23 the top right hand, P3774. That is a number that was used for this

24 document in a previous case, so it's written on the document. The

25 correct document in this case -- the correct exhibit number in this case

Page 25990

1 is P3154.

2 MR. SEPENUK: Excuse me, Your Honour. Is that a three-page

3 document, Mr. Stamp?

4 MR. STAMP: Yes, it is.

5 MR. SEPENUK: I received this last night at about 10.00. I

6 simply wanted the Court to know that.

7 MR. STAMP: Very well. If counsel is objecting to the use of

8 this, I'm not sure it's necessary, but I don't know what the comment is.

9 MR. SEPENUK: Well --

10 JUDGE BONOMY: He's not objecting, Mr. Stamp.

11 MR. STAMP: Very well.

12 JUDGE BONOMY: So --

13 MR. SEPENUK: I'm objecting, Your Honour, only to the extent that

14 if he doesn't think it's necessary, I'm not sure why we need it,

15 particularly when we saw it at 10.00 last night. I leave it to Your

16 Honour's discretion.

17 JUDGE BONOMY: I thought you were just setting up the position

18 that if anything emerges, then you might want to raise an issue, but at

19 the moment you weren't taking objection to its use.

20 MR. SEPENUK: Again, with Mr. Stamp's comment that he doesn't

21 think it's necessary, I wonder if we should proceed with it, particularly

22 when we saw it so late last night. Again, I would leave that to Your

23 Honour's discretion.

24 JUDGE BONOMY: What do you want to do, Mr. Stamp?

25 MR. STAMP: I really don't need to use it, so I --

Page 25991

1 JUDGE BONOMY: I mean, we can cope with the use of documents like

2 this, but as is so often the case it looks as though you're going to turn

3 to matters that don't actually require the use of this document.

4 MR. STAMP: I don't think it does, actually.

5 JUDGE BONOMY: Well, please proceed, and only if you really need

6 to refer to it, we'll come back and consider the issue.

7 MR. STAMP:

8 Q. Let's move on to your migration estimates or your critique of the

9 migration estimates. You see that section 7.1 of your report, and there

10 are some inconsistencies in the estimates. What is the inconsistency

11 that concerns you, because I see figures here ranging from 276.000 to

12 433.000.

13 A. Again, I'm sorry. I don't understand what your question is.

14 Q. Well, you have a section here entitled: Inconsistency in the

15 estimate. Do you have a complaint about these inconsistencies?

16 A. It's an observation that there -- that Dr. Ball has a wide

17 variety of numbers. One is 276.500 crossing the border.

18 Q. Right.

19 A. That number is then boosted by approximately 50 per cent to

20 404.000, and that's the total number again crossing the Morina border

21 that he estimates. Then when he performs his linear regression analysis,

22 for some reason that number drops to 352.700.

23 Q. Well, Dr. Ball's estimate of 404.000 is, well, pretty much

24 consistent with NATO's estimate of 433.000.

25 A. Well, actually, they're measuring two different things.

Page 25992

1 Q. But you put them side by side in your report.

2 A. I think I'm explicit in what the NATO number represents.

3 Q. Do you -- you said that for some reason Dr. Ball drops the figure

4 to 352.700. Did you discern what reason that was?

5 A. No. That's one of the problems I have with that.

6 Q. But didn't Dr. Ball in his report explain, firstly, that the

7 276.500 individuals that crossed the border, that the border records had

8 was incomplete, and that more persons crossed the border than what the --

9 than what were recorded. Many persons were not recorded. Would you

10 agree with that?

11 A. Yes. I recall Dr. Ball saying that.

12 Q. And that what he did, along with his colleagues, was to use the

13 surveys conducted by the Organization for Security and Cooperation

14 Europe, Physicians for Human Rights, Human Rights Watch, and other bodies

15 that were surveying the refugees in Albania to come to a statistical

16 estimation of the amount that crossed the border?

17 A. I think that's a mis-characterization of what Dr. Ball says he

18 did in his reports and testimony.

19 Q. What did he say he did?

20 A. I believe that according to his report Dr. Ball indicates that he

21 used information from the UNHCR and an organization known as the EMP to

22 take the 276.500 number and take that up to 404.000. I believe the

23 surveys from Human Rights Watch and others were actually used for a

24 different purpose.

25 Q. The ECG -- or the EMG is an Albanian government emergency

Page 25993

1 management group?

2 A. Is that a question?

3 Q. Yes.

4 A. I don't recall precisely. I think that sounds approximately

5 right.

6 MR. STAMP: May I just have a moment, Your Honours.

7 Q. To save time, I'll get back to that. The point I want to make

8 is: Are you aware that the number that Dr. Ball uses of 352.700, he gets

9 down to that because when he uses the surveys of persons who crossed the

10 border and who were in Albania in the camps, there was an indication that

11 some of them had crossed before the NATO intervention; in other words, he

12 used the surveys to come to an estimation, a statistical estimation, of

13 those who had crossed before. Do you recall that?

14 A. I don't recall.

15 Q. But you will agree with me that the estimate of 404.000 would

16 possibly include the persons who had crossed before the relevant

17 time-period?

18 A. That's a possibility.

19 Q. And would you agree with me that Dr. Ball would have had to use a

20 methodology to discover what proportion of that figure of those persons

21 had crossed the border before?

22 A. Yes, I would agree with that.

23 Q. Do you recall that he used --

24 [French on the English channel]

25 A. I'm sorry. Can you repeat that? I think my headphone went out.

Page 25994

1 Q. It's on now?

2 A. It's on now.

3 Q. You don't recall that he used surveys that were done in the camps

4 in Albania as a basis to make that estimation?

5 A. I'm taking -- I have a hard time with questions that are

6 negatives. I recall that he did use surveys, so I guess the answer is

7 no, I don't recall. So, yes, I do recall that he did use surveys.

8 Q. From the camps?

9 A. From the camps.

10 Q. Right. So to come to his figure, it was not just the EMG and the

11 UNHCR count and the border count. He also used surveys to come to adjust

12 the raw figure. Well, you have agreed to that.

13 This brings me to section 11.4.1 of your report, and I think you

14 testified about that yesterday. The point you are making and which you

15 have said yesterday is that more deaths were recorded on the 6th of April

16 than the two days preceding or following. In 64, you said only records

17 that can identify the date of death were used.

18 MR. SEPENUK: Excuse me, 64?

19 MR. STAMP: Footnote 64.

20 MR. SEPENUK: On what page are we talking about?

21 MR. STAMP: On 28, this is section 11.4.1.

22 Q. You used the raw figures of -- with identifiable data of deaths;

23 correct?

24 A. I'm not sure what you mean by "raw".

25 Q. The figures from Dr. Ball's data set or from the information that

Page 25995

1 you found online --

2 A. I used --

3 Q. -- that just had identifiable figures or identifiable dates.

4 A. Yes. I wouldn't call that raw. I would call those -- those were

5 what Dr. Ball characterized as the recorded deaths. And for some of

6 those deaths, there were -- the dates of deaths were not identifiable.

7 For those that weren't identifiable, Dr. Ball randomly assigned a date of

8 death to those. But it's very --

9 Q. You are going on --

10 A. Am I allowed to finish?

11 Q. No, because that is not what I asked you. Just answer what I

12 asked you, please. Dr. Ball used methodologies which you don't agree

13 with. You say it's random. He describes it, or Dr. Scheuren, the former

14 president of the statistical association, described it in the appendices,

15 to come to an estimated adjustment of the correct figures for the deaths

16 for those days, the two days preceding the 6th and 7th of April and the

17 two days succeeding the 6th and 7th of April, didn't they?

18 A. I'm sorry. I don't understand. I don't understand what you mean

19 estimated adjustment of the correct figures.

20 Q. Yes, thank you for that. Dr. Ball made an adjustment of the

21 recorded figures and used those figures in his assessment of the number

22 of deaths for those days, didn't he?

23 A. You're talking about the random assignment?

24 Q. Well, you call it random. I call it methodological. It's

25 probably outside of your field. All I want to know is this: That your

Page 25996

1 figures make no adjustments to the recorded dates of death at all;

2 whereas, Dr. Ball, as he has explained, makes adjustments to the recorded

3 dates of death; is that right?

4 A. Oh, that's correct, yes.

5 Q. Right. Would you agree with me that in conducting surveys or

6 when you review surveys that records figures in complex humanitarian

7 disasters, you have to apply statistical methodologies to adjust the

8 figures? Would you agree with that?

9 A. I think I make it clear on page 19 of my report with the

10 paragraph that begins: "Imputation procedures ..."

11 Would you like me to read that?

12 Q. Would you agree that you ought to adjust the raw figures from the

13 field because they are incomplete, necessarily?

14 A. Well, as I state in my report, imputation procedures such as

15 these may be useful in some circumstances, especially those in which the

16 imputation they control for biases associated with sample or

17 self=selection.

18 Q. Very well. I didn't ask you to read from your report. That

19 means you agree with me?

20 A. I think what I said speaks for itself.

21 Q. Very well. You did no such adjustment, did you? You just used

22 the recorded dates which you know are incomplete and unlikely to be

23 correct?

24 A. Are you indicating that Dr. Ball's recorded dates are incorrect?

25 Q. No, no. You when you presented here at page 28 and in the text

Page 25997

1 related to paragraph 64 that you made no adjustments to the recorded

2 dates of death. You just used them even though you knew --

3 MR. SEPENUK: Excuse me, Your Honour --

4 MR. STAMP:

5 Q. -- that they are incomplete and wrong.

6 MR. SEPENUK: Did Mr. Stamp mean paragraph 64?

7 MR. STAMP: Footnote 64, Counsel.

8 JUDGE BONOMY: Footnote 64, paragraph 11.4.1, and I think the

9 question actually relates to an adjustment of the numbers of deaths

10 rather than the dates.

11 MR. STAMP: Deaths, that's correct.

12 JUDGE BONOMY: Although, it's put as adjusted the dates. So

13 perhaps you should put the question again.

14 MR. STAMP:

15 Q. In your testimony today and in this passage where you said that

16 the figures or the figures from Dr. Ball's show that that the deaths were

17 lower for the 6th to 7th of April and lower on the two days after, you

18 are using figures that you, even with your basic knowledge of demographic

19 statistics in humanitarian disasters, know would be incomplete and wrong.

20 You would have to apply methodology to adjust them.

21 A. I don't think it's entirely clear that you must apply methodology

22 to adjust them.

23 Q. Well, do you -- do you know how to adjust these figures, having

24 regard to the survey data?

25 A. Well, in a graduate level econometrics course that I taught, one

Page 25998

1 of the issues you can deal with -- one of the ways you can deal with an

2 issue such as this --

3 Q. Can you answer yes or no, do you know, using a survey data, how

4 to adjust these figures?

5 MR. SEPENUK: No, I think that's part of his answer, Your Honour.

6 THE WITNESS: That is part of my answer.

7 JUDGE BONOMY: Indeed, I agree, Mr. Sepenuk.

8 Please continue with your answer.

9 THE WITNESS: That there are several ways that you can adjust for

10 missing observations. One is to simply ignore those observations. If it

11 can be demonstrated that there's no statistically significant difference

12 between the observations that have no date of death recorded and those

13 that do, then you can -- you're not going to introduce any errors by

14 excluding those observations. If, on the other hand, there is a

15 systematic difference between those observations, then, once you would

16 engage in some sort of statistical exercise to put those back in and make

17 those judgements, the hotdecking procedure simply randomly assigns them.

18 MR. STAMP:

19 Q. Well, with respect, you say the hotdecking procedure randomly

20 assigns them. You, as a statistician with your basic knowledge of the

21 field, were able or could have done a statistical model to adjust them in

22 a non-random way, couldn't you, just to test what Dr. Ball had done.

23 A. It could be done. There's other ways of doing it.

24 Q. But the you didn't.

25 A. I'm sorry. I didn't finish my answer, and I've lost my train of

Page 25999

1 thought.

2 Q. You didn't do a model to come to an adjustment, you didn't do a

3 statistical model to come to an adjustment?

4 A. No. I didn't another accepted technique which is to simply throw

5 out those observations.

6 Q. Did you -- well, I can accept that you simply threw them out

7 without applying --

8 A. I did not include them.

9 Q. Very well.

10 A. As a follow-up to the previous question, I believe the term

11 "randomly assigned" was actually used by Dr. Ball in his report.

12 Q. But you are using it here -- well, I won't get into any argument

13 about that. I think you're -- well, what I think at this point doesn't

14 matter. I just want to find out because I don't know if I got an answer.

15 You did a review of the surveys from the various organizations on the

16 refugee population in Albania, did you?

17 A. Those materials were not made available to me.

18 Q. How long have you been involved in --

19 A. Approximately two years.

20 Q. Sorry?

21 A. Approximately two years on this project.

22 Q. Well, I must confess that when I came on this case shortly before

23 it started, well, I was told that, in fact, your counsel had received

24 this material, but -- very well, you didn't review the material.

25 But you had the data sets, because the data sets are online, that

Page 26000

1 Dr. Ball and various people generated from this material?

2 A. That's correct, and they were also provided by the OTP.

3 Q. Okay. Now, it's -- isn't it a matter of putting these data sets

4 into a computer, devising a model to correct what you know are incomplete

5 figures, and come and say, On the basis of my own statistical

6 methodology, the adjusted figures would be so and so. Couldn't you do

7 that?

8 A. I don't understand the question. It would be useful if I knew

9 what you were referring to.

10 Q. I'm referring to the same figures where you used the -- only the

11 identified records of death, and I'm just trying to find out if you could

12 have formulated a scientific model to correct the figures. I just want

13 to know that. Could you?

14 A. Oh, I could have.

15 Q. I suggest to you that you improperly used figures that you knew

16 were wrong. You knew that this data had to be adjusted.

17 A. I think I already answered that question.

18 Q. Very well.

19 MR. STAMP: Your Honours, I don't propose to be long, but there

20 are some documents I wonder if I could hand to him -- well, at the

21 break -- I'll get to that.

22 Very well, Your Honours, I'll proceed and then I'll deal with a

23 matter at the break.

24 Q. Not only have you used these figures, the incomplete figures, to

25 draw these conclusions; but at section 7.4 of your report, you also

Page 26001

1 criticise Ball et al., as far as the migration estimates are concerned,

2 and I think you did that again today by saying: "Well, no one from

3 Leposavic or Novo Brdo was recorded as crossing the Morina border."

4 Are you saying, Doctor, that because no one from Leposavic or

5 Novo Brdo were recorded or was recorded as crossing the border, that, in

6 fact, no one did cross -- from those two municipalities did cross the

7 border?

8 A. I think I characterized it as "unlikely." If we read the section

9 heading, that says Dr. Ball's approach produces unlikely results.

10 Q. Are you saying that because the border crossing records, which

11 you admit would necessarily in a disaster environment be very incomplete,

12 do not record anyone from these two municipalities have -- as having

13 crossed the border, that it means that no one did? Is that what you're

14 saying?

15 A. Well, there is certainly a possibility that someone came from

16 there.

17 Q. Of course.

18 JUDGE BONOMY: You would have to have some basis for thinking at

19 least that that may have happened.

20 MR. STAMP: Yes.

21 Q. Now, Dr. Ball --

22 JUDGE BONOMY: So where is that, Mr. Stamp?

23 MR. STAMP: That would be --

24 JUDGE BONOMY: To use smoke and mirrors with figures and say,

25 well, somebody may have come from that municipality as well, there's got

Page 26002

1 to be some basis for it.

2 MR. STAMP:

3 Q. Doctor, this would be part of the disclosed data sets from the --

4 MR. STAMP: May I just ask the witness, instead of probably

5 testifying myself? I don't know if I could approach it that way.

6 THE WITNESS: I could use a break if he wants to testify for me.

7 JUDGE BONOMY: There's no need for that to happen, Dr. Fruits,

8 no. Mr. Stamp will just have to guide you through it to produce the

9 information he wishes.

10 MR. STAMP:

11 Q. Didn't Ball et al. use the surveys from the camps in Albania to

12 adjust the statistics because those surveys indicated that some, not

13 many, 513 and 36 out of 400.000 people had crossed the border from those

14 two municipalities -- withdrawn.

15 Let me ask you this: Did Dr. Ball use the surveys from the camps

16 in Albania to adjust the figures in respect to the persons who crossed

17 the border and didn't those figures indicated that some, quite a few

18 people, from those municipalities cross the border even though they were

19 not recorded at the border posts?

20 A. I don't know what the surveys say. I can only infer that the

21 surveys may have indicated that someone came from those.

22 Q. There, he says 513 from Leposavic and 36 from Novo Brdo, that's

23 his estimate, out of 400.000 people. Is there something in those figures

24 that would cause you concern? Is it shocking that 513 people or unlikely

25 that 513 people out of so many would have crossed from those

Page 26003

1 municipalities?

2 A. You can think of it this way: There were 276.500 people who were

3 recorded at the Morina border crossing. Dr. Ball's policy and panic

4 report indicates that 69 per cent of those had their municipality of

5 origin listed. It's surprising that 69 per cent times 276.500, out of

6 that, those people, say 150.000 people, then not a single person, not one

7 single person, was identified as coming from Leposavic.

8 Q. But this is precisely why you need the surveys from the camps,

9 don't you, because when you're dealing with so few people there might be

10 systematic differences. For example, just to give you an example, the 36

11 people from Leposavic -- or Novo Brdo may have come on one bus, one day,

12 during a period when there was a horde. There are many possibilities why

13 these would not be recorded?

14 A. Is that a question?

15 Q. Yes. There are many reasons why such small figures might not

16 appear in the border records.

17 A. Well, I don't know about that. Well, let me give the

18 interpreters a chance. I think your example may be at odds with what

19 Dr. Ball indicated because, if I recall correctly, Dr. Ball indicated

20 that everyone who came in a motor vehicle was recorded.

21 Q. No, no.

22 JUDGE BONOMY: What Mr. Stamp appears to have in mind is a very,

23 very big bus.

24 MR. STAMP: Thirty-six people. Well, we could reflect on the

25 evidence where people were stuffed in, but I was referring to 36.

Page 26004

1 JUDGE BONOMY: Not the 513?

2 MR. STAMP: Not in one bus, not in a bus, but perhaps on a train.

3 THE WITNESS: Well, there's a --

4 MR. STAMP: We had some evidence of that.

5 Q. Anyway --

6 A. I would like to back up for a second because I think you

7 characterized the number of people crossing the border as a relatively

8 small sample.

9 Q. Not as a sample, I didn't do that. I didn't call it a sample.

10 But, anyway, may I get on with my question --

11 MR. SEPENUK: Excuse me. Did Dr. Fruits finish his comment? I

12 don't know.

13 THE WITNESS: No.

14 MR. SEPENUK: Could he be allowed to finish, Your Honour?

15 JUDGE BONOMY: Yes.

16 THE WITNESS: 150.000 people recorded is a pretty sizeable sample

17 considering that there's only 29 municipalities that someone could have

18 come from. So, again, it's surprising that not even a single person out

19 of those approximately 150.000, perhaps even more, were --

20 JUDGE BONOMY: It's more --

21 THE WITNESS: -- not recorded --

22 JUDGE BONOMY: -- like 170.000, I think.

23 THE WITNESS: Thank you. It's tough to do math on the stand.

24 MR. STAMP:

25 Q. In any case, Dr. Ball's method was to use the surveys to come to

Page 26005

1 an estimate of people who came from these varying or these various

2 municipalities.

3 A. That's correct, yes.

4 Q. You further say that the indictment does not allege deportation,

5 and I think that's at page 17, third paragraph: "The indictment does not

6 allege the deportation in Novo Brdo. It is an eastern municipality with

7 a significant Serb majority. These factors, as well as the fact that no

8 one from the municipality was recorded as crossing the Morina border,

9 indicates that it is unlikely that Dr. Ball's estimates accurately

10 measure migration from Kosovo."

11 What was the indictment have to do with that? How could this

12 factor that the deportations from those two municipalities are not

13 mentioned in the indictment? How could that be a factor that affected

14 your assessment?

15 A. Well, it's a -- I think it's a piece of information that suggests

16 that perhaps there may not have been any sort of forced migration from

17 Novo Brdo. I'm not drawing any conclusions one way or the other, but

18 the -- it is not mentioned in the indictment; and so, therefore, it would

19 be somewhat surprising, if there's no allegations of deportations from

20 that area, that people would be deported from that area.

21 Q. And, therefore, that's a factor that you consider?

22 A. It's one of several.

23 Q. Very well. And, again, you go on at page 18, I think I have

24 called this already, to indicate that the numbers that Dr. Ball estimates

25 are higher than the border records. Well, I think you agreed with me

Page 26006

1 that the border records do not record the entire volume of people

2 crossed.

3 A. I think that might be a fair assessment.

4 Q. Later on or earlier on, you said that there is also flaws in the

5 municipalities -- municipality of origin estimates, at 7.2. That's

6 page 15. You said: "Only 70 per cent of the survey respondents,

7 however, are recorded on the register maintained by the Albanian

8 officials at the Morina border crossing. Somehow, by accident or by

9 design, the surveys on which Dr. Ball rely seem to have systematically

10 excluded those who names were recorded at the Morina border crossing."

11 JUDGE BONOMY: The figure recorded should be "7 per cent." It

12 says "70" on the transcript.

13 MR. STAMP: "7 per cent" is right, and I'm reading from page 15,

14 first paragraph of the report.

15 JUDGE BONOMY: What's your question?

16 MR. STAMP:

17 Q. The question is this: Are you aware from the data and the

18 material that Dr. Ball used that there were only 19.000 border records

19 naming persons who have crossed the border, approximately 19.000 border

20 reasons, meaning persons who had crossed the border? Well, approximate

21 19.000 as ballpark figure.

22 A. I'm not aware of that, but I'll accept it.

23 Q. That many times persons crossed the border in groups and only one

24 person from the group -- group's name was taken at the border. Are you

25 aware of that from the disclosure of documents?

Page 26007

1 A. That may be the case.

2 Q. Whereas, the count from the border in terms of how many people

3 crossed was 272.000. You are aware of that?

4 A. I thought it was higher.

5 Q. Approximately 272.000, I think.

6 A. I thought it was 276.500.

7 Q. All right. 276.500. You will agree with me, this is a slight

8 mathematical exercise, that 19.000 named persons out of 272.000 actual

9 persons who crossed Morina border is approximately 6.61 per cent?

10 A. I'll accept that, yes.

11 Q. Very well. And, therefore, you will agree with me that when

12 organizations like Human Rights Watch, OSCE were doing the surveys across

13 the border in the camps, they could get the names of everybody they

14 surveyed?

15 A. I don't know if that's true. Again, I'll accept it as true for

16 purposes of this discussion.

17 Q. Well, did you -- well, you said you didn't have a look at any of

18 that.

19 JUDGE BONOMY: Mr. Stamp, we'll need to break soon. Can you find

20 a suitable time?

21 MR. STAMP: Your Honours, to save time, I wonder if I could hand

22 him some documents. These are documents referred to in his report, so

23 that when we ask questions about it later on, he could look at them at

24 the break if he wants to.

25 JUDGE BONOMY: This's no problem with that, Mr. Sepenuk, is

Page 26008

1 there?

2 MR. SEPENUK: No, not at all, Your Honour.

3 JUDGE BONOMY: I'm sorry?

4 MR. SEPENUK: Not at all, no problem.

5 JUDGE BONOMY: Where are we in the cross-examination, Mr. Stamp?

6 MR. STAMP: I don't think I have more than 45 minutes. In fact,

7 I don't have more than 45 minutes.

8 [Trial Chamber confers]

9 JUDGE BONOMY: When we come back after the break we will be

10 without Judge Chowhan, who has to leave for urgent personal reasons.

11 We've discussed it earlier. We had thought it possible we might finish

12 this evidence before the break; but since we haven't, we think it only

13 right to continue. It's in the interests of justice to do so in his

14 absence.

15 Dr. Fruits, we have to have a break at this stage, that will be

16 for 20 minutes. Could you again leave the courtroom with the usher.

17 We'll see you at five minutes to 11.00.

18 [The witness stands down]

19 --- Recess taken at 10.32 a.m.

20 --- On resuming at 10.54 a.m.

21 [The witness takes the stand]

22 JUDGE BONOMY: Mr. Stamp.

23 MR. STAMP: May I?

24 JUDGE BONOMY: Yes, please.

25 MR. STAMP: Thank you very much, Your Honour.

Page 26009

1 Q. So when we left off, I was indicating to you that out of 19.000

2 named border records of the 276.000 persons counted that passed, this

3 would be 6.88 per cent of the total that crossed the border were named in

4 the border records. So would you accept that this could be the reason

5 why only 7 per cent of the named surveyed respondents are recorded in the

6 border records?

7 A. They may be reasonable.

8 Q. Therefore, that could explain your concerns at 72?

9 A. Yes.

10 Q. That is not an illustration of a fundamental flaw.

11 I gave you some documents earlier on. These relate to -- in your

12 report, if you look at 11.4 of your report at page 27, you make a point

13 here, and I quote.

14 "Dr. Ball does not indicate that he considered other reasonable

15 explanations for migration 'patterns' he describes. Exhibit 9 suggests

16 one reasonable alternative explanation. It shows that each of the low

17 points in the number of individuals and groups registered at the Morina

18 border coincide with the border being closed."

19 And you give a citation at footnote 61 where you refer to some

20 news reports, and those reports I gave to you to look at in the break.

21 Did you get an opportunity that look at that?

22 A. Briefly.

23 Q. Dr. Ball referred to two distinct low points, 6th and 7th of

24 April and the 22nd to the 24th of April. Which of the documents that you

25 cited -- did you find anywhere in those documents that you cited in that

Page 26010

1 footnote which indicates that the border was closed on the 22nd to the

2 24th of April?

3 Actually, Doctor, would I be correct in suggesting that none of

4 these documents that you cited indicates that the border was closed --

5 A. I can't tell you right now --

6 Q. -- throughout that period?

7 A. I can't tell you right now. I would have to look through these

8 documents. It looks like the list you gave me doesn't include all the

9 documents I cited.

10 Q. Well --

11 A. I may be incorrect because during the break I only briefly looked

12 at them.

13 Q. Well, do you have the documents that you cited?

14 A. I have all the documents I cited but I don't have them

15 highlighted in any way, so I would have to look through them to find the

16 information.

17 Q. And the next question would have been that you referred to a low

18 point in your cite -- in your report, and you also said the border would

19 have been closed for those two days the 29th to the 30th of March. Can

20 you find that in those documents?

21 Well, I'm going to suggest to you that those documents do not

22 indicate that the border -- the documents you cited would not indicate

23 the border was closed on those two occasions.

24 A. I just found the article regarding the March. It is the

25 article that is the first one listed BBC, Border shut on refugee tide.

Page 26011

1 Next to the map of Kosovo, it says: "The post at Morina near Kukes about

2 250 kilometres, 150 miles, north of the Albanian capital of Tirana was

3 shut after a total of more than 60.000 refugees had arrived."

4 MR. STAMP:

5 Q. Now, is that what you used to conclude that the border was closed

6 for two days?

7 A. Well, it was published on March 29th at 12.51 GMT.

8 JUDGE BONOMY: We'll need to go back here. I wasn't watching the

9 transcript at the time, but your answer has not been recorded because you

10 were reading too quickly for the interpreter.

11 THE WITNESS: I apologise.

12 JUDGE BONOMY: "The post at Morina near Kukes about 250

13 kilometres, 150 miles," and then it runs out of transcript.

14 THE WITNESS: I apologise.

15 JUDGE BONOMY: So could you repeat that, please.

16 THE WITNESS: After that, I say: "... north of the Albanian

17 capital of Tirana was shut after a total of more than 60.000 refugees had

18 arrived."

19 MR. STAMP: For the record, he's reading from P3152.

20 Q. And that section that you read, is that what led you to conclude

21 that the border was closed for two days?

22 A. Can you show me in my report where I indicate that was closed for

23 two days?

24 JUDGE BONOMY: This article relates to March?

25 THE WITNESS: That's correct.

Page 26012

1 MR. STAMP: I had asked previously about April, and he said he --

2 JUDGE BONOMY: This article doesn't relate to the 22nd to the

3 24th of April.

4 MR. STAMP: Well, I think he had indicated, Your Honours, before

5 that, he couldn't find it for that period.

6 JUDGE BONOMY: Yes.

7 MR. STAMP: So I asked him about March after that.

8 JUDGE BONOMY: Okay.

9 THE WITNESS: I'm sorry. I'm not clear where it says in my

10 report that I indicate that the border was closed for two days on March

11 29th.

12 MR. STAMP:

13 Q. Okay. I'll just ask you what you are saying. At one point, you

14 said, when the border was closed, it was closed for a relatively short

15 period, one to three days, with the exception of the 30th of March.

16 What time-period are you referring to here when you say one to

17 three days?

18 A. I'm saying that in some cases it was closed for one, in some

19 cases it was closed for two, and in some cases it was closed for three.

20 I didn't mean for you to take an average --

21 Q. From that report, how long was the border closed for according to

22 you on the 29th --

23 A. Well, I just put an indicator there that the border was closed.

24 I don't actually -- I don't distinguish -- in this exhibit, I don't

25 distinguish between -- about the amount of time. If it was closed on

Page 26013

1 that day, if it was closed for 12 hours or 24 hours, I just indicate that

2 the border was closed.

3 MR. SEPENUK: Excuse me, Your Honour. Just to clarify, I think

4 the exhibit Dr. Fruits is referring to is Exhibit 9; is that correct,

5 sir?

6 THE WITNESS: Exhibit 9 to my report.

7 MR. SEPENUK: I just wanted to make that clear.

8 JUDGE BONOMY: Thank you.

9 MR. STAMP:

10 Q. Yes. You have the border here closed in Exhibit 9 for the 30th.

11 Now, that is what you based your conclusion upon. Can I show you a

12 document.

13 MR. STAMP: Could we look at P1 -- P3153.

14 Q. Basically, you know, this is a report of the 30th of March, since

15 you are using newspaper reports to base your facts, that indicates that

16 the border was closed overnight. It's in the Agence France-Presse

17 report.

18 MR. SEPENUK: I'm sorry, Your Honour. Excuse me for

19 interrupting. Was this -- maybe I've missed it here. Was this one of

20 the footnote references on 61 that was referred --

21 MR. STAMP: No, it's not. It's a different report.

22 MR. SEPENUK: Oh. As I understand it, Your Honour, this is not

23 referred to in Dr. Fruits' report, footnotes 61, page 27. So, again, I

24 don't want to testify here, but I would have some doubt that he has seen

25 this.

Page 26014

1 MR. STAMP: That's why I put it on the screen.

2 Q. So, having seen that news report, you would agree with me that it

3 is possible that the Morina border post might have been closed overnight

4 on the 30th?

5 A. I don't understand. You're asking me to read this right now?

6 Q. No, no. I'm just asking you to read the first two paragraphs.

7 It's a newspaper, report as you can see, from Agence France-Presse, and

8 I'm asking: Having read those two paragraphs, would you agree with me

9 that it is possible, if you are going off newspaper reports, that the

10 border was closed overnight?

11 A. I typically don't make it a practice to comment on a document

12 unless I can read the full document. May I have time to read the full

13 document?

14 MR. SEPENUK: Also, Your Honour, to the extent that we're talking

15 about a newspaper report, it's obvious that we can't rely on the truth of

16 what was said in this report, particularly if it hasn't been looked at by

17 the witness.

18 JUDGE BONOMY: One of the things I think we're trying to find out

19 is if the witness himself relied on it.

20 THE WITNESS: I did not rely on this document. Let me

21 double-check on that.

22 MR. STAMP: No, no. This is not cited.

23 The answer to the comment made by Mr. Sepenuk is that, basically,

24 the witness has relied upon newspaper reports to come to conclusions of

25 fact. I suggest that the facts are not borne-out, and I'm showing him

Page 26015

1 another newspaper report; and I am asking him if, having seen that

2 newspaper report, he would agree with me.

3 JUDGE BONOMY: This one suggests that the border was closed on

4 the 29th into the 30th.

5 MR. STAMP: Yes.

6 Q. All I'm asking, Witness, is whether or not you are prepared to

7 agree with me that it is possible that the border was only closed

8 overnight for that period?

9 A. May I read the article?

10 JUDGE BONOMY: I thought you were doing that. Yes, please.

11 THE WITNESS: Yes.

12 JUDGE BONOMY: It's not all that difficult to read.

13 THE WITNESS: I'll try not to move my lips.

14 JUDGE BONOMY: What's meant by that comment?

15 THE WITNESS: That I'm a slow reader, and I have to talk while I

16 read. I've read the article.

17 So, it sounds like the -- it indicates here that the border was

18 closed at 10.00 p.m. and then re-opened early Tuesday. It doesn't

19 indicate what time on Tuesday morning. So the border was, in fact,

20 closed.

21 MR. STAMP:

22 Q. Overnight, do you agree with that?

23 A. That's correct, overnight, yes.

24 JUDGE BONOMY: The problem we have is finding the basis for your

25 determination that it was closed. It's simply that BBC report you

Page 26016

1 referred to?

2 THE WITNESS: The BBC report indicates it was closed, and then

3 this article seems to confirm that it was closed.

4 JUDGE BONOMY: Yes. But looking at your exhibit, what does that

5 indicate about the period during which it was closed?

6 THE WITNESS: If you look at the date that says March -- that

7 says 30 March, it's the first item on the left that says "border closed."

8 That indicates that one of the -- one of the low points in the border

9 crossings was -- coincides with the border being closed.

10 JUDGE BONOMY: During what period is the border closed? This is

11 very imprecise, and that's not a criticism. It's imprecise so far as

12 time is concerned, because the 30th is Tuesday when it would appear to

13 have been opened.

14 THE WITNESS: Well, I think the -- I think the difficulty with

15 this picture and the imprecision, you're correct, is that the border was

16 closed for a portion of the 29th for the late -- for the last hours of

17 the 29th, it appears. Then it was also closed into the early hours of

18 the 30th. It's unclear when it re-opened, but it was closed for some

19 time in the -- during the 30th.

20 JUDGE BONOMY: But you record, well, on this, if I'm reading it

21 correctly, a significant number of people crossing the border on

22 the 29th.

23 THE WITNESS: That's correct.

24 JUDGE BONOMY: And, virtually, well, a much smaller number on

25 the 30th, and it seems to be the 30th and the 31st that are the low

Page 26017

1 dates.

2 THE WITNESS: Yes.

3 JUDGE BONOMY: Okay.

4 THE WITNESS: So one way to think of it is: On the 29th, it was

5 open for 22 out of 24 hours of the day; on the 30th, it was open for some

6 indeterminant number that's less than 24. This may also highlight what I

7 would call the announcement effects. If news gets out that the border is

8 closed, people probably aren't going to leave to go across the border.

9 JUDGE BONOMY: The odd thing, though, is that it's closed

10 trapping thousands of people, which means or suggests to me that as soon

11 as you open it, you'll get a flood of people crossing.

12 THE WITNESS: Which you do on the 1st.

13 JUDGE BONOMY: Yes. But that's not consistent with the report.

14 Anyway, let's proceed, Mr. Stamp.

15 MR. STAMP: Thank you, Your Honours.

16 Q. You said yesterday when you testified, if I could find it, I

17 noted it's 2596 [sic] of yesterday's testimony -- well, before we get

18 there, there's just one thing I would like to clarify.

19 Yesterday, you also said at page 25959 - I see this note here -

20 in answer to a question from the Bench: "The difficulty I have as a

21 statistician is that you really can't come to a conclusion in a

22 round-about way, in which Dr. Ball does, in which you simply reject two

23 other alternative explanations, and then come to a conclusion that it

24 must be the third alternative. In this case, the third alternative being

25 the coordinated efforts of Serb or Yugoslav forces."

Page 26018

1 Is this a misrepresentation of what Dr. Ball has said? Can you

2 point me to where he has said that, having rejected the first two

3 alternatives then it must be the forces of Yugoslavia?

4 I want you to focus on "it must be the third alternative."

5 A. Oh, he did not say it must be the third alternative.

6 Q. Okay. You misspoke then. Very well.

7 MR. SEPENUK: Those were his words, I believe.

8 MR. STAMP: Sorry?

9 MR. SEPENUK: I think the witness can clarify. No, excuse me,

10 excuse me. I misread it. I think that's correct. I misread what was

11 said here.

12 JUDGE BONOMY: All Ball does is try to show a consistency with a

13 hypothesis; is that correct?

14 THE WITNESS: He does, yes, that's correct.

15 But by the same token, I think I could rephrase that statement to

16 fit what Your Honour just asked; that is, that you can't come to the --

17 you can't reject the other two conclusions of KLA and NATO activities,

18 and then come to a third conclusion that somehow the data is then

19 consistent with coordinated efforts of Serb activities, without actually

20 seeing the Serb activities.

21 JUDGE BONOMY: That's a rather different point from the point

22 about being clear in what it is that Ball actually claims.

23 Mr. Stamp.

24 MR. STAMP:

25 Q. You also said, yesterday, that in omitting the variable of the

Page 26019

1 activities of Serb and Yugoslav forces, you are committing an error known

2 as omitted variable error. Are you saying that there is no point in

3 trying to reject any number of hypotheses unless you include

4 consideration of all possible hypotheses or all variables?

5 A. No, you can't. You can't do an analysis where you include every

6 single variable, but the omitted variable bias relates to omitting what's

7 known as a relevant explanatory variable. As I understand it, Dr. Ball,

8 and maybe even the Prosecution, believes that the activities of Serb or

9 Yugoslav forces is an important, if not the most important, explanatory

10 variable.

11 Q. Very well.

12 A. If you exclude that variable, you've committed what's known as an

13 omitted variable bias.

14 Q. No. But are you saying, therefore, that you cannot statistically

15 seek to reject these two variables if one of them -- if the third

16 variable is not considered?

17 A. If you have a hypothesis that three variables possibly explain

18 the deaths or the migration, you can't statistically come to a conclusion

19 by only looking at two of those three because not only are you omitting

20 the variable that you're interested in, but you're also biassing. In

21 other words, your estimates aren't going to reflect reality on the

22 impacts of those other variables.

23 Q. No. I understand what you have said. So, basically, what you

24 are saying is, in seeking to reject the two variables, statistically

25 speaking you must include analysis of the third? You cannot

Page 26020

1 statistically seek to reject the two variables without --

2 A. That's correct. And as I state in my report, that it, in fact,

3 by excluding the Serb and Yugoslav forces' activities, you are more

4 likely to reject the hypothesis that KLA and NATO activities contributed

5 to it. In other words, you would be more likely --

6 Q. You have answered my question and you are repeating what you said

7 in the report.

8 Did you, assuming that Ball's data sets were correct, did you run

9 your own models to test whether there were patterns?

10 A. No.

11 Q. You state that it is possible to do so.

12 A. Can you point out where I say that?

13 JUDGE BONOMY: It may be page 30, 29 and 30.

14 MR. STAMP:

15 Q. Well, I think, at page 36, you said: "Economists have developed

16 statistical techniques to account for censored data," and you referred to

17 one model. To be precise, did you use this statistical technique to

18 analyse the linear regression model that Dr. Ball presented?

19 A. No.

20 Q. Assuming that the data sets that Dr. Ball are correct, could you

21 have done so?

22 A. You're saying if the -- so you're saying if the deaths were

23 correct --

24 Q. The data sets.

25 A. -- and if we had information on Serb and Yugoslav forces'

Page 26021

1 activity?

2 Q. Just the data sets that Dr. Ball used, assuming they are correct,

3 could you have done so?

4 A. Not without the Serb or Yugoslav forces' activities.

5 Q. In respect to the correlated relationship between migrations and

6 deaths, and that's all I'm asking, could you have formulated a model to

7 test whether or not Dr. Ball's conclusions that they are highly

8 correlated are correct?

9 A. I could only to the extent that the information on deaths and

10 migration were correct.

11 Q. Well, assuming those -- the information is correct, you could?

12 A. One could, yes.

13 Q. You say that the methodology used by Dr. Ball et al. was flawed.

14 Is there any reason why you didn't apply your own methodology to analyse

15 the correlation of deaths and migration?

16 A. The main --

17 Q. Assuming his data was correct.

18 A. Well, the main reason I didn't do it was because it wasn't my

19 assignment, another reason is that his data wasn't correct, and a third

20 reason is that Dr. Ball et al. did not provide sufficient detail of what

21 data they used to produce figure 2 for me to actually perform that

22 analysis.

23 Q. I believe you had said that you could do so only to the extent

24 that the information on the deaths and migration were correct. Did you

25 run your own tests or did you formulate a model to test whether or not

Page 26022

1 the additional information that you gathered on NATO air-strikes would

2 have affected the conclusions of Dr. Ball?

3 A. I only collected enough data on the NATO air-strikes to --

4 Q. Did you do what I -- did you run or did you formulate your own

5 model to test it?

6 MR. SEPENUK: I think he's answering the question, Your Honour,

7 and I think he should be allowed to answer.

8 JUDGE BONOMY: Please answer that question.

9 THE WITNESS: I did not run my own tests because I didn't have

10 the data from NATO to -- I didn't collect the data from NATO to perform

11 that test; and, again, it wasn't my assignment.

12 MR. STAMP:

13 Q. Well, the question is: Having Dr. Ball's data that he used - we

14 all know that data is sometimes incomplete - you having at your disposal

15 additional information, did you use the additional information and

16 Dr. Ball's data to -- in your own model to see whether or not the new

17 information you had changed his conclusions?

18 A. No.

19 Q. Maybe I'm -- how long would it take you to do something like

20 that?

21 A. Well, that's -- that's pure speculation because it would

22 require --

23 Q. Could you give me a ballpark time it would take you to do that

24 sort of analysis?

25 A. Probably another two years. You would have to recreate all of

Page 26023

1 Dr. Ball's data from scratch.

2 Q. No, I'm asking, just assuming that his data sets that you have

3 had are correct, how long would it take you to construct your own models,

4 using the various techniques that you refer to, to test their findings

5 and present it in writing?

6 A. It would probably take me several weeks.

7 Q. My information is that the time-consuming part of this work is to

8 create the data sets and that a qualified statistician could construct a

9 model to test the conclusions in a matter of hours. You are saying that

10 that is not correct. You would need weeks?

11 A. I have other projects.

12 Q. But if you were working on nothing else, how long?

13 A. Oh, it would still probably take at least a week. I'm a bit more

14 careful than other statisticians, I guess.

15 Q. Okay.

16 MR. STAMP: I think I'm finished, Your Honour, but could I have a

17 moment to just reflect.

18 JUDGE BONOMY: Yes.

19 MR. STAMP: Thank you, Your Honours. I have nothing further for

20 this witness.

21 JUDGE BONOMY: Mr. Sepenuk.

22 MR. SEPENUK: I have no questions on re-direct, Your Honour.

23 JUDGE BONOMY: Thank you.

24 [Trial Chamber confers]

25 JUDGE BONOMY: Dr. Fruits, that completes your evidence. Thank

Page 26024

1 you for coming here to assist us. You may now leave the courtroom with

2 the usher.

3 THE WITNESS: Thank you.

4 [The witness withdrew]

5 JUDGE BONOMY: I understand that completes the available evidence

6 for this week. Any comments that anyone wishes to make on any

7 outstanding issue or any procedural matter?

8 We assume that we will on the 13th of May and possibly the 14th

9 of May be hearing the evidence of two more expert witnesses, one on

10 forensic pathology and one on the constitution. We think that they are

11 the only, as at this moment, confirmed witnesses with specific dates on

12 which to give their evidence. If there is any other known to anyone,

13 please let us know now so that we can make appropriate arrangements.

14 Silence.

15 MR. VISNJIC: No, Your Honour.

16 JUDGE BONOMY: Mr. Visnjic.

17 MR. VISNJIC: [Interpretation] We still have a handwriting expert

18 to hear.

19 JUDGE BONOMY: Oh, of course.

20 MR. VISNJIC: [Interpretation] And at this point in time, I can

21 only assume that he will testify during that period; that is to say, the

22 first week after the break.

23 JUDGE BONOMY: Yes.

24 Mr. Ackerman.

25 MR. ACKERMAN: And I would suspect also, Your Honour, that a

Page 26025

1 handwriting expert from the Pavkovic Defence would be available at

2 roughly the same time. That's our plan.

3 JUDGE BONOMY: Thank you.

4 Well, we all find ourselves in different circumstances, some

5 obviously happier than others, but we're probably all grateful for a

6 break from court proceedings in whatever capacity we happen to be here.

7 So we wish you well and hope that insofar as you can you make the most of

8 the period of recess that we have for two weeks.

9 We shall resume on Tuesday, the 13th of May, at 2.15, and at the

10 moment we're scheduled to be in this courtroom.

11 --- Whereupon the hearing adjourned at 11.39 a.m.,

12 to be reconvened on Tuesday, the 13th day of

13 May, 2008, at 2.15 p.m.

14

15

16

17

18

19

20

21

22

23

24

25