²²¹⁸

Thursday, 14 March 2002

[Open session]

[The witness entered court]

[The accused entered court]

--- Upon commencing at 9.32 a.m.

JUDGE MAY: Yes, Mr. Milosevic.

WITNESS: PATRICK BALL [Resumed] Cross-examined by Mr. Milosevic: [Continued]

Q. I would now like to check some facts. The American Association for the Advancement of Science, in 2001, gave Flora Brovina its annual award. Do you know that?

A. Yes.

Q. As far as financiers are concerned, we tried to clarify something yesterday when we said that this was financed by the US government. The sponsors of your association are also an agency of the German Catholic Church, the Institute for Peace, the Ford Foundation, and some projects are funded by the governments of the Netherlands, Great Britain, and Denmark; is that correct?

A. Some of your -- some of the organisations you listed -- could you repeat them, please, so that I could review them.

JUDGE MAY: Put them to the witness one by one so he has a chance to answer.

MR. MILOSEVIC: [Interpretation]

Q. An agency of the German Catholic Church.

A. Although I'm not aware of all the groups that fund the entire ²²¹⁹ association, I do not believe that such an agency has funded the Science and Human Rights Programme, of which I am a part.

Q. All right. In your report, you said that there was a harmonisation of figures that was carried out together with the International Crisis Group, the ICG.

A. Perhaps there's a translation issue. I don't believe we used the term "harmonisation." Furthermore, I do not think we used any data from the International Crisis Group.

Q. But you did compare your figures with theirs?

A. Not with the International Crisis Group. We made comparisons to analyses presented by Physicians for Human Rights and for -- and by our two epidemiologists who worked with the US Centre for Disease Control, which was published -- their work was published in a medical journal.

Q. All right. Did you make a report for Burundi, for example?

A. I have never worked on Burundi, no.

Q. Did you do anything for this International Crisis Group or did you cooperate in any way with them?

A. I have met some scientists who have in the past collaborated with the International Crisis Group. I have met them at meetings, at scientific meetings. I have heard them present papers. I have never personally collaborated with the ICG, no.

Q. All right. You said that -- you said yesterday in your statement that 45 per cent of the data you statistically processed, you simply took over from the Prosecutor, if I understood you correctly; is that right?

A. The sources listed in the database that was provided to me were ²²²⁰ sources internal to the Tribunal, of various types. That's correct, 45 per cent.

Q. Forty-five per cent. And the rest are reports of the Albanian government and those lists that were compiled by Albanian clerks, and they handed them over to you at Morina and elsewhere, as you had put it. So that is the database that you've been operating with. It seems to me that I've heard --

A. That's not quite correct.

Q. In what sense is it not correct?

A. There are distinct systems of data. Each examines one of the kinds of information we are looking for. The Albanian government data is used only in the migration analysis. The lists to which you refer are used in the analysis of killing. The discussion to which I just referred, the 45 per cent sources contributed by the Office of the Prosecutor's internal sources and the corresponding 55 per cent of the sources which were primarily from the Yugoslav government, these data were on KLA activities only. The fourth series of information on NATO activity comes exclusively from Yugoslav government sources.

In order to understand how the analysis is done, I think it is quite important to be clear about each source.

Q. And then since you're talking about sources of the Yugoslav government, did the Office of the Prosecutor give to you the reports that the government handed over to the Office of the Prosecutor? Let me clarify this. I heard here from Mr. Tapuskovic, I believe it was, Tapuskovic, an amicus curiae, when he spoke during one of the ²²²¹ hearings - Mr. Tapuskovic, please correct me if I'm misquoting you - that the government of Serbia sent 26.000 pages regarding activities of the KLA to this institution. Is that right? Did you obtain from the Office of the Prosecutor these 26.000 pages about the activities of the KLA, these 26.000 pages that were sent by the government of Serbia?

A. I don't know what the government of Serbia sent. What I do know is that there were many sources in the data among the -- in the data on the KLA exclusively. There were many sources that referred to Yugoslav government activity. Our interpretation of the sources is that they were primarily open sources, that is, information that had already been made publicly available. As I mentioned, we began our analysis of the patterns of KLA activity by using information that was on a Yugoslav government website that covered casualties to Serb forces inflicted by the KLA.

Q. So my question is: Did you obtain from the Office of the Prosecutor materials of the government of Serbia or of Yugoslavia, whatever you prefer, that pertain to KLA activities? I am asking you, did you obtain this from the Office of the Prosecutor?

A. The sources I received in the database refer to a number of Yugoslav government sources. Whether these sources are the same as the reports to which you have referred, I don't know. I don't know. The sources are listed in Appendix 3 in terms of the originating or authoring organisations which produced this information. Perhaps that will help clarify the question.

Q. Yes. But none of the things I've been referring to are contained there. All right. This question is quite clear. This side across the ²²²² way gave you 45 per cent of the data for your database. I see in your report, on the 19th of February, 2000 [as interpreted], that is to say, two or three weeks ago, you say: [as read] "The Office of the Prosecutor asked us to look at the analysis on pages [sic] 8 and 9, bearing in mind the relationship between the killings and refugee flows."

And your ultimate conclusion is: [as read] "In this review, we noted that the correlation neither proves nor implies that the killings led to refugee flows or vice versa."

So the Office of the Prosecutor put forth certain requests to you in respect of the expert activities that you are to carry out, and you give answers. As far as I can see, even up to the 19th of February, 2002. How intensive were the contacts between you and your co-authors from the OTP while these statistical surveys that you presented were being made?

JUDGE MAY: What co-authors from the OTP?

THE WITNESS: I have no co-authors from the OTP. That seems to be a lack of clarity. My co-authors come from the American Statistical Association, Carnegie Mellon University, and the American Bar Association, Central and East European Law Initiative, in addition to the American Association for the Advancement of Science.

The OTP asked for a clarification because in Figures 8 and 9 - not pages - in Figures 8 and 9, there were some questions they had about whether the same analysis would reveal what we would expect it to reveal with respect to killing and migration. Again, we found that there's a ²²²³ strong relationship in time and space between killing and migration. That is, in the same places and at the same time when one is strong -- when killing is at a high point, migration also tends to be at a high point. And when killing tends to be at a low point, migration tends to be at a low point. And that is consistent over different times and places throughout the period.

The analysis that the OTP asked me to clarify here is to examine whether, when migration is at a peak, if there has been some presence of killing just prior to that; that is, in the same period or in the previous two-day period, have there been one or more killings, reported killings? Similarly, when killing is at a high point, has there been, in the same period or in the subsequent period, some migration activity? We find that in both relationships, we find very high levels of presence, much higher than we found in Figures 8 and 9. This is -- in fact, we find roughly double the proportions of presence.

JUDGE MAY: Dr. Ball, you've given your evidence about that.

THE WITNESS: Okay.

JUDGE MAY: We'll see if there are any more questions.

MR. MILOSEVIC: [Interpretation]

Q. Well, I didn't want to go into these questions related to killings now, but since you have given such an extensive explanation, I would like to remind you that it says here, in connection with what the OTP asked you, it says here, on the 19th of February this year: "We have noted in this paper that the correlation neither proves or implies that killings brought about refugee flows." That's what it says here. ²²²⁴ Then the OTP also asked you, as you put it here, during a telephone conversation that took place last week: "The representatives of the OTP asked me to apply the peak method related to the correlation between killings and refugee flows." And you also say that this observation is in line with the original findings. Then finally, you say: "However, we would like to caution that these observations are not a confirmation that it was killings that caused refugee flows."

That is what you said here. That is what you have written here. Is that correct? Do you abide by that position?

A. The reason that this analysis was not included in our original report is that we believe that this method is appropriate to reject a hypothesis. It is not an appropriate method to confirm one. We did not, therefore, use it to try to confirm the method, although, as I said a moment ago, the statistics imply the confirmation. I am cautioning here in the sections that have been quoted, I'm cautioning that although the numbers seem to be favourable to the arguments and the conclusions we have reached, that we should not use these in that manner. That is not statistically sound. That is the extent of the finding. I confirm that we have concluded that there is a strong correlation over time and space between killing and refugee flow, but that confirmation should not be drawn from this argument. That is the point of the sections that have been quoted.

Q. All right. But since you are going into this problem in such detail, you have a few manipulations - that's what I'd call it - in terms ²²²⁵ of the number of persons killed, estimates. On page 3, you say the analysis includes the estimate of killings, about which none of the four sources had been informed. That is the main problem, because NATO and the KLA also relate to killings, and this has to be based on true data as well. Is that correct or is that not correct?

JUDGE MAY: Do you follow the question?

THE WITNESS: Not quite.

JUDGE MAY: Would you clarify the question, Mr. Milosevic.

MR. MILOSEVIC: [Interpretation]

Q. The question is that I believe that what I quoted here is indeed a problem, because when analysing the NATO and KLA relationship to the killings, he enters that problem -- he approaches that problem with estimated figures, not true figures, actual figures. Am I being clear? I think I'm being very clear.

A. All of the analysis of the killings of ethnic Albanians, with the exception of Figures 8 and 9, all of the rest of the analysis, however, is based on the estimates, including the residual analysis in section 5.3, the analysis of the residual patterns which we discussed in depth yesterday. So the exception on the analysis of killings of ethnic Albanians here is only on Figures 8 and 9.

Q. All right. But there is a difference, a big difference, between your estimates and what you have as facts, because on page 5, it says the estimated number of killings obtained by statistical analysis is 10.356, with an error margin of 9.002 to 12.122. And on page 17, the table of exhumed bodies is a total of 4.211, a total of 4.211, and identified ²²²⁶ bodies are 45.4 per cent. Identified Albanian victims actually amounted to 1.912, so that is 45 per cent of 4.211. And all analyses consistently apply the invented figure of 10.356. Can you explain that?

A. Ten thousand, three hundred fifty-six is not an invented figure; it's an estimated figure. The estimate is based on sound and well-known statistical principles that are in wide use throughout the world and have been subjected to both rigorous preparation and review.

Q. Have you finished your explanation?

A. Yes.

Q. Is it correct that the total you have of identified Albanian victims, those you can identify as Albanian, that you have certified figures only amounting to 1.912?

A. No, that is not correct. The number is 4.400. Those are the victims we can identify by name.

Q. And what about the difference between 4.400 and 10.356? What is in between the two? And quite a bit of scope is provided there. What about that, in relation to what you say you have identified?

A. As I suggested in my previous answer, the estimate of 10.356 is made by well-known scientific methods which have been used in a wide variety of statistical contexts throughout the world. The techniques are explained in significant detail in Appendix 2.

Q. Well, now I understand this a bit more, and I hope those who have been following this too as well, why this statistical research of yours can only be used for educational purposes and for the purpose of providing information, not for the purposes for which it has been used. ²²²⁷

JUDGE MAY: It's suggested that you can't use it as evidence. What's your answer to that?

THE WITNESS: I'm afraid, because I'm not lawyer, I don't know in this kind of level.

JUDGE MAY: It's a comment, Mr. Milosevic. Yes, let's move on.

MR. MILOSEVIC: [Interpretation]

Q. All right. Then I have the following question: In view of this cooperation with the OTP, a question for you personally. Do you believe that the Prosecutor is making a maximum effort, a maximum effort to prove that NATO aggression did not cause an exodus of the population?

JUDGE MAY: It's not a matter for the witness.

THE ACCUSED: [Interpretation] I think that the question is for the witness because he cooperates with the OTP.

JUDGE MAY: It's a comment.

MR. MILOSEVIC: [Interpretation]

Q. And do you think that this paper of yours is very important for the Prosecutor?

JUDGE MAY: It's not a matter for the witness, not a matter for the witness. He's simply giving evidence. It's not for him to comment.

THE ACCUSED: [Interpretation] All right. That's an answer too.

MR. MILOSEVIC: [Interpretation]

Q. Do you remember, Mr. Ball, that some time ago you said that the NATO claims about 100.000 persons missing in Kosovo are absurd? Do you remember having said that? Do you remember that you said that?

A. Yes, I did say that. ²²²⁸

Q. That's what you said. So my question is: In your opinion, what was the reason for such assertions to be made by NATO, these that you had called absurd? In what opinion, what was the reason underlying such assertions?

A. I have no idea. My claim was that the number is too high and the estimates bear that out.

Q. Mr. Ball, did you in the -- the advisory board of the Hacktivism group of international computer hackers, are you in the management board of that group which is known as the "Dead Cow Cult"?

A. I'm not on the management group. I advise them in their efforts to try to help young computer programmers move away from illegal activities and direct their activities toward productive and legal activities.

Q. And what is the purpose, the mission of that group which you are assisting?

A. I think they do many things. The only thing I do with them is to assist them in their effort to help young computer programmers move away from illegal activities.

Q. At the conference of hackers, DefCon 9 on the 14th of July in California [as interpreted] last year, did you say: "It would be very nice if you would applaud the extradition of Slobodan Milosevic to The Hague, and I hope you are as excited as I am about his trial which is about to start in nine months"?

A. Two corrections. The conference was in Las Vegas, Nevada, not in California. And there seem to be some translation slippages between my ²²²⁹ exact quote and what you quoted.

JUDGE MAY: What did you say, Dr. Ball?

THE WITNESS: That's the essence of it. I don't think we need to parse it too much more closely.

The context of the conversation was that I was explaining to young computer programmers what international human rights are. You will note in the paragraph previous to that quotation I had to explain what the Universal Declaration is, what the Geneva Conventions are in a very, very simple sense to people who have no idea of what human rights, international human rights law might mean and why those are important things to respect. My comments were very well-received, and we noticed a lot of excitement in the room around the possibility of doing good rather than, as I said earlier, doing illegal or harmful things, which we discourage.

There was an indictment against Mr. Milosevic, and thus anyone who supports international justice should welcome the extradition and welcome the trial as an opportunity to hear the evidence relevant to that. I support international law and international human rights. This is where I have placed my career. And therefore, it seems to me entirely logical to support the comment that I made.

THE ACCUSED: [Interpretation] I understand --

JUDGE ROBINSON: Mr. Milosevic. It's not very clear to me. Were you just now paraphrasing what you said or offering an explanation for what you said? I'm particularly interested in the passage, "... and thus anyone who supports international ²²³⁰ justice should welcome the extradition..." Is that what you said at the conference?

THE WITNESS: No, Your Honour.

JUDGE ROBINSON: Because what the Presiding Judge asked you to tell us was what did you say in relation to that matter at the conference.

THE WITNESS: I'm sorry that I -- Your Honour, I do not have the quote immediately available to me. There were just some infelicitous phrasing that I heard in the quotation made by the Defence that did not sound like my words. That was the basis for my suggestion that we needed to correct them a bit.

The essence of what he has quoted is essentially correct. My explanation of why I said that is that I support international law and, therefore, welcome the extradition. That is my explanation. I did not go into that complex detail in the talk. I was speaking to a room full of primarily 19- to 22-year-olds, so one has to be very simple.

JUDGE ROBINSON: Proceed, Mr. Milosevic.

MR. MILOSEVIC: [Interpretation]

Q. I did not ask you about your motivations when making that statement. I asked you whether it was correct, whether it was accurate. And it does not mention international law anywhere. And I'm quoting it again because I see a mistake in the interpretation, because I never mentioned California, and you later corrected me that Las Vegas was not in California. Maybe something was added in interpretation maybe by accident.

JUDGE MAY: He has accepted that was in essence what he said, and ²²³¹ he's now given an explanation for it. Whether it was in California does not matter.

THE ACCUSED: [Interpretation] No, that doesn't matter. I'm taking California as an illustration of the error in interpretation probably. Maybe I was reading the quotation too fast and I did a disservice to the interpreter.

MR. MILOSEVIC: [Interpretation]

Q. So the quotation has nothing to do with international law. It says: "It would be very nice --"

JUDGE MAY: We have that point. Now, what is the next question? What is the next question? You --

THE ACCUSED: [Interpretation] Mr. May, why are you interrupting me all the time?

JUDGE MAY: Just ask the next question.

THE ACCUSED: [Interpretation] He is talking about --

MR. MILOSEVIC: [Interpretation]

Q. Well, my question is as follows: Do you, when you are saying that it's nice to applaud the extradition of Slobodan Milosevic to The Hague last week, do you believe that regardless of that attitude, your personal attitude, that is, do you believe that despite your personal attitude, your work related to the indictment against me can be considered as objective?

A. Science does not depend on who the scientist is. I believe that any statistician or demographer or sociologist competent in the methods described in the appendices could take the same data and arrive at the ^{2232 BLANK PAGE 2233} same conclusions. However, I believe that supporting international law does not in any way prejudice one's objectivity. I support international law. Thus I do not believe -- but I do not believe that my objectivity is in any way prejudiced.

Q. Well, since you are applauding my extradition and you say you support international law, do you know that precisely under that law which you say you support, you may not consider me guilty in any respect unless I am proven guilty?

A. Yes.

Q. Therefore, you applaud my punishment, and you provide this fabricated evidence, which is serving the policy that we have noted at the beginning of the examination. Is that right or not?

A. No, that's not right.

Q. And do you remember saying, during that speech in Las Vegas: "In the '80s, we relearnt that people who oppose the American support to wars are targets"? Are these your words?

A. I was asked by people in the audience if they should have physical or data concerns should they do something that would contradict American policy. I replied that during the 1980s, there were some domestic opponents of the United States who were investigated and harassed. This was a domestic question and a domestic concern.

Q. Well, precisely that. People who oppose the American support to wars are targets. Have you changed your views on that since?

A. I was referring to something in the past. I have no knowledge about anything like that more recently. But again, this is a domestic ²²³⁴ concern that I raised about a domestic question, internal to the United States.

Q. So why does one set of criteria apply to the ethical approach within the United States and a different set of criteria apply to the ethics outside the US?

A. I don't know that -- I don't think my comment had anything to do with ethics.

Q. Well, it is an ethical belief that somebody is a target if he or she opposes a certain policy. You are just placing it within the context of the US, as something specific to that country. Why would that differ in terms of within or outside the USA?

JUDGE MAY: Do you understand the question?

THE WITNESS: No, I'm sorry, I don't.

JUDGE MAY: No. He doesn't understand the question. If you want to ask a relevant question, ask it, but this seems to be irrelevant.

MR. MILOSEVIC: [Interpretation]

Q. All right. Your entire statistical construction is based on the premise that Kosovar Albanians were telling the truth, that they are telling the truth; is that right?

A. No, that's not correct.

Q. Is it partially correct?

A. No, it's not. This is better construed as a test of whether it's plausible that the truth was told. It's the reverse of what has been implied.

Q. You yourself, at the seminar sponsored by the Project for War ²²³⁵ Crimes 2000, said and admitted that they were concerned over the possibility that Kosovar Albanians were not telling the truth when speaking to journalists and others; is that correct?

A. That was a concern, and in fact a journalist at that conference discussed a widely-known case in which she had been misled. My response was to point out that these methods test that sort of problem and in fact will reach the same conclusions even if some fraction of the witnesses have lied.

Q. Let's leave witnesses aside. Let's add to those truthful statements of the Albanian side the documents you obtained from the Albanian government and the documents you yourself have discovered in the Morina border crossing. What is not truthful among those things?

A. As far as I can establish, all of the documents that I have received are truthful in their statistical interpretation.

Q. Statistically speaking, when something is true, it doesn't mean it's actually true, but let's deal with that now. Is it correct, in addition to the Albanian guards, border guards, and clerks, that people who collected data for your paper were activists, human rights activists, who worked for the Human Rights Watch and the Institute for Human Rights Studies based in Tirana?

A. My partner organisation in Tirana is the Institute for -- or was, for the migration study, the Institute for Policy and Legal Studies, not for human rights studies. They were my partners, as I said, only for the migration study. I also had partners, as you suggested, at Human Rights Watch, and their data -- I used their data for each part of this analysis. ²²³⁶

Q. Migrations are the central thesis of the Prosecutor; in fact, he calls them deportations. So your partner from Tirana is your partner in the main issue of migrations. On top of that are doctors for human rights. Is that true?

A. Data from the organisation Physicians for Human Rights was used in the migration study.

Q. Now, my question is: Is it correct, is it true, in your estimate, of course, that these groups - so that institute from Tirana and other groups - that they had a vested interest in encouraging and justifying the international intervention in Kosovo, as you call NATO's war against Yugoslavia? So my question is: Do you believe that these groups had a vested interest in encouraging and in justifying the international intervention in Kosovo?

A. I don't know what their interests were or are. I don't work for them.

Q. I'm asking you about your opinion. What could be their interest?

JUDGE MAY: He says he doesn't know. That's sufficient.

MR. MILOSEVIC: [Interpretation]

Q. And have you noticed that they made a portrait of a variety of human rights violations to galvanise public opinion and raise hostility against the Serbs and the desire to punish them? Have you noticed that?

A. I'm sorry. Who are you talking about?

Q. I'm speaking about these groups who were involved, this institute from Tirana and the others who worked for you in gathering data.

A. I don't know of any reports issued by the Institute for Policy and ²²³⁷ Legal Studies in Tirana. Human Rights Watch has issued several human rights analyses of conditions in the region.

Q. Does it seem to you from your cooperation with them that they have an identical or similar attitude to all of this as the OTP, for which you are testifying now?

JUDGE MAY: That's a matter of comment. Witness, you needn't answer.

THE ACCUSED: [Interpretation] Why would this be a comment, please?

JUDGE MAY: It's purely a comment on the OTP, the Prosecutor. It's not for the witness to answer.

MR. MILOSEVIC: [Interpretation]

Q. All right. Is it true that when you compared the data of this Physicians for Human Rights group, who selectively interviewed selected refugees and asked them only about killings in their families, when you compared this to Human Rights Watch data, the data did not correspond, did not coincide?

A. Physicians for Human Rights conducted a random sample survey, choosing people by a method to assure randomness through the population being sampled, to clarify the idea in the question that they were selected. Human Rights Watch, on the other hand, was investigating particular events, not sampling randomly at all. We noted the differences between these two sources and in fact found them to be complementary. However, the data from PHR are not used in the analysis of killings in the current report. They enter into the current report only as one of the ²²³⁸ three sources of survey information used in the migration analysis, and thus have only a very small impact on the conclusions.

Q. So you made a selection of what you will and what you will not use even within that circle of groups who gathered data for you?

A. Yes. We chose the largest ones.

Q. If I understood your explanation correctly, your study's mainly based on data obtained at one border crossing to Albania.

A. The migration study is based on one border crossing. The killing analysis is based on a much wider selection of information.

Q. You described that place as a very chaotic one and said that 40.000 people crossed that narrow road in one day.

A. On one of the days, yes. That was a high point.

Q. In the interview in May 2000, you described the way in which you obtained this information, and if my information is correct, you said that Serbs fired a shell at the administrative building on the border, that you took out records from the rubble and then you gave that to a 17-year-old son of a local Albanian, who scanned them on a computer over a few days, and the documents were then handed over to guards at the border and were relayed to a group in charge of collecting data in Tirana. Is that an authentic description of what you told about this?

A. The border post had been damaged. Local Albanian officials told me there had been some sort of an explosion which they believed to have come from the Kosovar side of the border. There was some broken glass and some broken wood inside the border post. I did collect the documents there from among the broken glass, and I took them back to Kukes in ²²³⁹ Northern Albania.

We did -- I did have an assistant, the local Albanian assistant, the 17-year-old to whom you refer, and he helped us scan the documents, but I scanned most of them myself. I was also assisted by my Albanian translator, a Kosovar Albanian translator.

There were more than 690 pages, 690 pages of the portion used for this analysis, as well as some others which we ended up not using that were not border records and but were other kinds of things. We didn't realise that at first. So there was a great deal of work to be done, and I appreciated the assistance.

We then took the scanned images -- I returned the documents, of course, to the border guards. They weren't mine. But after we'd inventoried them, we returned them to them, and I then did return with the scanned images to Tirana [Realtime transcript read in error "Toronto"] where I worked with a commercial firm to have the scanned images keyed into a database.

Q. It says here in the transcript that you [In English] "scanned images to Toronto." You said "Toronto" or "Tirana"?

A. Tirana. Sorry.

Q. [Interpretation] Tirana then.

A. Yes.

Q. You said yesterday that more than half of the Albanians left precisely over that Morina border crossing. Is that correct?

A. Half of the Kosovar Albanians who left Kosovo crossed through that border. I believe so. That's my estimate. ²²⁴⁰

Q. And can you show this to me on the map, where this border crossing is, on your map.

A. You mean -- are you referring to Figure 3?

Q. I haven't got that list of maps here. My resources are very meager, my resources of work.

JUDGE MAY: Have you got a copy of your report in front of you, Dr. Ball?

THE WITNESS: Yes.

JUDGE MAY: If you look Figure 3 - put it on the ELMO - perhaps you can show us where you mean.

THE WITNESS: It's right there. There we go. It's in this area. To be more specific, we'd need a map that had the roads marked.

MR. MILOSEVIC: [Interpretation]

Q. I couldn't see this well, please.

A. This area.

Q. Can you see that what you have been pointing out is not the Morina border crossing? Morina is much further to the north. You are showing the crossing Vrbnica, south of Prizren. And it is logical that many more Albanians used that border crossing to cross the border, because from Prizren to the border where there is a border service, where on the Albanian side there is also a border service, there is a normal road that people took, and Morina border crossing is much further up to the north.

JUDGE MAY: Let the witness deal with this. Would you be assisted by a more detailed map, Dr. Ball? ²²⁴¹ Just a moment. Just a moment.

Would you be assisted by a more detailed map?

THE WITNESS: Yes, Your Honour.

JUDGE MAY: Yes. Provide a more detailed map. One of the exhibits.

MR. NICE: I'm not sure if this map is going to assist. We're still waiting for the map that I told you about. We'll have another look at another map in a second.

JUDGE MAY: Let the witness see that map and see if it assists.

THE WITNESS: I see the confusion. The border crossing to which the question refers is on the Kosovar side, Vrbnica. Or I'm sorry. My pronunciation is probably incorrect, but V-r-b-i -- V-r-b-n-i-c-a is the spelling on the Kosovo side of the border.

The Albanian guards on the Albanian side of the border refer to the tiny post on the Albanian side as Morina. There is another village substantially to the north.

This is going to be hard to do. No, no, no. We have to move way, way this way. Well, I guess we can start here.

The -- this is the town to which I believe the question refers. That is not what we're talking about here.

I'm just going to move it down here. As the question suggested, there is a road from Prizren to the border here, and as the question referred, there is a small town on the Kosovar side called V-r-b-n-i-c-a. There is also a tiny settlement here, several houses and the border post, which is, as the question suggested, ²²⁴² an official border crossing point. The Albanian guards at that point, in conversations with me through my translator, referred to this point as Morina as well.

So what that confusion is, I think, is -- probably comes from the same name being used to refer to -- being used to refer to different points. But indeed this is the point at which the records were collected, as I discussed a second ago.

MR. MILOSEVIC: [Interpretation]

Q. So you put the Morina border crossing at the wrong place, and the documents were given to you by border guards from the Morina border crossing. I am not saying this just in passing, Mr. Ball. I am saying this because I think that you have been deceived.

JUDGE MAY: Just a moment. Let the witness answer that. You're suggesting he's being deceived.

Yes, Dr. Ball.

THE WITNESS: I collected the documents. I saw them being prepared. I saw the process by which the documents were prepared and I later collected them. I very strongly doubt that I was deceived. This was the name that the guards used to me in their references, and that's the name I reported in the various documents in which I have discussed this. There's another name -- another town, substantially larger, with the same name. That's quite common. There are many towns throughout the region that have the same names, as we discuss, actually, in the current report.

MR. MILOSEVIC: [Interpretation] ²²⁴³

Q. There are not two border crossings by the name of Morina. There is only one Morina, one border crossing of Vrbnica, and you put it in the wrong place, Morina, the Morina border crossing. And then you said according to the records collected at the Morina border crossing, over half of the Albanians crossed there. And that is as if you were saying that a million cars cross the Brooklyn Bridge in one day. It's impossible. That possibility is precluded altogether. This is undoable at the Morina border crossing. It's not a big border crossing anyway. It's just for small border traffic. There's only a macadam road there. There is no real across-the-border traffic there.

JUDGE MAY: Will you assist with that? You were at the border crossing. Is the description which the accused gives the right one?

THE WITNESS: Not at all. Actually, this -- at this border crossing here, there is quite a -- there's a two-lane blacktop road as well as official border posts on both sides of the border, on the Kosovar side and on the Albanian side. There is a significant presence of guards on the Albanian side. I don't know about the Kosovar side. And it's easily capable of having 40.000 people through it, although it will be very, very crowded when that happens.

MR. MILOSEVIC: [Interpretation]

Q. Such explanations, a person can explain any kind of deceit. Anything is possible. But --

JUDGE MAY: Thank you. Let the document be removed.

MR. MILOSEVIC: [Interpretation]

Q. Did it perhaps cross your mind now, when you see this border ²²⁴⁴ crossing, Morina, how you put it in an inaccurate map, in the wrong way, I hope, I hope through no fault of your own. You are a researcher. I imagine that you don't do things like that. But now, now, can you imagine the following situation? Think about this and give us an answer. That these are fabricated papers, that Albanian clerks fabricated these papers, and that this was their mistake, that they put the Morina border crossing there. Because Morina, due to its inaccessibility, is easier for fabricating data because international observers do not get there. And then they gave you a truckload of papers, allegedly from the Morina border crossing.

JUDGE MAY: Let the witness answer -- Milosevic.

MR. MILOSEVIC: [Interpretation]

Q. -- where hardly --

JUDGE MAY: Let the witness answer. If you stop making -- if you stop speeches, Mr. Milosevic, and ask questions, we'd get on more quickly and get through this evidence more quickly.

Now, then, what is suggested is this: that you were handed fabricated papers. Now, can you deal with that?

THE WITNESS: I think it's very unlikely. I collected the papers from a place that had I not collected them there, they would likely have been ruined by the weather in a day or two or whenever the next rainfall were to come.

They did not look fabricated to me. In fact, they looked the same -- as we saw the other day, they looked the same to me when I collected them in June as the ones I had seen them preparing at that ²²⁴⁵ border point in early April -- or excuse me, in early May. So it seems to me highly unlikely that they were fabricated.

MR. MILOSEVIC: [Interpretation]

Q. In view of these errors that we have established, and I hope that we have shown this quite unequivocally, it seems to me that it is highly likely - not highly unlikely but highly likely - that they were fabricated.

I would like to draw your attention to what you gave us yesterday. This is a list of crossings at Morina. This was distributed to us yesterday by the usher here in Court. This is your paper, with an English translation. In all fairness, instead of the number of persons in the group, it said the type of tools, but that was easily corrected. All in all, they probably gave you an authentic paper from the border for these purposes, and they had it translated for you because you cannot read Albanian.

A. No, that is not correct, sir.

Q. What is not correct? Do you speak Albanian?

A. No, but there were several other misunderstandings in the question. First, I selected that document, I selected that document, from -- that page from among the 690 pages which I collected at the border. Second, the document was -- all the documents, the 690 pages, were keyed into a database, which selected the statistical information from the documents, including the locations of residents, the address of each of the parties crossing the border, and their quantity of people in that group. ²²⁴⁶ The other information on the document that's in Albanian was unnecessary in the database, so I did not have to have the documents translated nor did I have to speak Albanian in order to interpret the statistical information that's in those documents.

Q. Even if we were to reduce life itself to statistics only, even then it is worthwhile seeing what a document pertains to, because that provides some information as well. In the English translation here, it says [In English] "Note." [Interpretation] "Note," and that is empty. And in the original --

THE ACCUSED: [Interpretation] Could you please place it on the overhead projector, the original. The one you gave me yesterday. You can take my copy. I'm not going to keep it as a memento. You can have it. In all fairness, it's a photocopy, not an original. Put it down a bit, just a bit, a bit, so that you can see the date on the top. [In English] Please, put it down a little bit to see the date. Okay. But up, up now. No, no. To see the date and to see the note at the same time, please. Okay.

[Interpretation] Take a look now. Take a look now. The date is on the top, the 7th of April, 1999, and further down, there is a note. Could you please zoom in on the note, down here. Down here, it says that the information pertains to the time from the 3rd until the 7th of April, 1999. Therefore, from the 3rd until the 7th, even according to their data, 209 persons had crossed in five days. That means an average of 40 persons per day in this critical period, concerning which you have been claiming that across Morina, only in one ^{2247 BLANK PAGE 2248} day - I don't know what you said - 40.000 had crossed or 60.000, whatever.

You can now remove it from the ELMO. So this deceit with Vrbnica, with Morina, with their alleged lists, I hope deserves no further comment.

THE WITNESS: Perhaps we could move the document down again, please. Thank you. There we go. All the way. Excuse me. There we go. The documents came in bunches. This was page 43 of the section that it came in. As I mentioned in my document - excuse me - in my presentation, after the ceasefire declared by the Yugoslav government on the late evening of the 6th/early morning of the 7th, the border was closed on the Kosovar side. It was closed after the ceasefire. And so after the massive flow of the previous days, the guards described to me that when the flow stopped, they closed this section of the document, which is what the note refers to. Again, noting at the top that this is number 43, this is number 43 of its block, its section of documents, but the documents did not always follow a day or two-day period. They seemed to correspond to shifts or to moments when the guards could catch their breath. They would stop a numbering sequence, put a band around it and put it in the file, start a new sequence thereafter. So I just want to clarify the confusion in the question that this document, this page by itself, refers to the period from the 3rd to the 7th. That note apparently pertains to this section of documents, which includes many, many others.

MR. MILOSEVIC: [Interpretation] ²²⁴⁹

Q. You have just explained what I have been asserting. That is precisely what you've just explained. You analysed two-day periods of flows, of movements, in order to draw certain conclusions, and now you have said yourself that this is given for several days and that you cannot exactly determine what pertains to which day, or which date, rather --

A. I'm sorry. That's not correct. That's not correct. That's not what I said. The date of the document specifically identifies what day this page refers to. However, the numbering sequence at the top does not always refer to the pages that are relevant to a particular day. The guards went along, dating each document specifically, and when midnight occurred, they would draw a line across, put the new date in the margin. This document is the final page from the 7th of April, in the early morning hours, and the note says this block, apparently, of 43 pages, is part of this section. There may actually be another block before this in this same period. I don't know. We would have to refer to the original images. But the -- again, to clarify, the list of people on each document is clearly identified by date for each document, and then again, when they broke, there's a line across, a new date.

Q. You know everything, Mr. Ball, except for how these documents reached that rubble and who fabricated them.

JUDGE MAY: No need to answer.

THE ACCUSED: [Interpretation] All right. All right. I know about that.

MR. MILOSEVIC: [Interpretation]

Q. I have a question: As far as I have understood, your methods of ²²⁵⁰ analysis were first used for estimating population and legal immigration, when you first started using these methods.

A. The methods I describe here for analysing killing; is that what your question is?

Q. I mean your methods, your statistical methods. They were first used for estimating the population and the number of legal immigrants. That was the primary use of your methods. And they were considered to be innovative, as far as I managed to find out.

A. I don't know to what you refer when you speak of legal immigrants. The methods we have used for analysing killings have been widely used in censuses around the world, yes. That is their central application in scientific demography.

Q. Yes, but your methods were considered to be innovative, actually. That's what I've been saying. Were your methods assessed as innovative?

A. I believe these methods are innovative in human rights. They are not innovative with regard to scientific methods more broadly.

Q. Well, you know full well, just as I do, that "innovative" means that in the scientific community, this is not considered to be reliable.

JUDGE MAY: He's already dealt with all this. We've heard it several times. He's described the methods. He said they were well-known. I don't think we can take this any further.

THE ACCUSED: [Interpretation] I don't know what you're objecting about, Mr. May. I don't understand.

JUDGE MAY: I'm objecting, and in fact I'm stopping you repeating yourself and going over the same ground over and over again. Now, let's ²²⁵¹ move on to another topic.

MR. MILOSEVIC: [Interpretation]

Q. At one seminar, you stated that refugee flows -- I have to continue from where the witness stopped a while ago. You stated that refugee flows did not coincide with mass killings.

A. Yes. That's a conclusion, actually, in my first report.

Q. Is that correct or is it not correct?

A. Given the data I had in January and February of 19 - excuse me - January and February of 2000, I did not believe that the patterns I observed in migration coincided with the massacres, the massive killings reported by the various human rights organisations and in the -- what was then very recent OSCE report, as seen, as told. Subsequently, when we looked at the overall patterns of killings, including the far more numerous killings in small groups - one, two, three at a time - we discovered this very close relationship. It seems that while massive killings of dozens of people may be in a different cycle than refugee flow, the small-scale killings, again much more numerous, do quite closely follow these patterns. This was, I believe, a very interesting result of the additional analysis.

Q. You have been operating with the figure of 10.000 persons killed. You established 4.000, the identity of 4.000. The 6.000 whose identity you did not establish, as for them, do you, as a statistician, accept that these 6.000 imaginary or assumed persons can be manipulated and can be put into various days, whichever way you want in this analysis of yours, in these two days that you've been analysing? Because the number is so big ²²⁵² that you can prove anything with it. Is that correct or not?

A. No, it is not correct.

Q. I expected you to say that it is not correct, but how did you manage to distribute, then, these 6.000 killings, the difference between those whose identity you established and those who you are assuming? How did you take care of that in your statistics? Where did you put them in your statistics?

A. When we did the estimations, we did them in several different ways. Notably, all the ways produced the same overall total or a total very close to it. But in order to analyse the patterns in time and space, we did the estimates for each point in time and space. That is, we did not simply estimate 10.300 and then distribute them about as we wished. Instead, we took each point, each day for each of the four regions, and made -- or excuse me, each time period for each of the four regions, and then made the estimate for that time period/region point. So the distribution of the 6.000 estimated deaths for whom we don't have names occurred as part of the process of analysing each time/space point. So where those 6.000 occurred, where and when those 6.000 occurred, was controlled tightly by the estimation process. It was not something over which the analysts could control.

Q. So you distributed the assumed dead into assumed time points by applying some kind of statistical methods. How can that be a serious way of doing it? Tell me.

A. If we knew all the data, if we had every piece of possible information, this would indeed be a trivial task. Scientific methods have ²²⁵³ been developed over the last several centuries because we rarely have all the data. We know a great deal about the reliability of those techniques. They've been used in many contexts. Many different things have been analysed, from wildlife populations to medical experiments to censuses to surveys. We know the properties of the uncertainty. We know how much we can or cannot conclude, based on analyses that contain missing data. Those techniques have come into play here. And so I believe that this study, done to, in my opinion and I believe in the opinion of my co-authors, done to the highest scientific standards, this study respects that. We respect that there are areas of uncertainty, and in my direct testimony we discussed one point where we believe there is substantial uncertainty. That is how I can be confident drawing these conclusions.

JUDGE MAY: Yes. It's now just after 11.00, and we'll adjourn. Mr. Milosevic, how much longer are you going to be with this witness, do you say?

THE ACCUSED: [Interpretation] You're asking me?

JUDGE MAY: Yes.

THE ACCUSED: [Interpretation] Well, as you know, I do not have the possibility to prepare for cross-examination, so I cannot estimate how much time I need. I'll try to cut some of the questions short, because I think it's quite clear what this is all about, but I won't manage to do that with other questions. I don't know how much time I will take. As much as necessary.

JUDGE MAY: We shall expect you to finish within half an hour. You will have then had the same time as the Prosecution, plus another half ²²⁵⁴ hour.

THE ACCUSED: [Interpretation] Objection, Mr. May.

JUDGE MAY: We are adjourning now. We'll adjourn until half past 11.00.

THE ACCUSED: [Interpretation] I have an objection. [In English] I have objection, Mr. May.

JUDGE MAY: We'll deal with it afterwards.

--- Recess taken at 11.04 a.m.

--- On resuming at 11.36 a.m.

JUDGE MAY: Yes, Mr. Milosevic, your objection.

THE ACCUSED: [Interpretation] I have an objection to the limitation you imposed on my cross-examination which you explained by the time used by the other party. I think that anyone can see it's totally incorrect, because the other party came here with an extensive written alleged expert report, so they didn't need to go into detailed examination. They just scanned through this exhaustive material, and in my cross-examination, I have to deal in detail with the material that the other party has presented here. So it's absolutely incomparable. On the other hand, the other party's trying to go as fast as they can through all these computer-made diagrams, maps, tables, and create the impression that all this fabricated data constitutes a complex of scientific argument supporting the indictment, and it is evident that it is very easy to refute that great errors have been made and that the data itself originates from the other warring party and had a vested interest in fabricating such data. ²²⁵⁵ So my objection is that you cannot limit my cross-examination. I don't know how long I will take, but I certainly cannot finish in half an hour.

JUDGE MAY: Mr. Milosevic, the reason that time limits are being imposed in this case is in order to make as expeditious use of the time as possible. Now, we bear in mind, first of all, that you have been defending yourself and therefore are entitled to some leeway, first of all, because you have to prepare your own cross-examination; and secondly, because you're not a professional advocate. All that we bear in mind. On the other hand, what is not permissible is the use of cross-examination as a vehicle for making speeches, for making comment, and for repeating evidence which has already been given. All that is impermissible, and when you indulge in it, you'll be stopped. Now, we have in mind that this is important evidence, and we have in mind the point that you make about a report and this being your only opportunity to refute it. But in my judgement, and I speak for myself, a great deal of the time so far has been wasted. What is required is relevant questions. You claim that the evidence is fabricated. You should put relevant questions to deal with that. Now, in order for us to manage the time effectively, it would be helpful to know how long you anticipate you need to put your questions. It shouldn't be a difficult question for you to answer.

THE ACCUSED: [Interpretation] I have told you that it depends on the course the examination takes. Certainly not too much time, but on the other hand, half an hour is certainly not enough. I'll try to make it as ²²⁵⁶ expeditious as I can. It's not in my interest either to make this longer than necessary.

[Trial Chamber confers]

JUDGE MAY: We'll go for half an hour and then we'll review the position.

Do you have any questions, for the amici?

MR. KAY: There will be some questions, probably 15 to 20 minutes, in addition. If Mr. Milosevic covers those points, we won't be repeating them.

JUDGE MAY: Thank you.

MR. MILOSEVIC: [Interpretation]

Q. During the break, I checked some assertions that you denied, and I would like to ask you a few questions about this. Namely, I asked about your cooperation and adjustment of data to the data of the International Crisis Group, and you said that was not true. However, on the website of your AAA association, and that is website hrdataaas.org/kosovo/index/html [as interpreted], titled "Political Killings in Kosovo from March to June 1999," in the column called "Statistical Analysis of Data," it says: The method of killing people in Kosovo coincides with migrations, and this claim corresponds to the data obtained from the International Crisis Group; and then others are enumerated as well.

Another heading says: The role and consequences of cooperations with NGOs regarding the Kosovo conflict efforts have been made to synchronise data gathered from the International Crisis Group and others. Third, when you enter the website of the International Crisis ²²⁵⁷ Group and when you type "Patrick Ball," you get a report on Burundi. Will you please tell me: Is this true or not?

JUDGE MAY: Which part are you asking about?

THE ACCUSED: [Interpretation] About all the three parts, because the witness said he had no cooperation whatsoever with the International Crisis Group, and from their own website and from the website of his association, we can see that the questions that are being dealt here have been harmonised with them. And my reason for asking is that it is common knowledge what this International Crisis Group represents.

JUDGE MAY: Just a moment. Let the witness answer. Can you deal with that, Dr. Ball?

THE WITNESS: Let's deal first with Burundi. That's the first one. I have no idea what my name is doing in reference to anything having to do with Burundi. I have never done any work on Burundi or, actually, on any of the countries in that region, with International Crisis Group nor with anybody else, so I have no idea why my name is there. With regard to the report on the website of my organisation, the report is "Political Killings in Kosovo," as you titled it. That's correct. It is the intervening report between the migration report and the current report. We sought the cooperation of the International Crisis Group, but they declined. The reference that you have found may mean that -- I mean, I don't remember the exact quotation and I don't have the report in front of me, but I think it means that we find our results to be roughly comparable to what they have reported. We did not, I repeat, use any of their data, simply because they didn't give it to us. ²²⁵⁸

MR. MILOSEVIC: [Interpretation]

Q. All right. That was one of the questions I wanted clarified. Let us come back to the map which shows that it's not only an omission or an accidental mistake when you put Morina on the Vrbnica border crossing. On that map, 2.1, in the report from January, you drew Morina in the territory of Albania. That's one element.

Second, on that Vrbnica border crossing --

JUDGE MAY: Just a minute. If you're going to ask questions, where is this map?

Dr. Ball, do you have it?

THE WITNESS: Yes, Your Honour. I believe he's referring to a map not in the report that's been entered into evidence but, rather, in this report "Policy or Panic?"

MR. NICE: I think I have enough copies of this report for the Chamber to look at it. And it can be found, I think, on page 12. I also have other maps, but I'll deal with those in re-examination. I'll hand in my colour version and I'll stick with black and white for the time being. Maybe this can be dealt with comparatively informally at this stage and decide on whether the document needs to be exhibited in due course.

JUDGE MAY: Yes. We have the map on page 12. Dr. Ball.

THE WITNESS: Yes. The question, I think, is: Why is Morina on the Albanian side of the border? As I answered before, in my conversations with the border guards, in translation, they referred to the small cluster of buildings and homes there only a few hundred metres from ²²⁵⁹ the border as Morina. And so my reference to the town of Morina -- or it's not really even a town, it's a tiny number of buildings and homes, refers here to this point, as I pointed out on the map in the earlier conversation.

I do not refer to the other considerably larger town to which the defendant has earlier referred.

JUDGE MAY: So we can have this clarified; according to this map, the crossing is on the road from Prizren to Kukes. Is that right?

THE WITNESS: That's correct, Your Honour.

MR. MILOSEVIC: [Interpretation]

Q. That's -- that's where Vrbnica is. That's not the location of Morina.

Now, about that Vrbnica, there was no Albanian border service there. After that crisis in Albania in 1997, that building was abandoned, looted, and shepherds and cattle were there, not data. That's my assertion.

JUDGE MAY: Dr. Ball, tell us what you saw when you went there.

THE WITNESS: Yes. I went there and there were thousands of people, including dozens of -- perhaps hundreds, I'm not sure, of journalists from all over the world.

Here at this point in the road south of Prizren, there is a border crossing. There are buildings on both sides of the border, on the Kosovo side and on the Albanian side. At the time I collected this data, I did not cross the border. I was only on the Albanian side. There's a small border post here, and as I suggest, it's ²²⁶⁰ immediately on the border. It's drawn here a little bit off the border, but it's actually immediately on the border. There are several other houses there, farmers. But at the time of the -- which is covered by this report, March to June of 1999, there was a great deal of traffic on this road from Kosovo into Albania, and there were many international observers as well as the Albanian border guards here. It's not abandoned. There's quite a few people there.

MR. MILOSEVIC: [Interpretation]

Q. In that previous report, page 8.11, you said only three persons had data from the border, which later went from Tirana to the database. Who are those three persons?

A. Which note do you refer to, sir?

Q. That is in the previous -- the earlier report, page 8, item 11. It says: "Only three persons had the data. They went later from Tirana to the database."

My question is, and there's no reason to look for them there: Who are those three persons who were the only ones who had the data?

A. Who handled the data at the border, there was myself and my translator and the young man who assisted us. We conveyed -- I personally conveyed the data to Tirana, whereas I said in my earlier reply the data were typed or keyed into a database.

Q. So you are one of those three persons, and the other two were your assistants. There was no one else; right?

A. I see your note here. Since they were keyed, we're talking about three different people. The note you're referring to specifically, which ²²⁶¹ is actually quite important, I believe, for your question, says that: "Only three people have had access to the border data since they were keyed." Okay? So this refers to a different step. The three I described earlier were the three people who obtained the data and conveyed it to Tirana. Okay. There were four people in Tirana who did the coding process, which I can describe in a moment, but also the keying, the point at which the data are actually typed into a database.

The three people that - excuse me - had access to the data after it were keyed were myself and two people who assisted me in Washington. At no time in that process was the data out of my immediate control after they were sent from Albania to me in Washington.

Q. In that first report, on page 1 and page 29, on page 1 you say in the first paragraph that killings were not a mass phenomenon, and on the other hand, in those places where there was killings, there was no refugee flow. So there is no correlation between the two. Is that correct?

A. No, it's not correct. I think I answered this before. My conclusion in this report was that massive killings did not coincide with the heaviest points of refugee flow. By "massive," I was coming from Guatemala, where I was involved in the statistical analysis of the killings of hundreds of people, and so I was sensitive to - excuse me - sensitive to the difference between massive killings and individual or selective killings.

The data available when I wrote this report, which I wrote between September and February of 1999/2000, the data I had then were about the ^{2262 BLANK PAGE 2263} instances of massive killings, dozens of people at a time. Those massive incidents did not correlate precisely with the flows that have been documented in this report.

Subsequently, as I described in my earlier reply, we have discovered that the far more numerous small-scale killings of one, two, three people, small numbers of people at a time, those far more numerous killings do in fact correlate quite closely with refugee flow. So this later finding, including vastly more data, is much more precise and enables us to reconsider the earlier finding.

I think it is good scientific practice, when one finds something which contradicts an earlier finding, to make that clear and explore why there's a difference, and I believe that the explanation I've just given you and that I gave in my earlier reply explains why we have observed this difference.

Q. Yes, but you encompassed there not only those whom you identified, that is, 4.000, you included all of 10.000. So the assumed 6.000 were distributed across the assumed time period; is that correct?

JUDGE MAY: You've already answered that.

MR. MILOSEVIC: [Interpretation]

Q. All right, then. Did anyone intervene to make you change that as compared to your earlier report?

A. No. In fact, I remember well the moment of astonishment that I felt when I saw the killing graph for the first time. When I saw it, I assumed I had made an error, because the correlation was so close, and I had hypothesised earlier that the correlation would not be close. When I ²²⁶⁴ found that, I immediately began every measure I could consider to test whether I had made an error. So I separated the data sets on to different computers and re-ran the analysis, I checked it according to several different ways of calculating the two-day period totals, and I found the correlation to be quite robust, by which I mean it appeared in each of the analytic methods and techniques I employed.

Q. And what do you mean when you say "datum" or "data"?

A. In this case, the data include, as I have described, the records from the border guards, in combination with the counts provided by the UNHCR, OSCE observers, and reported that the actual data I used were the reports UNHCR made available on a daily basis in Tirana; the various surveys that I conducted and that were conducted by partner organisations and shared with me, again, as I described yesterday; the lists of deaths that were used to prepare the analysis of killings, again by ABA/CEELI and their partners; the exhumation records; the interviews conducted by Human Rights Watch and the interviews conducted by the OSCE - those are the killing data - and as we have described and discussed at some length this morning, the data analysing KLA activity; and finally, the data published by Yugoslav press sources in the Yugoslav government on NATO airstrikes. These are what I call data. There are intermediate data points when one does the analysis, but these are the data, I believe, in the sense implied by your question.

Q. Let us now move on to a concrete question that you have been trying to deal with statistically. In practical terms, as your paper shows, you have put three questions that you need to answer. The first is ²²⁶⁵ whether they were fleeing from the conflict, as you say, between the Yugoslav troops and the KLA; then whether they were on the move in order to avoid NATO airstrikes; or was their departure the result of an ethnic cleansing campaign? You put these questions with the aim of establishing which was the cause out of all of these three elements; is that right?

A. Let me refer to the actual language of the hypotheses in the report to correct some of the minor misstatements made in the question. So I don't want to take as read those -- the statement of the hypotheses by the question. The hypotheses are in the report. The purpose of a statistical analysis is not to find a cause, but rather to evaluate hypotheses and to determine whether data are consistent or inconsistent with the hypotheses. This is actually a somewhat different logic than the question implies. We cannot sustain a hypothesis or prove it. That's outside the realm of these techniques. Instead, we can reject hypotheses if the data are found to be inconsistent. As I explained yesterday, we rejected, of these, two hypotheses. We reject the hypothesis that action by the KLA, as measured by the variables we described, motivated Kosovars to leave their homes. We reject that hypothesis, with a few qualifications. We also reject the hypothesis that NATO airstrikes created local conditions or directly or indirectly motivated people to leave their homes. We find - and this is I think the point of most important clarification - we find evidence consistent with the hypothesis that Yugoslav forces forced people from their homes, forced Kosovar Albanians from their homes, and killed people. We find that evidence to be ²²⁶⁶ consistent with it. And our observation about the consistency has to do with this striking coincidence of the dramatic decline in people leaving their homes and being killed at the very moment that the Yugoslav government announced a ceasefire. This is a striking coincidence, and especially in light of the --

JUDGE MAY: You've made the point.

MR. MILOSEVIC: [Interpretation]

Q. All right. It's not three questions. You have reduced this to three hypotheses. Have we clarified that matter now?

A. Yes, they were three hypotheses. That's correct.

Q. Are you aware of the fact that on the 24th of March, 1999, a war started, waged by NATO against Yugoslavia?

A. Yes.

Q. Do you assume that a war is a very complex situation and from many points of view it is also chaotic, and it is therefore difficult to simplify matters and to reduce everything to three hypotheses only, the way you had done it?

A. We look at the data with the idea that there may be some broad patterns in the data on refugee flow and killing, and those patterns may be quite clear. The clarity of the patterns may be surprising in light of what seems to be chaos. I think that this is one of the points of science, is to find patterns and clarity in the midst of chaos. When we look at these patterns, they are in fact much less chaotic than would be implied by the question. These are the bases for our conclusions.

Q. Yes, but you assume that such a complex and chaotic situation, as ²²⁶⁷ war is, cannot be reduced and simplified to three hypotheses only. Can you assume that people were fleeing from the war, the war, the war that included all sufferings, all fears, not only one of the three fractions that you turned into hypotheses and that you wish to single out? Are you aware of that?

A. Is the --

JUDGE MAY: Rephrase the question. May there be other reasons than the hypotheses which you used for people to have fled?

THE WITNESS: Yes, there may. These seemed to us the most obvious hypotheses since these were the three parties with significant arms involved at this time. However, this method does not exclude the possibility that there may be other causes.

MR. MILOSEVIC: [Interpretation]

Q. Yesterday you said that the subject of your statistical survey was actually to establish the common cause of exodus, the common cause, the reason. That is the wording you used, I believe.

A. We used the word "common cause." We did not use the phrase "establish." What we are looking for in the analysis of common cause is we are looking for patterns which move together over time and space. Such coherence, the connection of patterns over time and space, correlations, in the statistician's language, strongly suggest the existence of a common cause. They do not establish it, nor do we make the claim that that is established. We conclude, based on such a strong suggestion, that such a cause existed. But again, the scientific language is quite formal here.

Q. Of course. This is not even contested that there was a common ²²⁶⁸ cause, but did it occur to you that the common cause of the exodus was war? You said there was a cause and then this cause was stopped. There was a war going on from the 24th of March onwards and then that war stopped. While the war was on, the exodus was going on as well. When the war stopped, the exodus stopped.

Did it occur to you that this common cause was the war, not this - how shall I put it? - this quest and elimination between and among only three hypotheses that you have put forth by simplifying an extremely complex phenomenon as war is?

A. Is the question that when the war stopped, the refugee flow and killing stopped? Is that the question? I'm sorry.

Q. Yes. I'm asking you, did it occur to you that this common cause was the war? Because everything happened from the 24th of March when the war started. And then when the war ended at the beginning of June, when this common cause ceased to exist, all of this stopped. The war stopped; all of this stopped.

JUDGE MAY: Just let the witness answer.

THE WITNESS: Actually, it did occur to me. Excuse me. So we tested it. We tested the idea given the data we had. In particular, we were interested in what happened during the ceasefire period, the ceasefire declared unilaterally by the Yugoslav government - excuse me - on -- again, as I said, on the evening of the 6th of June, the early morning of the 7th. So I looked at the four-day period from the 7th to the 10th, and again as we've discussed, there was relatively little -- there were relatively few killings or - and little refugee flow. The ²²⁶⁹ graphs are clear about that.

What's fascinating to me is that the number of -- the number of KLA activities recorded and the number of NATO airstrikes documented again by the Yugoslav government and press sources, both of those patterns increase dramatically, doubling and tripling relative to their numbers in the earlier period. So it in fact was a unilateral ceasefire during this period since these other two parties dramatically increased their activity at the time when refugee flow and killing declined to tiny fractions of their earlier levels.

So while war, in a very broad and, in my opinion, vague sense may be a cause, what is clear is that the statistical relationships are not equal here. One party declared a ceasefire, the refugee flow and killing declined drastically, but at the same time, KLA and NATO activity increased drastically. There is no corresponding increase in the patterns of killing and refugee flow.

This helped us to be increasingly confident in our rejection of the hypotheses that I earlier stated.

Q. I have been talking about simplification in terms of three hypotheses of this common cause that is called "war." It lasted, it produced these consequences, it stopped, and then the consequences stopped. Are you aware of the fact that in this war, it was not Yugoslavia that attacked itself?

A. There was a conflict that -- between NATO and Yugoslavia, and there was clearly an insurgency by the KLA. What language in formal legal terms you use, Mr. Milosevic, I don't know. ²²⁷⁰

JUDGE KWON: Okay. It seems to me that what is suggested by the question is that the people might have fled because of war and not a single factor of that.

THE WITNESS: Uh-huh.

JUDGE KWON: How could you simplify those complex factors into three simplified hypotheses? That seems to be the question. Could you answer it? Could you help us in that sense?

THE WITNESS: I do so by means of a theory, which is that if -- if -- in the presence of war, there would be -- certain things would occur, and those things should leave us some way to count them. Some of them would be countable. We look for things, in fact, that are countable as indicators of our -- of our theory about what's going on. The indicators we found were battles between the KLA and Serb forces. This seemed to be a reasonable indicator of the local presence of this broader idea called "war." Similarly, casualties inflicted on Serb forces by the KLA would be an indicator of the presence of this broader thing called "war."

I don't claim that these indicators, the things we can count and measure - deaths, battles - I don't claim that these indicators encompass the complexity of this enormous idea, but they should be present.

JUDGE KWON: Thank you. You may go on, Mr. Milosevic.

MR. MILOSEVIC: [Interpretation]

Q. So then these first two hypotheses, that is to say NATO and the KLA, is it clear to you that, in this war, NATO and the KLA were acting with the same objective in mind? Did you give any thought to that? ²²⁷¹

JUDGE MAY: Can you assist or not?

THE WITNESS: I did not give any thought to that idea.

MR. MILOSEVIC: [Interpretation]

Q. And now, now can you look at this idea, this concept that NATO and the KLA worked with the same objective in mind, that is, a single hypothesis, the concerted activity of NATO and the KLA?

JUDGE MAY: He did not consider that. Therefore, he cannot give evidence about it.

MR. MILOSEVIC: [Interpretation]

Q. All right. Even if we are not bringing together many things but, rather, separating them, we are looking at a complex issue such as war. Even if we were not to link them together at all, do you think that this question boils down to combat action only of the Yugoslav army and police, of the KLA, and of the NATO bombers respectively? Or this complex phenomenon called war, does it contain many other phenomena that also affect --

JUDGE MAY: You have made this point. He has answered the question. Now, kindly move on to another topic or we'll bring this to an end.

THE ACCUSED: [Interpretation] That is the core of the matter, because the witness himself has said that it is certain that there are many other elements.

MR. MILOSEVIC: [Interpretation]

Q. I am going to mention some elements now, and you're going to tell me whether -- ²²⁷²

JUDGE MAY: He said there may have been other elements. These were the three that he concentrated on. Now, if you want to put some other elements, you can, but the witness may not be able to help.

MR. MILOSEVIC: [Interpretation]

Q. But I want to put a question to the witness. So other phenomena that affect people are, in my opinion, political means, propaganda, threats of the KLA, orders of the KLA, the media war that waged, et cetera, and the leaflets, the black and red ones that were thrown, in the colour of the Albanian flag through which the citizens were ordered to leave the territory of Kosovo. War is not only a bullet and a bomb.

JUDGE MAY: No. Ask a question. I told you before, it's not a time for making speeches.

Now, the question appears to be this: Were there, in your view, other matters or did you consider other possible causes for, it seems to be, the refugee flight, namely, propaganda and threats of the KLA?

THE WITNESS: This was an occasional topic of discussion among my co-authors and myself. We did not consider the effect. What we did, however, is observe that propaganda is unlikely to cause killing in the same way, and thus when we found the close correlation, the correspondence between the patterns of killing and the patterns of migration, we felt that that -- that the propaganda explanation was unlikely.

MR. MILOSEVIC: [Interpretation]

Q. Why are you simplifying matters? Why are you reducing it to propaganda only? I'm also talking about threats that were not empty threats only, because the KLA killed many Albanians as well in order to ²²⁷³ discipline them. I'm talking about orders issued to the population to leave --

JUDGE MAY: Let's deal with it one by one. The next question is threats. Did you consider threats made by the KLA?

THE WITNESS: No. We considered it unlikely on the same basis.

MR. MILOSEVIC: [Interpretation]

Q. So this has no effect, propaganda, calls, threats, orders issued by the KLA to the population to leave the territory. Is that the way it is in your opinion? Or the media war either that was part of this war as well?

A. As I said and I think as is clear from the report, we did not consider these hypothesised causes presented by the question in a direct way. However, given what we have considered, it seems to me that they are unlikely causes.

Q. All right. But let's think logically about this. You have linked all reactions to combat action only, to fire, as if we were talking about a herd that reacted to shooting only while the shooting was going on, as if it weren't people who were at stake, who think about all these things, who have reason, who are susceptible to pressure, to instructions, to media manipulation. Homo sapiens think and bear in mind all these elements, not only the momentary situation when somebody is shooting and somebody is killing. Do you bear in mind this complexity of the situation that you have simplified, reducing it to these three hypotheses only?

A. Certainly we considered the effect of something that has happened ²²⁷⁴ in the recent past in addition to the events that are occurring at the immediate time. We had an extended discussion of this yesterday around the statistical term of a "lag." Again, that means that we considered the effect of KLA activity and NATO airstrikes in the previous four days, the current period, two-day period, and the previous two-day period. It is in this way that we have considered time.

I did additional analyses that go under the statistical language of - pardon the jargon - autocorrelation, which looks at patterns over time, looks for those sorts of relationships. The one that I found that was meaningful in this analysis was the immediately-prior period. It only appeared to be meaningful in the variables we discussed yesterday about the KLA having -- relating to migration in certain regions at certain times, but that's where it appeared.

Q. For example, are you aware of the columns of Albanian refugees who were returning to their villages after the bombing, and they were bombed along the way by NATO as they were returning? Do you think these were messages as well, that orders had to be carried out, that Kosovo had to be left?

JUDGE MAY: Do you understand the question?

THE WITNESS: I think so. I think the question is am I aware that NATO bombed Albanian refugees on the road.

JUDGE MAY: That seems to be the first part, yes.

THE WITNESS: There were some incidents of that kind reported in the Yugoslav government and press sources. We counted them in our data.

MR. MILOSEVIC: [Interpretation] ²²⁷⁵

Q. I am asking you about the effect of such a message, when a column of peasants who are going back to their village are bombed and at the same time there are orders and requests put through the media that Albanians should leave Kosovo. Did you take that into account as an effect of this message or did you only add up the persons killed by NATO in these columns?

A. We did not take that into effect. And to clarify, the analysis of the NATO airstrikes is not number of people killed; it's number of airstrikes.

Q. The number of airstrikes is something that we will look at later, but by enumerating all of these individual elements of war - far be it that I have managed to mention all of them - the question is: Is it clear enough that looking at all the phenomena related to war in their cumulative effect, can be defined as a common cause and that this cannot boil down to --

JUDGE MAY: You have been making this point over and over again. The witness has dealt with it. Now, in light of your application for more time, we're prepared to give it to you until the adjournment. There will be no more time after that. It will give you nearly three hours to cross-examine this witness, which is more than adequate.

THE WITNESS: Your Honour, may I have two or three minutes of personal privilege if we're going to continue?

JUDGE MAY: Yes. We'll adjourn for five minutes.

THE WITNESS: Thank you.

--- Break taken at 12.29 p.m. ²²⁷⁶

--- On resuming at 12.35 p.m.

JUDGE MAY: Yes, Mr. Milosevic.

MR. MILOSEVIC: [Interpretation]

Q. So we are discussing the inviability of these simplified premises in statistical analysis, even if the data were accurate, and they simply cannot be, in view of their source. Do you suppose that all these complexities of phenomena encompassed by the war, as a fact, as a phenomenon, should be explored for a scientist to arrive at valid conclusions?

A. This study sets out very specific hypotheses and it makes conclusions based on the best data available to us. I am not competent to enter into a larger methodological or epistemological debate.

Q. All right. If we are talking about statistical data, does the fact that this report is based on records made by immigration officers on the Morina border crossing and research made in refugee camps in Macedonia and Albania, and even Bosnia and Herzegovina is mentioned, does this put a question mark over the validity of your report and its conclusions?

A. No.

Q. And does the question arise in your mind, as the person who researched this: Who are these administrative officers on the border crossing of Morina who were able to engage so quickly after the aggression in drafting these records? And you said the largest report came in the first stage, in your own words.

A. These were the regular border guards who worked that post. They may have been supplemented with guards brought from other posts, given ^{2277 BLANK PAGE 2278} that this post was experiencing such heavy flow.

Q. At this first stage, do you think that the issue of this readiness and alertness of border guards or clerks was important, the fact that they were ready to deal with this very large inflow and quickly make records?

A. I don't know if they were ready in some sense. They were able to do a good job within the limitations of their personnel and capabilities. I discussed in some detail yesterday how I evaluated their work.

Q. Did it cross your mind that it can only be explained by the fact that the appearance of refugee columns were preplanned and organised in advance in order to create an impression around the world that they were a product of an ethnic cleansing campaign, and other things that you are claiming, rather than the war in all its complexity, as is obvious?

A. There are several questions there, some of which I think I answered earlier, about the war's complexity. But on your new question, did it occur to me that this was preplanned, the answer is no.

Q. Will you tell me: Who could have expected these records to be accurate and truthful, knowing the objectives of the Albanian secessionist movement and the attitude of Albania towards Serbia and Yugoslavia and the so-called Kosovo problem?

JUDGE MAY: What is the question?

THE ACCUSED: [Interpretation] Precisely what I asked.

MR. MILOSEVIC: [Interpretation]

Q. Who could possibly expect these records to be truthful, knowing the objectives of the Albanian secessionist movement and the attitude of Albania towards Serbia and the so-called Kosovo problem? ²²⁷⁹

JUDGE MAY: He's dealt with these questions. Let's move on.

MR. MILOSEVIC: [Interpretation]

Q. You have said that the research which has allegedly been conducted in refugee camps were based mainly on interviews. Who conducted these interviews?

A. We reviewed the sources of the information in these reports several times. But to review again, in answer to this question: The surveys conducted to supplement the migration analysis were conducted both - excuse me - first, by teams that I organised myself; second, by interviews conducted on a random sample population basis by Physicians for Human Rights; and third, interviews of refugees conducted by Human Rights Watch.

In the second phase of the study, the analysis of killing patterns, there are three interview-based sources in addition to the exhumation records. The first were the interviews conducted by the American Bar Association, Central East European Law Initiative and their partners; second, interviews conducted by Human Rights Watch; third, interviews conducted by the OSCE. These are the interviews conducted in order to support this -- in order to do this. They were conducted by very different organisations, at different times and places, and yet they tell a remarkably similar story.

Q. We are talking about --

JUDGE ROBINSON: Mr. Milosevic, just a minute. Who were the people interviewed?

THE WITNESS: Kosovar Albanian people who were either in refugee ²²⁸⁰ camps or back in Kosovo, depending on which phase of which survey we're discussing. They were Kosovar Albanian civilians.

JUDGE ROBINSON: I'm not sure if this is what Mr. Milosevic was inquiring about, but I would be interested, in any event, to know whether your analysis took account of the -- what would be a natural prejudice on the part of the interviewees.

THE WITNESS: We did not take account of it in a direct sense; however, statistically what I think is interesting is that our method for the killing analysis depends on finding the same people reported by the same or different witnesses to different projects. We found a relatively high rate of those overlaps among different reporting processes. That is to say, the same victims appeared in Human Rights Watch interviews, OSCE interviews, ABA/CEELI interviews; and in exhumation records, not always in all of them, but you see a very high rate of overlap among two systems or among three systems. If people were making this data up, they would have had to coordinate a fabrication among more than 15.000 interviews conducted over an almost two-year period and given to four different -- or three different organisations, as well as fabricating exhumation records. The level of fabrication required is, in my opinion, deeply implausible. It is in that way that, at a statistical level, we are confident of the robustness, that is to say, the resistance of our findings to fabrication.

JUDGE ROBINSON: Yes, Mr. Milosevic.

MR. MILOSEVIC: [Interpretation]

Q. I think precisely the opposite. I think they were coached, because the statements, depending on their education attainment level, ²²⁸¹ were identical. But let us make one thing clear. We are talking about an interview as a methodological procedure, requiring a very qualified and unbiased approach. Is that correct?

A. For the purposes to which we put these interviews, all that I required was that people write the story down. We're looking here just for a list of people described as dead, and the time and place of the witness's best recollection of where and when that person was killed.

Q. You know very well, being a sociologist, that the interview, as a methodological procedure, cannot be conducted with one single question: Please describe what happened.

A. The interviews were done in different ways. As you'll find in Appendix 1, there's some discussion of the interview process. The Human Rights Watch interviews were conducted by a qualitative method, which proceeds along the lines of the interviewee's story. The ABA/CEELI and OSCE interviews were conducted using a -- to use a tiny bit of jargon here, a semi-structured interview process, which asked open-ended questions, for example: Why did you leave your home? The witness would report, in some phrases, his or her reply. Those would be recorded. The interview - or excuse me - the questionnaire process for the OSCE and the ABA/CEELI teams was quite similar.

Physicians for Human Rights used a much more tightly structured interview process.

Again, the only use we put those interviews to in this work was to calculate from the story the time it took from someone's departure from their home to crossing the border. That's the only use we put those ²²⁸² interviews to in this study. Although the interviews that I designed and that we did in the refugee camps in Albania and other refugee-gathering locations in Albania, although that was a much more elaborate interview, the only use we put it to in here was to calculate the transit time, as I just described - the time it takes people to get out from their home, to get to the border - as well as to provide a check on the distribution of people's origin locations over time, as reported in the border data. These methods are described in "Policy or Panic?"

Q. Precisely what you have just mentioned, that you used their statements to establish the time of departure and the time of border crossing. You said yesterday you had established that the overwhelming majority crossed the border the day following the date of departure. Is that correct?

A. No. I'm sorry. What we established is that the overwhelming majority of people had crossed the border on the same day -- not the overwhelming majority. More than half the people crossing the border on any given day left their homes that same day. That was our -- what we discovered.

Q. That is even less likely than what I mentioned in my question, because that Morina crossing, and even the Vrbnica crossing, make it implausible for anyone to cross the border on the same day as leaving their homes, if we bear in mind the enormous mass of people. Do you know anything about the distance, the difficulties of the journey? Do you understand that it's very unlikely that such a great number of people at that first stage crossed the border on the same day they left their ²²⁸³ homes?

A. There are two independent counts of the number of people crossing the border at that point. The numbers largely agree. And when they disagree, the numbers reported by the UNHCR and OSCE observers is always higher, always greater than what was registered by the border guards. So I can't address the plausibility of whether it happened except to say that I think that all available evidence says that it did happen. So plausibility is no longer an issue.

Q. I understand that plausibility is not your area of expertise, but it's -- when I'm saying that it's unlikely for them to have made a move on the same day as crossing the border, then it puts a question mark over their other statements if something that is physically impossible is mentioned.

JUDGE MAY: Ask a question, Mr. Milosevic.

MR. MILOSEVIC: [Interpretation]

Q. Can we consider, in view of the fact that we agreed that the interview as a methodological procedure has to be conducted neutrally and by experts, by qualified people, can we consider as neutral the way that interviews were conducted by the party which perpetrated the aggression?

JUDGE MAY: No. That's not a proper question. THE ACCUSED [Interpretation] The question is very clear. I don't know if it was put correctly.

JUDGE MAY: You can change the suggestion that it was conducted by the party which perpetrated the aggression. That is not a proper question. ²²⁸⁴ What you can deal with, Dr. Ball, is this: Do you consider as neutral the way in which the interviews were conducted?

THE WITNESS: For the purposes of obtaining the data required for this analysis, the interviews were neutral.

MR. MILOSEVIC: [Interpretation]

Q. And in which way -- except for the statements of refugees, what else is there to support the claim that only a few number -- that only a small number of Kosovar Albanians fled Kosovo because of NATO bombing and that it followed certain constant patterns so it had to be coordinated? What is there to support this claim?

A. Well, I believe that is a conclusion from my first report. However, I would point to the current report that we're looking at as evidence here and say that we have now given the analysis of refugee flow a thorough statistical test by two different measures and found conclusively and consistently that NATO airstrike patterns do not explain the patterns seen in refugee flow. Furthermore, the close correlation of refugee flow with killing patterns over both time and space suggests more strongly than in the first report that there is some sort of coherent or common cause of the two phenomena.

Q. Yes, but NATO's airstrikes were also coordinated with the activities of the KLA on the ground, and they constituted one comprehensive form of attack on Yugoslavia.

JUDGE MAY: No. That is a speech. Now, have you got any more questions? Because the time is practically finished.

MR. MILOSEVIC: [Interpretation] ²²⁸⁵

Q. The main conclusion that Yugoslav authorities planned and carried out a campaign which was centrally organised in order to ethnically cleanse from Albanians at least certain areas of Kosovo, as claimed, is not proven in any way except by hypothesis. So I'm asking you: If the Yugoslav authorities planned and carried out a centrally organised campaign, where is that plan? What is it called and who made it?

A. My conclusion here and the conclusion of myself and my co-authors is that the evidence is consistent with the hypothesis that Yugoslav forces conducted a systematic campaign of killings and expulsions. Again, we have found the evidence to be consistent with the hypothesis, and I've described in some detail why and how we reached that conclusion. That is the extent of my expertise, is to speak to the statistical evidence related to the hypotheses which we proposed. Beyond that, I'm unable to comment further.

Q. All right. But you are aware of the statement of one of NATO Defence Ministers, the German Minister, Rudolf Scharping, who said that there was a plan, the Horseshoe Operation, and this claim was later refuted even by his own associates as a lie. Are you aware of that?

JUDGE MAY: Do you know anything about this?

THE WITNESS: I read about it in the press, that's all.

JUDGE MAY: He can't deal with this.

MR. MILOSEVIC: [Interpretation]

Q. Fine. If there is no such plan, and there is none, how can this campaign be conducted in a planned and organised manner?

JUDGE MAY: He's here to give statistical evidence, which he's ²²⁸⁶ given. He cannot answer that question.

THE ACCUSED: [Interpretation] All right.

MR. MILOSEVIC: [Interpretation]

Q. How do you explain that if everything you are presenting in statistical terms -- if I may ask you to agree with certain things, if we added up, in terms of maps 1, 2, 3, and 4, it turns out that 471.000 people crossed over into Albania. So where is the difference, the difference of 850.000 people mentioned in the indictment? It is unimaginable that all of them crossed over to Macedonia or Bosnia and Herzegovina.

A. The number to which you refer includes border crossings other than the one that I studied. The number of total refugees, 850.000, does not come from my most recent work. I can't comment on it. However, it's important to remember that refugees went places other than just Macedonia. They went to Montenegro. Some went to Bosnia-Herzegovina. So should you seek a larger number, it seems appropriate to look at all the destination countries.

Q. Very well. So it's not within the area of your interest. But dealing with what you have, how do you explain the large difference in the number of refugees, according to your maps 1, 2, 3, and 4, 471.000, and the diagram which indicates 248.000 people? How do you explain that difference?

A. Where does 248 come from?

Q. I mean the chart that you show for the first, second, and third stages. ²²⁸⁷

A. Can you give me a --

Q. You have the right -- I don't see how you marked the chart, but this is a rectangular chart. 25th March, 12th April, 24th May. That is the chart which shows in the first stage, second stage, third stage, and when you add all the numbers up, you get 248.000.

A. Can you tell me which report it's in --

JUDGE MAY: Can you assist with this at all?

THE WITNESS: I don't know what he's referring to. I would look at it, but I don't know.

JUDGE MAY: Mr. Milosevic, the witness can't assist with this matter. Now, you've got one more question to ask him and then we must adjourn.

THE ACCUSED: [Interpretation] I have more than one question, Judge May.

JUDGE MAY: We have given you a time limit and it's now come, so you've got one more question.

MR. MILOSEVIC: [Interpretation]

Q. Please, this is very important. One of your claims is that NATO bombing did not produce waves of refugees, while your own data go against it, because the percentages of NATO airstrikes coincide with the numbers of refugees by stages.

I will read to you the data. You said the number of targets -- you mentioned 942, with the proviso that at the first stage the number of targets was 541, and that is from the 24th March to the 6th of April. Out of the total number of targets, that is 58.55 per cent. And in that first ²²⁸⁸ stage, from the 24th of March to the 6th of April, you quote in your text that 236.000 people fled, which means at the first stage, according to your own text, amounts to 61.14 per cent, which is an extremely large degree of coincidence. 58.8 -- 55 per cent of targets --

JUDGE MAY: [Previous translation continues]... just break this up so the witness can follow it. Now, what is being said? Just break it up.

MR. MILOSEVIC: [Interpretation]

Q. Do you follow me?

A. If you can refer me to what you're looking at; which report, what page number, what table?

Q. I made for myself, because I have no resource to any assistance here, I made copies of your report, but you know that what you wrote in your text that in the first stage, from the 24th of March to the 6th of April, 236.000 got out. In the second stage --

JUDGE MAY: Just pause -- pause there. Do you agree with that figure?

Just a moment. Just a moment. Let him deal with it. Do you agree with that figure?

THE WITNESS: I think that is approximately correct. We'd need to refer -- I don't want to be held to that number without checking it here, but I think that's approximately correct.

JUDGE MAY: Yes. We'll accept that.

MR. MILOSEVIC: [Interpretation]

Q. You can check it on the tape. I'm reading your data. I'm not ²²⁸⁹ reading my own.

So in your text, you're talking about 236.000 in total out of which in the first stage 236.000, in the second stage 83.000, and in the third stage 67.000, and the total is 386.000.

JUDGE MAY: Wait a minute. Two hundred and thirty-six thousand in the second stage. Does that sound right?

THE ACCUSED: [Interpretation] [No interpretation]

JUDGE MAY: All right. 83.000 in the second stage. Does that sound about right?

THE ACCUSED: [Interpretation] That's true.

MR. MILOSEVIC: [Interpretation]

Q. In the first stage, 236.000. Is that correct? The second, 83.000; and in the third, 67.000. Is that correct?

JUDGE MAY: Just let the witness --

MR. MILOSEVIC: [Interpretation]

Q. And--

JUDGE MAY: Yes?

THE WITNESS: Those are approximately correct, yes.

JUDGE MAY: Yes.

MR. MILOSEVIC: [Interpretation]

Q. Now, look at the number of targets, bomb targets, NATO airstrikes. 924, as you put it, in the first stage, 541 in the second stage --

JUDGE MAY: Wait a moment. Let's break it down. It's suggested there were 924 in the first stage. Does that sound ²²⁹⁰ about right?

THE WITNESS: No. Where are you getting that data, please, so we can refer to it here? I don't remember that. I don't think -- the bombing numbers that I know of are much smaller than that. Remember, we're counting airstrikes, not people killed or something else. If you could give me a direct reference.

MR. MILOSEVIC: [Interpretation]

Q. You wrote in your text a total of 924 as the number of targets in Kosovo. 924. In the first stage --

JUDGE MAY: Where will we find this in the text?

THE ACCUSED: [Interpretation] In his own text, Mr. May. I can't leaf through it now. You can check this data. It can be found in the text.

JUDGE MAY: He doesn't agree with you. But will you come to the point?

MR. MILOSEVIC: [Interpretation]

Q. So what is the essence?

THE ACCUSED: [Interpretation] Please take the time to hear me out. And if you don't believe this data is in his own text, you can check it later. In fact, he himself can do it.

MR. MILOSEVIC: [Interpretation]

Q. We have verified -- deal with the number of refugees. Now we are dealing with the number of targets, 924 --

JUDGE MAY: [Previous translation continues] ... yes. What is the point that you're making to him? ²²⁹¹

MR. MILOSEVIC: [Interpretation]

Q. I have been saying this: Out of the refugees number, in the first stage -- 61.14 per cent crossed over in the first stage, because 236.000 out of --

JUDGE MAY: Does that sound right?

THE WITNESS: That sounds right.

JUDGE MAY: Yes.

MR. MILOSEVIC: [Interpretation]

Q. At the same time, the number of targets in the first stage is 58.55 per cent. 58.55 --

JUDGE MAY: What does that mean?

MR. MILOSEVIC: [Interpretation]

Q. -- and 61.14 --

JUDGE MAY: What does that mean?

THE ACCUSED: [Interpretation] That means that the percentage of refugee outflow in the first stage almost completely coincides with the number of targets in the first stage.

JUDGE MAY: Yes. We have the point. What is your comment on that?

THE WITNESS: I don't know where his number of targets comes from. That --

JUDGE KWON: How about the percentage? Almost 58 per cent was bombed during the first phase? That seems to be the question.

THE WITNESS: 58 per cent of what? Are we talking that 58 per cent of the airstrikes occurred in the first stage? That is certainly ^{2292 BLANK PAGE 2293} false. If we mean that 58 per cent of the municipalities in Kosovo were bombed during the first phase -- is that what the question is? That's possible. We'd have to look at that municipality by municipality. The question, though, it seems to me, for a causal argument, has to do not with whether it occurs in a long period but whether there is a sufficiently close correlation in time and place of bombing intensity and refugee outflow to hypothesise or to sustain the hypothesis that the bombing pattern has caused it, and we have conducted that analysis and rejected that hypothesis.

JUDGE MAY: Can you take it any further than that?

THE WITNESS: I don't think so, no.

[Trial Chamber confers]

JUDGE MAY: Yes. We will adjourn now. After the adjournment, we will hear the amicus.

THE ACCUSED: [Interpretation] You won't let me put any more questions?

JUDGE MAY: No. That's right, we won't.

THE ACCUSED: [Interpretation] What was that?

JUDGE MAY: We will adjourn until twenty to three.

--- Luncheon recess taken at 1.12 p.m. ²²⁹⁴

--- On resuming at 2.43 p.m.

JUDGE MAY: Yes, Mr. Kay. Yes, Mr. Milosevic.

THE ACCUSED: [Interpretation] It was my understanding that you would not let me continue my cross-examination. I just want to give some explanations that the witness requested, and I did not have these things at hand.

924 bombings. That particular figure comes from maps under B, from the first report. 924 bombings: In the first stage, 541; in the second one, 241; and in the third one, 142. And the percentages calculated showed that the percentages of refugee flows and the percentages of the targets engaged totally coincide. They correspond to each other, actually.

Secondly, the chart that the witness asked me about is the chart that is Chart 1, also from Annex 1 of the first report, and the total is 248.000: In one -- in first stage, 164; and in the third stage, 49 [as interpreted]. And Maps 2.1, 2.3, and 2.4, the total is 471.000. The second stage is 331 and the last one is 67 [as interpreted]. So according to the maps, it is one figure, and according to the charts, it is another figure.

In addition to that, the witness said that he did not operate with the figure of 850.000 refugees.

JUDGE MAY: You've had your chance to explain the matter, and that's sufficient. We're going to hear -- no, no. We're going to hear now some cross-examination from the amicus. ²²⁹⁵ Yes.

Questioned by Mr. Kay:

Q. Dr. Ball, have you ever appeared as an expert witness in court before?

A. No.

Q. The data that you used for the compilation of your report and research, is that available for an independent inspection?

A. The data for the migration study were placed on the Internet -- were placed on the Internet over a year ago. The data on killing were disclosed to the Defence along with the report. The information on KLA activity and NATO activity were derived, as I described, primarily from open sources, which we detailed. And so we believe that the data are available to the Defence.

Q. Is there any aspect of the data that is not available? Is there any aspect of your research that is unable to be disclosed or is unavailable?

A. The 45 per cent of the sources used to analyse the KLA data are the -- are in the domain of the Office of the Prosecutor, and I would direct the inquiry to them.

MR. NICE: Can I just help with that, because I was going to deal with it in any event. That material, subject to any redactions for proper reasons that might be approved by the Chamber, will be available to any expert who wants to have a look at it.

MR. KAY: I'm grateful.

Q. In relation to your research and study, what would be the leading ²²⁹⁶ international journal in which a study such as this would be published for reading and inspection by other members of your profession?

A. Probably demography, the journal of -- one of the journals of the Population Association of America.

Q. Has your report been sent to that publication for publishing?

A. Not right now, no.

Q. Do you intend to do that?

A. Some portions of the -- I'm sorry, there's a ringing.

Q. Yes. There's a technical hitch, obviously, going on.

MR. KAY: Shall I ...

JUDGE MAY: It can't be interpreted.

[Trial Chamber and registrar confer]

JUDGE MAY: Apparently the technician is on his way. Yes. It seems to be -- try it now.

MR. KAY: Thank you. I don't think that was me.

JUDGE MAY: Yes.

MR. KAY:

Q. Dr. Ball, just to repeat my last question ...

MR. KAY: I'll try again, shall I, Your Honour?

Q. Dr. Ball, just repeat my last -- no.

MR. KAY: We have something coming into the courtroom, Your Honour.

JUDGE MAY: Try again.

MR. KAY:

Q. Dr. Ball, I'll just repeat my last question. ²²⁹⁷

JUDGE MAY: We're going to have to adjourn. Ten minutes.

--- Break taken at 2.52 p.m.

--- On resuming at 3.02 p.m.

JUDGE MAY: Mr. Kay, it's apparently been fixed.

MR. KAY: Yes.

JUDGE MAY: It sounds all right.

MR. KAY: I will move on to another subject as I know time is of the essence.

JUDGE MAY: Yes. I should say that we will in fact sit until 5.00 because we've had these breaks.

MR. KAY: Yes.

Q. I want to deal with general matters concerning data, Dr. Ball. Data you have used depends upon the accuracy of the account or record given; that is right?

A. Well, the data depends -- every piece of data has some precision associated with it, if that's what you mean.

Q. Yes.

A. Yes.

Q. And the record made of data has to be accurate as well?

A. The record made of data. I'm not sure what you mean.

Q. Let's just take a simple example in this case of the border guards in Morina. The account given to them must be accurate.

A. Well, what people say to the border guards when they say, "This is my address," yes, that must be accurate.

Q. Yes. The date they left the country, that must be accurate? ²²⁹⁸

A. It must be accurate when it's written down by the border guards on the form, yes.

Q. And for any study such as yours, you are only able to act upon the records what you're given?

A. That's not quite right, actually.

Q. Why not?

A. Because this concern that you're raising is the point of a kind of study called sensitivity analysis. And we have done substantial sensitivity analyses in both the migration study as well in the present study, looking for, quite literally, the sensitivity of our findings to imprecision in the data.

Q. The record made by the person writing a document upon which your data is based must also be accurate; is that right?

A. No, it mustn't. Well, that's not quite right, actually. I'm sorry, I will have to refer to my previous answer. We -- it is certain that some data are in error. The question is then: Does that error impact our conclusions? That is the point of doing a sensitivity analysis.

If the findings are not subject to change, the conclusions are not subject to change as a result of making very, very different kinds of assumptions about the precision of the reporting, then we say the conclusions are robust to imprecision in the data. So it -- I think it's your term "must" that I'm objecting to. That's not quite correct, actually.

Q. In relation to the raw data, the forms completed by the Albanian ²²⁹⁹ border guards, are they available for inspection?

A. They are not currently available for inspection. I could -- I had, at the time that I obtained the records, a brief note of agreement with the Albanian government that I would maintain those records in confidence. I could inquire of the Albanian government.

Q. Yes. You've heard the allegation in this case that these could be fabricated.

A. I've heard that allegation.

Q. And a part of that allegation, if it's to be carried forward, would require examination of the documents that you yourself have relied upon.

A. That would be one way to test the allegation. However, I would refer you to Appendix A in the "Policy or Panic?" study where we tested the patterns in the border records against the surveys and found them to tell substantially the same story. We have several different surveys. We looked at some of the most important patterns we found in the border data and we found the same patterns. So I think that for statistical purposes we have already investigated the possibility that there might be some problem, of whatever kind, with the border data, and we have rejected that.

Q. Moving on now to an aspect of your analysis, which was the dates of the migrations. As I understand it, the data that you have worked upon has been compiled by an assumption of the period of time taken for each of those individuals to migrate from their village through to the border crossing of Morina. ²³⁰⁰

A. Sorry. What's the question?

Q. As I take it, you have made an assumption.

A. No, we didn't make an assumption. I'm sorry.

Q. Right.

A. Let me clarify that. We looked at the survey records that I've described in the past and we determined from the survey records how long each of the respondents in the surveys had been in transit from his or her home village to the border point. So we then have a set of transit times that correspond to a particular -- leaving a particular village at a particular time. Understand so far? I'll explain how this applies.

Q. Can I just clarify one matter: Is it an estimate, then, of the time of each individual, when they would have left the village to have got to the border crossing?

A. That's closer, yes, yes. We estimate, actually, for each person crossing the border, how long ago they left their homes, and we make that estimate based on the survey information.

Q. And it's that migration which -- or time of migration that really underpins your research, doesn't it?

A. Well, originally I thought it would, which is why I did a sensitivity analysis precisely on the question you're asking, and we simulated very, very different kinds of migration or transit time structure from people's departure to the point of getting to the border. That distribution, the amount of time it takes the people who cross the border on a given day - How long ago did they leave? Some people left today, some people left yesterday, some people left the day before - that, ²³⁰¹ to a statistician, has a certain shape. I simulated -- and in the report here that's been on the Internet for a year and a half or a bit more, I simulated very, very different distributions. I said: Well, what would happen if the transit time process was completely different? And I did that in six different ways. All of those results were nearly indistinguishable from the original. It turns out that the transit time process makes very little difference to the conclusions which have been presented in this report and subsequently in the report which my co-authors and I have presented to the Court.

Q. One other aspect of data as well that I'd like to deal with here is the dates of killings and identities of who has been killed, because, would you agree, this seems quite an important part of your research?

A. The dates of killings are quite important.

Q. Yes.

A. The names are important, but I think less so for technical reasons.

Q. Shall we just start with date of killing.

A. Sure.

Q. As I understand it, there are instances that you have relied upon when people have given a date, and that's been a date with a named person that you've been able to positively identify.

A. That's correct.

Q. There have been other occasions when accounts have been given of a body being referred to by the side of the road or in a house, which is a death that has formed part of your statistical material. ²³⁰²

A. Well, let's clarify that a bit. The people who were identified without name we referred to as the anonymous victims. The anonymous victims played a role in this analysis only for one purpose. It's important to be clear. For this purpose only did we use the anonymous victims, and that was to establish that removing all the anonymous victims who could possibly have been accounted for by name as a named victim, there nonetheless remain quite a few; at minimum 2.000-some anonymous victims.

The only use we put this information to is to say there have to be more. There have to be more than the named victims. We do not believe that the anonymous victims provide an adequate basis for a statistical estimate. They do not provide sufficient information for us to do analysis. So we used only the named victims. But we observe that net of the named victims, nonetheless, many thousands of anonymous victims remain who had names. These people had names, they are just not documented.

Q. In relation to deaths as well, was it possible to distinguish in your research as to which of the victims were Serbs, which were Kosovo Albanians, whatever ethnic group they may be? Was there any distinction you are able to draw within the figures that you have given?

A. Yes.

Q. Could you explain further.

A. We attempted to identify names that appeared to be ethnically Serb names. Because this was a study -- this is a study of ethnic Albanians, it was important to exclude the Serb victims from the calculations. They would get their own calculations in a different study, I presume. But for ²³⁰³ this study, we wanted to exclude them from the lists. We sought lists of surnames, and when we -- surnames of Albanian and Serb surnames, and when we found names, we checked each of the names on our lists against the Serb -- the list of Serb surnames, and when we identified them, we removed them from the list and did not use them in the calculation.

Q. So the identification process that you underwent had to be accurate to get the right people for your statistical study.

A. The identification of the ethnicity of their name?

Q. The people who identified them had to be right about the person they had identified.

A. That's correct.

Q. And any problems associated with the identification process might cause your attribution of those particular deaths to be flawed.

A. Well, "might" encompasses, to a statistician, a wide range of probably. I think it's possible that misidentification could increase the error range, the margin of errors that we have reported. However, I doubt it, because in order for the misattribution to affect these results beyond the error that we have already calculated, those misattributions would have to be not only very large in number but they would have to be very systematically distributed through these four independent projects. The combination of those two assumptions seems to me very unlikely. So I do not consider it probable. It is possible, perhaps remotely possible.

Q. We've heard about carbonised bodies, bodies that are unrecognisable. How were they dealt with in your study?

A. Well, if they were unrecognisable, then -- and they were ²³⁰⁴ identified, for example, by an exhumation, then they would be part of the deaths -- the bodies exhumed there were not identified. In reference -- in Appendix 1, you will find that we excluded approximately half of the exhumation records because they did not have names. So a carbonised body that was never identified would be in that category. On the other hand, if the person had been seen to have been killed by someone else, if there was a witness to the death and if that witness subsequently reported that death to one of -- either the OSCE or Human Rights Watch or ABA/CEELI or its partners, that death would appear in the lists.

If we can take a second on this observation, it's actually quite insightful, because this is precisely what a statistician is looking for by the combination of data with different kinds of data that are going to be excluded.

Q. Yes.

A. So if, for example, a death occurs and there are no witnesses to that death except the perpetrators, that person may nonetheless appear in an exhumation, but of course, he or she would not appear on a list created by witnesses because no one would have witnessed it. The example that has been posed by the question is the opposite problem; a death which cannot be identified in exhumation but might well have been identified by witnesses.

To a statistician, these are biases or exclusions or data that can't get into a list which complement each other when you combine the lists. This is exactly the thing which this method, the multiple systems ²³⁰⁵ estimation, is strongest at, is identifying those sorts of conditions.

Q. Moving on to another subject, I'd like us to turn to page 11 of your final report, and Figure 8, which is the table headed "Timing of KLA attacks with killings and refugee flow."

Have you got that before you, Dr. Ball?

A. Yes.

Q. Thank you. Looking at this table here in relation to your particular study, killings preceded or coincided with peak in relation to the KLA, from your study, am I right in thinking matched 38 per cent