 My name is Patrick Ball. I was one of the experts in the trial, and I spoke about the statistical finding consistent with the prosecution's hypothesis that acts of genocide were committed. There's a little, here it is. The fundamental conclusion that we did, and I'll walk through how we got to that, is a comparison of the mortality rates between the indigenous community and the non-indigenous community in the three municipal municipios of the Ischial region. In particular, through integrating four databases and estimating the deaths that we could not observe, we estimated that 2,147 deaths occurred as a result of killing by the army, excluding other state forces, killing by the army to indigenous people in that community. According to the census of 1981, there were approximately 39,000 people of indigenous ethnicity living in that region, which yields a crude mortality rate of 5.5%. That is 5.5% of the indigenous people alive in April of 1982 had been killed by July of 1983. Now, to a statistician, genocide, it is not enough for an argument about genocide to show that many people of the target community have been killed. We also need to show that the people who were not targeted were not killed. That is how we would understand, in a statistical sense, targeting that some people were not killed. And the not killed category in this hypothesis would be the non-indigenous people who were living in the three municipios of the Ischial region during the same period. And we estimate that 41 people from the non-indigenous people, a non-indigenous community, were killed. Of course, that's a much lower rate because it's also a lower population basis, so it's about 0.7% of those people were killed. From a statistician's point of view, the fundamental conclusion here is not necessarily either of these numbers independently, but rather their comparison, the ratio between the two, which in public health terms is understood as a relative risk. That is, the ratio of the proportion of indigenous people killed to the non-indigenous people gives us the difference in the probability of being killed between the two communities. That is, we can interpret this number, 7.9, as that if you were indigenous, your probability of being killed by the army was eight times greater than if you were non-indigenous in this period, in this region. Now, these, again, are only homicides committed by the army. And we are relating it to the census of 1981. Both of those are interesting points that I'll touch on later if there's time, which there almost certainly won't be. But perhaps you'll come to my talk tomorrow at LASA, where I'm going to problematize all of these numbers in great depth and make it much more complicated. For those of you who've been saying that 200,000 people were killed in the armed internal conflict in 1960-1996, that number is probably not right. That's, it's not right. I calculated it, so I can tell you, for the truth mission, it's not right. We'll talk about it tomorrow. We'll come to the topic. So we'll skip forward to the next slide and say that, again, this data came from, first, the census of 1981. This is the page from that census for Nebach. And then from four data sources. First, the International Center for Human Rights Research, the CIIDH, then the RME Project, the Catholic Church, also the Commission for Historical Clarification, and the National Program for Compensation. These are all named individuals. These are not lists of groups. And this is a big point that I'll talk a lot about tomorrow. These are people who are identified by at least one forename and one surname, and for whom we have at least a year of death, if not month and day. We have relatively few days of death. And we have the municipio of death. So these are well-identified records. Anything else was excluded from this analysis, and we calculated those exclusions by means of a bit of math. So let's talk a little bit about the notion of registration and under-registration. A lot of times when we talk about statistics, we're talking about what we have in our hand. We have a list. And we count the elements in the list, and we say, that's how many people died, because that's how many we have in the list. And a lot of times, we're worried about how many people we can verify. So we filter the list. We take a bunch of people out of the list who may have been killed, and we have their names, some reason we can't verify them. That's not how statisticians do it. Statisticians are interested in what the total population of deaths is. And our job, then, is to start with what we can observe, represented here by these white intersecting circles, in order to estimate the entire population, which is what we can observe, plus what we cannot observe. And we use a bit of probability theory to estimate from what we observe to what we cannot observe. This is not simply so we can get a higher number. Not at all. In fact, the total number is of relative unimportance to most statisticians. Rather, the point is that if we are going to make an argument about patterns in which we are comparing one category to another, whether that's March to June over an analysis over time, whether it's Chahul to Kotsal, if we're doing an analysis over space, or, in this case, the indigenous to the non-indigenous community, we must have unbiased estimates of both in order to compare them fairly. So it is completely inadequate to work with only what you know, because one community may have told you their stories at a more enthusiastic rate than the other community, which would create false statistics. So we fix the statistics using some probability theory. And the way we do it is this. Now, I actually did this proof for Judge Barrios in court. And I was delighted when I got to the, and that is how we estimate the unobserved deaths. She smiled, looked at me, and nodded. The universal indication that a light bulb has gone off. She got this proof. And I was delighted. I looked at the defense table. There was somewhat less in the way of light bulbs. So we worked from, as I said, I'm not going to do the proof here. There's not enough time. Please come tomorrow, where I will be less respectful of my time limits. And I will make sure to do the proof. The key here is that we're using the intersection pattern. The rates, or excuse me, the levels at which the data sets document the same events. We're looking at, for example, across all these events, across all these data sources, if we consider the number of people who are documented by the CIIDH, RME, the CEH, and the PNR, there are two of them. Two people, among the thousands of people documented, only two people show up in all four data systems. While at the same time, there are about 1,000 who are in the PNR and not in any of other data sources. There have been people who've raised questions about the quality of the PNR. I think actually, from a statistical point of view, the PNR is at least as good as the other data systems and probably a lot better than one of them, which I will not name here. But it's the one that is often considered the most popular. It's not the commission. Have I hinted enough? All right, I'll move on. So because we've done an estimate, an estimate is bounded by an error interval, a confidence interval. Sometimes late people call these a margin of error. It's not quite a margin of error, because it's not plus and minus the same amount. The error is asymmetric because of the way we're doing the calculation. But this estimate of 2,147 people killed by the army, indigenous people in these regions during this period, is in a confidence interval between 1996 and 2325. In our report that we gave to the court, we provided the technical background for how the calculation is done for the confidence interval. The point of presenting these confidence intervals, the key point here is that these intervals come nowhere near each other when we compare them. So this is the bar for the number of indigenous people killed during the period of Rios Mont's government. The dark blue area, the number of people we observe to have been killed, the light blue area, the number of people we estimate to have been killed. The width of the bar indicates the population basis. It's proportional to the population basis, about 39,000 people. And this error whisker, this little line here, indicates the range of probable estimates. This green bar next to it has all the same characteristics. Its width is proportional to the number of non-indigenous people who are alive in that region in that period. The dark area is the number we observe. The light area is the area we estimate. You notice that they're very, very different. And that's really the whole point to this argument. The whole point is that they are so different that it is completely implausible that that would occur by chance. And so to a statistician, we would conclude that, therefore, the evidence is consistent with the hypothesis that the Guatemalan army committed acts of genocide against indigenous people. Now, this graph is also a little bit of a curveball because we're slipping in other periods. These are a series of 16-month periods that proceed and follow the government of Rios Mont. And what I think is interesting about this graph is that it's clear that the rate of killing against indigenous people is substantially higher, substantially higher, during the government of Rios Mont than in previous and succeeding periods. Now, it may seem that the rate of killing against non-indigenous people declined slightly, but it's within the error. We cannot reject the hypothesis that these are all the same rates here, these little green bars, because they all have overlapping confidence intervals. But we can say that the rate of killing against indigenous people during the Rios Mont government is substantially greater than in previous or following periods. The prosecution asked me also to make some comparisons between killing in the Isil region and recent genocides, particularly Rwanda and Bosnia. Now, comparable statistics, reliable statistics do not actually exist. We know quite a bit about Bosnia, actually, thanks to the Research and Documentation Center in Sarajevo. Terrific job and nearly complete enumeration of deaths in the Bosnian conflict. But for Rwanda, the statistics are appalling. We really have very, very little clarity. The only place that we have anything like a clear finding is in the Kabuye prefecture. And if we consider all victims, we cannot distinguish the victims by ethnicity. We find that approximately 20% of all the people alive at the beginning of the genocide in Kabuye had been killed by the end of it. And the reason that I present that statistic is that it's actually pretty close to the total killing rate across the period, 1979 to 1986 in the Isil region, which comes out at about 22%. So that's quite interesting. That in some sense, there's a sort of a commonality of this notion of 20% of total deaths in these other genocide cases. Srebrenica also, about 20% of the people alive at the beginning of July, 1995 had been killed by the end of 1995, nearly all of whom were killed in the 3A period as the enclave was emptied. So I found that evocative and we presented that. So let's come back to this notion of bias and why it's so hard. The problem is that while we're living in these white circles, while we're living in the world of documented deaths, it's impossible for us to tell the difference between the reality on the left where in fact, we only document about a third of the deaths and the reality on the right where we've documented almost all of them. It's impossible to tell while we're only looking at the documented deaths, whether we've got only a little fraction of it or we've got nearly the entirety. And that's important because it might be that for one community, say the indigenous community, we have very good documentation. It could look like this, while for the non-indigenous community we have poorer documentation. And that would create a bias which would be unfortunately a bias in the direction of the prosecution's argument. And whenever you're doing a scientific argument, you want to create biases that go against the hypothesis you're testing. So that's the key to motivating our estimation. But it also leads us to look at some of the data sources, in particular the PNR and the census, and ask ourselves the question, if there were some systematic problem in this data source, would we nonetheless, and we corrected it, would we nonetheless reach the same conclusion? So what we did is we took the records that are only documented by the PNR. We took records that did not appear in Remy, in the commission, or in the CIIDH, but did appear in the PNR. And we systematically, well randomly, deleted sections of those records. So we deleted up to half of them. And then we recalculated the ratio, the relative risk, of being killed by an indigenous person relative to a non-indigenous person in this period in this time. And we found that by deleting half the records in the PNR, we still have a relative risk of almost five. Our argument is sustained if we assume that half of the people who spoke only to the PNR made their stories up out of thin air. And I think that that is a deeply implausible claim since the PNR's records were, PNR's findings were substantiated by death certificates and investigations and other things. So while there may be some PNR records which are falsified, this analysis shows that the conclusion that we draw is not affected by the possibility of in fact a quite substantial number of falsified records. Similarly, it is known to demographers that censuses systematically under register people who are not urban and people who are of excluded ethnicities. And this Guatemala's a classic case. So mathematical demographers and applied demographers did a lot of research on Guatemala in the 70s and 80s looking at the early censuses, the censuses in 1951 and 1964. And they found in those censuses that indigenous people were under registered by about 15%. Now from my argument, this is quite problematic because I'm using the census of 1981 as the denominator. And if the denominator is too small, I will be, if the indigenous denominator is too small, I will be inflating artificially the estimates that I'm making of their homicide rate. Well, if we assume that in 1981 there was an under registration worse than those of the earlier censuses, which again is implausible, censuses get better over time, not worse. But if there was a very severe under registration, 20% of indigenous people, while non-indigenous people were fully registered, we would nonetheless still have a relative risk of six and a half between the two ethnicities. So again, our conclusion would remain very strong. So that in 15 minutes is what took me an hour and 25 on the witness stand. Although it would be helpful if someone in here could ask me how division works. Thank you very much. My expertise was in the trial and that incredible and frustrating trial was about racism and genocide. I say everyone's know that it's my, well, my expertise. The first thing that I tried to get with the attorneys was how the racism was a structural and historical element through history. How it starts with the racism in the colony and then it starts in the spheres of the 19th century, calling people degenerate, bad people, inferior races. And how it comes to debate, a very great debate around the 30s about the degenerative race and the improvement of the race, the eugenesia, and the extermination of indigenous people. Since the end of the 19th century and beginning in the 20th century, the imaginary of racism and practical racism in the elite was very common. The extermination of the population was a very common thing in the newspapers. So that ideology of racism goes through the elite of the power you know my book about Lina and Racismo and that established and dispersed to everybody in the population. With the war, I think it's the worst part. The war was everybody start seeing the Mayan people as terrorists, as subversives, as guerrillas, communists. And we have heard many testimonies where they were subversive insurgentes, guerrillas, supported groups. Then in that part, in the 80s, all Indians were subversives where it was necessary to cut off the seed of evil, to cut the seed from the roots and to normalize. So they have to be Ladini Sarlos or to erase the Mayan identity. It was very clear in the Sophia plan when they treat them like animals or like objects or they are named like Filo Eno and when they kill the children, they say that word chocolates. I didn't know where it was chocolates, why? It was referring to the color of their skin and the woman's too. So I think that the Sophia plan refers to extermination and total annihilation of the population. And it also mentioned the possibility of destroying the population and the community ties. Many people of their testimonies or in the trial report that a high number of women and children were murdered and were insults and humiliated and wrapped. Okay. So I think that we could affirm that in the case of Guatemala, it was not only reasons what it happens or only genocides. It was a two ethno-seed, against the cultural, the cultural Indian Mayan culture. But the problem was the process of dehumanization and devaluation of the other. They treat them as animals, carries a heavy burden of racist and a stigmatization of the other, understood as inferior or expendable and situation of ex, is ex-executive rebel of Mayan and women's. But the thing is how to prove the genocide. But the problem is the intention or the intent to prove of the proposal of to prove a genocide. The premise of the intention to eliminate an ethnic group like that cannot be denied by simple assuming that they were subversives or because the roles they played in the conflict. The intent to destroy the group can be inferred from certain assumptions related and interconnected to the crime of genocide. Which are the assumptions that all the olocaust and all other genocides, Alruanda, Bosnia, Dafur has the massacres. How do you, can you difference between genocide, crime and as delesa humanidad or just collective massacres, no? The presumption of Causen mass murder, so genocide, as Verdeja, Ferenstein and all the specialists or the expertise of genocide said is to cause mass murder and genocidal massacres to children, women, elders, civilians to destroy housing and culture and religious images to sites. The existence of clandestine graves and mass graves, the dehumanization and dispersal, the depersonalization of the victims, the declaration of target group as public enemy, intent to erase the elements of ethnic identity. The total of partial destruction of an ethnic group are through an systematic planning of human extermination or the public statement of people involved in extermination. The model, the type, it's the Holocaust, the Jewish Holocaust in Germany. And other Holocaust or other genocide like Bosnia or Dafur or Rwanda has four, five or six intents of elimination. We have the nine, the same as the Holocaust. We are the only one that has the same name of presumption intention of exterminate an ethnic group like that. So my remarks or my conclusions is that historical structural racism, stereotyping, stigmatization of indigenous people through the history of Guatemala and the institutionalization of the violence of the violence and violations contribute, help and facilitate the perpetration of genocide. And it reminds one of the more powerful ideological tools to justify the genocidal massacres in the country. Since racism was already internalized in the minds and in the hearts of the elites and the perpetrators of genocidal violence. Now we know that in the urban middle class it's already there in the minds and in the hearts of the urban Latino middle class of the city of Guatemala. The structural historical racism ideology contributed to shape the characteristic of the state, of the racist state, discriminatory, in equal authoritarian and use the ideological and repressive institutions against the indigenous population in times of crisis and domination. I think that the consolidation of racism as an ideology of the state reached its apex and showed its peak intensity with the crisis of the oligarchic military government, the emergence of the popular revolution of movement and the application of the counterinsurgencies. The problem is that the counterinsurgencies, the manuals of the counterinsurgencies were applies without limits due to the presence of the racist ideology in society. The historical structural context of the racism and the intensification of the stereotyping and stigmatization of the indigenous people. In the case of the Mayanic Shield indigenous people were they identifiable as public enemies of the states and members of the guerrillas. So that was the pattern and modus operandi through all the history in Guatemala. It's not the only one. It has been in Quetzaltenango, in Pazizilla in other moments of the history in Guatemala. So I think that the genocides, it's increased it, it's internalized in all the country in all the classes and meanwhile more in the Latin-Urban middle class. In that case, I think that it could appear again in any moment in the future because with this trial we have seen in the papers, in the newspapers, in the mass media that the blogs or in the commentaries of the urban Latinos middle class were so racist, were so cruel that they didn't have any empathy, any compassion, any kind of poor people. What has happened? Oh, what a terrible thing. They didn't have any compassion. They said, no, it's a pity. They didn't die anybody, everybody in the war. So that is really the most worried thing that we should reflect for tomorrow. Thank you.