Tuesday, January 29, 2013

An Interview With Iraq Body Count


The U.K.-based Iraq Body Count (IBC) is the premier organization for tracking deaths and violence in Iraq. Since its founding in 2002, it has kept a running count of the casualties caused by the Iraq War, which is constantly updated. Its work has also been included in various studies, and is a constant reminder of the costs of the conflict there. This is an interview with two members of IBC, Hamit Dardagan and Josh Dougherty.

1. What was the impetus for creating Iraq Body Count (IBC), and what year did it start?

The principal impetus for starting IBC in late 2002 was to ensure that the deaths of civilians in the forthcoming war would stay firmly in public view, remain a prominent part of debate and discussion about the war, and be considered soberly and appropriately. More specifically, it was to ensure that the cumulative number of reported civilian deaths would become an accessible part of the public record.

A notable, but not untypical, example of discussion detached from any pertinent data was Tony Blair’s eve-of-war statement to the U.K. Parliament that “[Saddam Hussein] will be responsible for many, many more deaths even in one year than we will be in any conflict.” This might or might not have been true according to the best evidence available at the time, but to make a quantitative claim without so much as mentioning any relevant data is surely unacceptable, and all the more so when so much is at stake for those affected. Indeed, our data proved Blair’s unsupported claim to have been simply false. More civilians were killed in just the first two months of the war in 2003 than were killed by Saddam Hussein’s government in the year before, or indeed for many years before. One would have to go back at least to the early 1990s to find any plausible evidence for deaths on a similar or greater scale. Blair’s claim went mostly unquestioned, but, given that it dealt with an essentially numerical question was of the sort that would not have been permissible for a politician to make without closer examination if it had been about, say, taxation.

One of the ways in which IBC has developed since that time is our recognition that no one else, currently, is fulfilling this role with the same rigor, at least not publicly. Thus from a role, which has been primarily directed at the West, we are now moving to one that recognizes our potential role as the source of record for Iraqis too, and will shortly introduce an Arabic translated version of the site that takes us more visibly in this direction.

2. How does IBC collect its information on deaths in Iraq?

We primarily collect our data from online news media or other publicly available reporting sources, supplemented by official records where and when these become available. This needs to be done 7 days a week. From these reports we extract not just the numbers killed and injured, but a range of detailed variables such as the location, time and date, weapons used, perpetrators, victim demographics such as age and sex, profession and marital status, and name; all wherever possible, of course, as these things are not always reported.

Surprisingly, perhaps, we have some demographic information on at least 42% of the dead, and the weapons involved for 93% of incidents. One of the most important pieces of information that is still woefully limited is the victim name; currently this is recorded for less than 8% of the dead.

One might think that once a death or a particular violent incident is reported by the press, it’s in the record, but because there are so many such reports, spread across so many different sources, particularly when the violence is at elevated levels, with the latest news pushing yesterday’s aside, or after a week or so, totally overwhelming it, it is all too easy to lose track or confuse events with others.

Very soon the scale and number of these incidents is lost from view, and for ordinary news consumers at least, all that’s really noticeable are the most unusual or large-scale events. And individual news agencies never manage to comprehensively cover all reported deaths. Generally it’s just those within their own network of correspondents and contacts.

So there are two kinds of biases that we have to work against: the first is the editorial impetus to push ‘regular’, unremarkable news such as the violent death of, say, a single, ordinary Iraqi into obscure corners of the press. It’s reported, but demoted to the “inner pages” in print terms, or hidden many paragraphs into an article. This is the main reason why an ordinary news consumer will be astonished to learn that, provided you go looking for it, and have access to subscription news databases such as Lexis-Nexis, much more is reported about Iraqi deaths than one would imagine.

The second bias is that the press more consistently reports larger incidents than smaller ones. Thus there is essentially blanket coverage by the local and Iraq based international media of larger-scale events, but when incidents are smaller, and involve fewer deaths, the coverage becomes more patchy. One therefore needs to monitor as full as possible a range of sources and reports, and carefully combine them to provide a more comprehensive picture, paying due attention to potential errors such as double counting.

It is our view, based on years of painstaking engagement with this work and close review of its limitations, that, at least where Iraq is concerned, it is in the main only incidents where a very small number of persons were killed that are likely to have been completely missed. The importance of this depends on how many of the deaths occur in this way, how quickly this coverage tails off at the smaller incident sizes, and whether there are cumulative sources available such as monthly reports from morgues or hospitals that are unaffected by incident size.

What is also certain is that finding all these reports, and stitching them all together, is not easy. As we’ve discovered, it takes a dedicated research effort to pull all this data together, and something much more rigorous than an ordinary monitoring of the news to extract and organize each item of useful information contained in that reporting.

3. What separates IBC from other groups that publish death counts such as the Iraqi government or icasualties used to do is that the numbers are constantly updated. Can you explain some of the process behind that?

I would say that the sharpest distinction between IBC and the data on casualties released by the Iraqi government, typically published simply as monthly totals, is that we continually publish disaggregated data specifying the time, place, and a systematic description of each deadly incident, with the number of people killed or whose bodies were found in each case. Of course, IBC’s data can be totaled at the end of every month to produce monthly numbers, but the user of our figures can look much deeper than this, and see what they are based on.

Given that IBC’s monthly figures have usually been higher than those released by the Iraqi government, for theirs to be correct and ours to be wrong, specific incidents in our database would have to have never occurred, or many incidents would have to have had lower death tolls than reported, which is all quite unlikely. Basically our figures are transparent enough to be verifiable, which is arguably one of the reasons they are taken seriously. Aggregate figures are much more impenetrable. We don’t know what specific deaths or incidents are included in them. For all we know, there could even be some deaths in the lower government totals that are not present in IBC, but without incident by incident listing, there is just no way to know.

By contrast to government figures, iCasualties, one of whose two co-founders took inspiration from IBC does publish disaggregated information like we do. Unlike us, their focus has primarily been on tracking and listing deaths among US and Coalition military forces. While they do also provide a listing of Iraqi casualties, it’s been my understanding that this has been more a secondary part of their project, which does not attempt to cover a wide range of reporting sources or claim to be a comprehensive record of reported deaths. As such, every time I have checked their site, their Iraqi civilians totals were lower than IBC’s for any given month. Perhaps a more important distinction is that iCasualties does not extract or present the range of other variables, such as listed in answer to Question 2 above, which thorough information-extraction process explains the 2-week or so delay before our formal data is published.

Much of the value of IBC’s work has not been in compiling an accumulated number, but in tracking all the associated information about victims and incidents, such as their age and sex, the location of their killing, and the weapons that killed them. These have formed the basis of several scientific papers published in major journals, co-authored by IBC and other researchers, which provide new insight into the human impact of the war in Iraq and perhaps other modern conflicts as well.

The constant updating, which I’m pleased you noticed, is important to IBC both as a means of underlining the fact that the violence itself is ongoing, which our data merely reflects, but also as part of the transparency we try to bring to the process: it is easier to trace recent facts than very old ones.

4. How did the Wikileaks release of Iraq war related documents affect IBC’s casualty figures?

There is no question that the WikiLeaks data has contributed greatly to public knowledge and understanding of the war’s impact on Iraqis. First, we estimated after an initial analysis of the documents that they would add around 15,000 previously unreported Iraqi civilian deaths to our total. We have formally added over 3,300 of these deaths to our database so far (listed here: http://www.iraqbodycount.org/database/incidents/list/wl2012), and this work continues. In addition to what they contribute in terms of numbers, the records also contain a wide range of specific details about particular incidents and victims, including thousands of victim’s names, which adds considerably to our understanding of the pervasive, day-to-day tragedies that have occurred throughout this war.

The WikiLeaks data also confirmed that as far as our other sources are concerned, it is primarily some of the smaller incidents that they tend to miss. We are finding that it is only incidents of this kind that we are able to add to our database from the logs.

5. IBC not only keeps aggregate numbers for those killed in Iraq, but tries to keep track of people’s names that lost their lives as well. Why is that important?

The importance of names can scarcely be overstated. One need only look at almost any  war memorial to understand this. It must always be remembered that we are not dealing simply with interchangeable units or numbers on a balance sheet, but rather individual human beings with unique lives and identities. Names help keep this perspective at the forefront. There are also technical reasons, such as the ability to differentiate between victims with greater reliability, for the bereaved to know the fate of loved ones who are presumed missing, and for those who are still missing to be correctly identified as such, that is, that they remain truly unaccounted for. But one of the main goals for any history of a war must be to treat its human losses in the appropriate manner; to record not just how many died, as important as this is, but who died.

6. There were several surveys early on in the Iraq war such as by ORB and the two published in the Lancet journal that estimated hundreds of thousands up to one million deaths in Iraq. What was Iraq Body Count’s opinion of those reports?

We think the surveys published about Iraq have been rather a mixed bag. The most credible have been the Iraq Living Conditions Survey (2005) by the United Nations Development Program (UNDP), and the Iraq Family Health Survey (2008) by the World Health Organization (WHO). Both have produced estimates that are higher than the corresponding IBC numbers over the periods covered, but within at least a plausible range. The surveys published by the Lancet (2006) and the polling group ORB (2007) are, however, clearly very large overestimates in our view.

We discussed some of our initial skepticism of the 2006 Lancet estimate shortly after it was published here: http://www.iraqbodycount.org/analysis/beyond/reality-checks/. Since that time, there have also been multiple academic papers and other analyses that have been critical of that survey and called its findings into question for a variety of reasons. Taken together, we think it’s clear that that survey was simply wrong, and by a large margin. As for the ORB poll (2007), this was never taken as seriously as something like the Lancet survey, as its estimate was derived from a fairly crude opinion poll better suited to other purposes, and was not published in a serious journal. But it did initially make some public impact, perhaps in part because it seemed to somewhat ‘corroborate’ the very high estimate of the Lancet survey, and quite likely also because “a million” is an attention-grabbing and memorable number. However, a closer look indicates that the ORB poll was also a very big overestimate, and of insufficient quality to provide reliable corroboration whatever its findings. The problems with the ORB poll were discussed in detail in a paper co-authored by a member of the IBC team here: https://ojs.ub.uni-konstanz.de/srm/article/view/2373.

7. Violence in Iraq has gone through trends from the initial invasion, to terrorism immediately afterward, to the civil war period, to the current one. What is the leading cause of death in Iraq these days, and is that the same or different from the civil war years?

I’d like to suggest that you and your readers answer this question for yourselves, using a tool we’ve produced – and haven’t really a good name for yet – that allows you to specify weapons, time-frames, provinces, and a number of other variables, and produce a graph (or download the comma-separated data) to this or other questions on matters for which we’ve collected formal data that allows descriptive statistics.

We’re still working on this tool to make it more detailed, but the point is to allow readers to perform their own investigations, realising that people have other interests than those which strike us as most important. Of course when there are trends which we find noteworthy we will continue to point these out, but as we’re dealing with a very large data set there is a lot of scope for people to explore it along the lines that most concern them.

The latest version of the tool is at http://www.iraqbodycount.org/analysis/numbers/2012/

8. IBC has deaths going up for the last three years, yet said in its year-end analysis for 2012 that security was largely unchanged. Can you explain that?

We see the security situation as largely unchanged because the changes have been relatively small over the last three years. Actually, over the last four years each year has fallen within a pretty narrow band of between 4-5,000 deaths, much narrower than changes across previous years of the war. It’s highly unlikely that any two years will have the exact same number of deaths, so it’s mostly a matter of what degree of change is sizable enough to consider a significant change in the situation on the ground rather than just statistical ‘noise’. For example, the change in our totals for 2010-2011 is from 4,073 to 4,137. While that is indeed “going up” in an absolute sense, the difference is far too small to consider a national trend. It must also be remembered that simple variations in reporting patterns, rather than changes in the violence, or that having just a few more or a few less big bombings in one year than the next, could account for small changes in totals. The change between 2011-2012 is somewhat larger than 2010-2011, but still much smaller than previous notable changes in yearly levels across the war. Thus we note the increase in 2012, but viewed broadly, we don’t see this as clearly indicating any significant change in the general security situation across the country.

One point we need to always make in these discussions of trends, lest they become too abstract, is that of course, once one removes the “abstraction layer” of applying an annual or other time-frame, then 4,000 vs 4,073 deaths is not an “improvement” on the latter number: it simply represents 4,000 more people who have been killed and can hardly be considered an improvement on the previous state of affairs for those new victims and the bereaved. Deaths unfortunately can only ever be cumulative.

9. Finally, the Iraqi government’s figures and IBC’s have been going in opposite directions for a little over a year now with the official numbers decreasing. Can you explain the difference in statistics?

We can’t, for the reasons touched on above. Without seeing the disaggregated incident-level data, there’s no way to tell what deaths we have that they do not, or vice versa. If we view the different totals as implying that their figures are challenging ours – that our figures are somehow too high – there’s no way to tell which of the deaths in our totals are being ‘disputed’ in the first place, and so no way to resolve the dispute. Moreover, there has been no clear methodology described for the data that’s been released by the government in recent years. It is reported as being compiled from data from the ministries of Health, Interior and Defense, but that’s basically the full extent of what is known. The underlying practice of how this is done, and therefore what might account for something like deaths going unrecorded, remains impossible to determine. What we can say is that we stand behind our numbers, which by contrast can be examined in  detail, and think that whatever the precise explanation for these discrepancies, that explanation will primarily involve some unspecified limitation of the methods being used to produce the Iraqi government’s monthly figures.

SOURCES

Iraq Body Count, “Iraqi deaths from violence in 2012,” 1/1/13

No comments: