11 January 2008Michel Thieren
A third assessment of post-invasion violent deaths in Iraq was published on 9 January 2008 by the New England Journal of Medicine, a prestigious platform for medical research and scientific debates edited in Boston, Massachusetts. The lead article in the journal - "Violence-Related Mortality in Iraq from 2002 to 2006" - reports the results of an inquiry by the Iraq Family Health Survey Study Group (IFHS), involving collaboration between national and regional ministers in Iraq and the World Health Organisation (WHO). It finds that 151,000 (between 104,000 and 220,000) people died from violence in Iraq between March 2003 and June 2006. When such a politically sensitive figure is published, it is critical to turn statistics into words and explain what the new evidence tells, what it does not, and how far it confirms or invalidates the previous ones.
In October 2006, an article in openDemocracy discussed two important statistical analyses independently published by scientifically reliable sources - respectively the Iraq Body Count project and the medical journal the Lancet - in relation to excess violent mortality attributable due to the invasion in Iraq (see "Deaths in Iraq: how many, and why it matters", 18 October 2006). Both figures pointed to very different magnitudes, respectively 48,000 and 600,000: the former being generated indirectly through scientifically screened media reports and the latter through population-based surveys. The studies were grounded in and consequently equally justified by very different methodological approaches, but the dichotomy in their findings became a source of political appropriation by those (broadly in the pro- and anti-war camp) seeking to support their respective, preordained positions.
In scientific terms, 48,000 was likely to be too low because indirect reporting - even through the best possible screening and verification process - will always suffer from under-reporting and missing events; by contrast, 600,000 was considered as possibly too high because it was based on a very reduced sample that may have been too small to be representative of the entire population of Iraq. The most recent figure ranges in the middle, in that it is three times higher than the IBC's media-based figure and four times smaller than the survey-based number. The third assessment The Iraq Family Health Survey's broad and collaborative data-collection exercise has been undertaken to allow the Iraq government and the World Health Organisation to update the Iraqi national health database and provide enough health information to design appropriate national health policies and programmes.
The survey thus covered a wide span of health aspects, including measures of mortality due to violence. In order to generate valid and reliable estimates, the survey designers ensured that the sampling would be large enough to curtail any doubt about its ability to represent all Iraqis. From that perspective, having investigated 10,000 households - more than five times more households than in the Lancet survey (which was based on research by a team from the Johns Hopkins Bloomberg School of Public Health in Baltimore - the January 2008 estimate is very strong. In addition, the conduct of the survey carefully incorporated a series of checks and balances to enhance quality: such as intense training of survey teams, verification processes, phase-based implementation, multi-agency participation and cross-checks, and coordination mechanisms.
The final analysis and computation compensated for a series of possible biases - such as the under-reporting of deaths because people have moved away from households and relocated across them, the impossibility of visiting some households for security reason, and the effects of migration of Iraqis to neighbouring countries. Although adequately controlled, these biases are still present, and this makes the final estimate of "151,000" the one that is, for that survey, the closest to the true toll. The survey released by the New England Journal of Medicine, therefore, concludes that between 104,000 and 220,000 people died in Iraq during the three years after the coalition forces invaded Iraq in March 2003, with the highest probability that the true number is 151,000.
The survey, unlike the Bloomberg School / Lancet one, does not say that this estimate represents the excess violent deaths attributable either directly or indirectly to the invasion. It is a toll of violent death in a critical period of the recent history of Iraq. "The deaths don't include car accidents and they don't include unintentional injuries", says Dr Ties Boerma, director for health statistics at the WHO. "They just include intentional injuries and armed conflict. In fact, the armed-conflict deaths are more than 80% of the deaths we got reported." Another aspect of the storyline comes from trend analysis. When disaggregated year by year, the death-toll does not show the escalation of violence the survey published in the Lancet found (200+, 500+, 900+ deaths per day). In the IFHS/WHO survey, the daily deathrate was 128 the first year, then 115 and 126 in the two years after a trend corroborated by the Iraq Body Count project. The sharp difference comes in the measurement associated with the second but more with the third year and can be grounded in a sampling problem which was an issue in the Lancet study. Who holds the truth?There remain three numbers: one too small, one too high, one somewhere in the middle. More data-collection exercises could complete this macabre series of anonymous statistics, each one raising technical issues and feeding scientific debates. The authors of the study published in the Lancet are now arguing that many deaths may have been kept unreported, by implication questioning the independence of a survey conducted with Iraqi national authorities; this would suggest that the international community itself, the World Health Organisation and other participating funding agencies would have trumped scientific excellence to abide by a political agenda.
In suchways,the numbers game can be turned away from a evidence-based search for truth, adding its cortège of replicas to a scientific chorus that falls in amnesia before the unchanged humanitarian reality on the ground: healthy people die in Iraq because politics has tragically failed there.
The next agenda As in 2006, the final question should not be formulated in statistical jargon, even if the study published on 9 January seems to have the last technical word on that matter for the reasons invoked above. In a period when the United States presidential campaign is gathering pace, the preoccupation of the candidates should be with the jus in bello - or how the war in Iraq has been and is being conducted. Civilians die in Iraq because there, politics has simply ended, Thomas Hobbes's Leviathan crashed, and Clausewitz's praise for organised conflicts been discredited.
The numbers game in Iraq may not be over, but the implacable humanitarian argument continues to speak for itself, independently and away from scientific arenas: many lives are lost in Iraq for no reason. By 2009, the elected fifty-fifth US president may refer to that humanitarian argument - perhaps with the help of some mortality numbers - and conclude that invading Iraq was not such a great idea, and that nation-building can be accomplished and democracy exported without military intervention. One of the deaths in Iraq was that of Sergio Vieira de Mello, a man with inalienable humanitarian convictions and immense peacemaking capacities (see "A world of dignity", 24 August 2003). He died - with twenty-one others, many of them United Nations colleagues - on 19 August 2003 in Baghdad. In his life, he made a difference both to entire populations trapped in war-zones and to anonymous individuals who by chance crossed his hectic path here and there. His immense legacy represents a Socratic standard that can provide fine-tuned answers to a key question: not how to end the numbers game in Iraq and elsewhere, but how to use what we know about numbers to begin politics and stop the killing.
http://www.opendemocracy.net/article/conflicts/iraq_handover/numbers_game_revisited
