MODERN TIMES

Art Hobson

ahobson@uark.edu

NWA Times 28 Oct 2006

 

How Many Iraqi Deaths? 

 

              A recent survey implies that the Iraq War has killed twice as many Iraqis in four years as Saddam Hussein did in 20 years.  Is this plausible?  President Bush didnÕt think so.  When informed of this survey, he instantly pronounced it Ònot credible.Ó 

              Most estimates of war deaths rely on ÒpassiveÓ data such as military reports or hospital reports.  ÒIraq Body Count,Ó which estimates 47,000 civilians killed, is the best-known of these. The last guess from President Bush was 30,000 civilian dead. 

              The new survey utilizes ÒactiveÓ interviews of a statistical sample of households throughout Iraq, much as the Gallup Poll samples the U.S. population.  It was conducted by four members of the Johns Hopkins School of Public Health and published in the Oct 21-Oct 27 issue of the British medical journal ÒLancetÓ (available in the university library). 

              The survey used a method called Òcluster samplingÓ thatÕs standard in epidemiology:  50 geographical locations, called Òcluster centers,Ó were randomly chosen from all over Iraq.  Interviewers surveyed the 40 households nearest to each cluster center.  So about 2000 households, having nearly 13,000 household members, were interviewed in all.  Interviews were done by trained 5-person teams of medical doctors.  As you can imagine, the process was dangerous for the interviewers. 

              Many precautions were taken to ensure truthful, accurate responses:  no incentives were provided, the survey purpose was carefully explained, oral consent was obtained, deaths were reported only if the deceased had lived in the household continuously for 3 months before the event.  Most importantly, at the conclusion of any interview in which a death was reported, surveyors asked to see the death certificate.  92 percent of these families produced a death certificate; the remaining 8 percent were judged to be truthful in their claims of losing the certificate (remember, thereÕs a chaotic war on), because the pattern of deaths in these households was no different from those with certificates. 

              The survey tallied all Iraqi deaths, whether combatant or civilian, because this distinction is difficult to make in any objective fashion.  The cause of death was noted, including both non-violent causes such as accident, old age, disease, etc., and violent causes such as explosions, gunshot, air strikes, etc. 

              The results, stated as mortality rates per 1000 people per year, were as follows:  During the 15 months immediately preceding the U.S. invasion in March 2003, the mortality rate was 5.5 per 1000 people per year, while during the 40 post-invasion months the mortality rate was 13.3 per 1000 people per year.  Ponder those numbers:  Since the war began, Iraqis have been dying at more than twice the normal (if the Saddam Hussein era can be called ÒnormalÓ!) rate.  Nearly all the excess deaths were violent, but some were non-violent, perhaps from such causes as the danger of getting expectant mothers and others to hospitals quickly during wartime—a common cause of wartime deaths.  

              Extrapolating to the entire population, during the 40 wartime months there were an estimated 655,000 excess (above the pre-war rate) deaths, 601,000 of them violent and 54,000 non-violent.  In addition, there were 495,000 non-violent non-war-related deaths (the pre-war rate). 

              Like the Gallup Poll, these results come with statistical uncertainties.  Because of the difficulty of surveying under wartime conditions, the number of clusters was only 50.  Epidemiology surveys usually use 100 or even 1000 clusters.  The small number of clusters meant that uncertainties were much larger than the usual 3 percent associated with the Gallup Poll.  For example, the researchers could state with 95 percent confidence that there were between 393,000 and 942,000 excess Iraqi deaths during the war years, of which 426,000-793,000 were violent. 

              The estimated number of violent deaths, 601,000, is twice the number killed by Saddam Hussein during his 20-year rule.  So the annual rate of violent deaths has increased by a factor of 10 since Saddam was replaced by our invasion.  Even the lowest plausible number, 426,000 violent deaths, is enormous in a nation of only 27 million.  Extrapolated to a nation the size of the USA, 426,000 deaths becomes 4.7 million!

              Right wing attempts to discredit this article have gone into overdrive.  The Wall Street Journal hastily published an article by Steven Moore claiming that the survey was statistically flawed.  Moore was formerly director of a pro-Iraq-war website, and a former advisor to Paul Bremer, head of the U.S. provisional authority that governed Iraq right after the invasion.  Moore argues strenuously that the small number of clusters invalidates the survey.  But in fact the small number of clusters simply implies a wide uncertainty range, and even the lowest numbers in this range are far larger than previous estimates. 

              The survey authors published a similar survey of Iraq in 2004, after 18 months of war, with results entirely consistent with the recent results. 

              The Lancet editor, apparently expecting controversy, wrote an editorial supporting the validity of the recent survey, pointing out that the article was reviewed by four expert peers who provided detailed comments, and that all reviewers recommended publication with relatively minor revisions. 

              The Lancet results are certainly more credible than previous passive estimates, and probably correct.  They are shocking and incredibly important.  They imply not only that this war was a dreadful mistake from the start, but also that far more people are killed in modern wars than had been thought.  This has vast implications for anyone contemplating the benefits and costs of any future war.  We will be making a big mistake if we allow our widespread ignorance of science and mathematics, and our right-wing biases, to obscure this survey. 

LINK TO ART HOBSON'S HOMEPAGE