mousey musings

Plenty of publicity is being given to the new study estimating more than 600,000 fatalities in Iraq. The study is problematic, although it appears to use standard techniques for estimating deaths. Here is my best estimate of the total number killed in the Iraq War as of October 9, 2006. ALLIED FORCES: 2988 Killed (from iCasualties.org) IRAQI POLICE/MILITARY (Pro-Allied): 5507 Killed (from iCasualties.org) IRAQI MILITARY DURING CONVENTIONAL WAR (Anti-Allied): Figures here range from under 5000 to about 30,000. In general, estimates of military dead during wars tend to be inflated. The later estimates tend to be lower ones. Given the range of figures, my best guess is that 10,000 Iraqi soldiers died in the six-week conventional war against allied forces. The only significant digit is the 1. That is, the real figure could be anywhere between 5000 and 14,999. Still, 10,000 is the best estimate I can come up with. INSURGENTS: The range of estimates is 5000 to 50,000. As with other figures for battle-deaths, the lower estimates are more likely to prove correct in the long run. The 5000 estimate is a little long in the tooth; al Qaeda claims that 4000 foreigners have been killed, and everyone seems to agree that domestic insurgents are doing more fighting than the foreigners. The Iraqi government suggests figures in the 7000 range while the US government's own estimates have exceeded 15,000. My best guess is again 10,000 with the same caveat about having only one significant digit. CIVILIANS: This is the hardest one to answer. The Iraq Body Count Project records media reports of civilian deaths, which provide a range of about 44,000 to 49,000. These are almost certainly underestimates, since not every civilian death is reported by the media, particularly after major engagements in urban areas. On the other hand, the recently-released MIT study in The Lancet extrapolates from family interviews to a figure of more than 600,000 civilians killed. This method of counting deaths is known to dramatically overstate fatalities for two reasons: the same death is reported by multiple family members, and the populations interviewed are nearly always those most heavily affected by war (refugees, people in areas of recent fighting, etc). This study claims to avoid the second problem (I have sincere doubts about the alleged randomness of its sample), but adds a new one: it compares mortality rates with a "baseline" before the invasion, which requires us to believe that death rates from disease would have remained constant absent the war. I suspect they were on the rise anyway due to sanctions. A reasonable low estimate would be 50,000 (only a few thousand unreported deaths) while a high estimate would be 100,000 (more than half of deaths go unmentioned by the media). I tend to suspect the lower numbers are closer to the truth, given the well-known tendency of news media accounts to exaggerate casualties in virtually every war. I would estimate 50,000 to 80,000, with a mean estimate of 65,000 -- but the range of error is huge. Adding up the "best guesses" I can make: 2988 + 5507 + 10,000 + 10,000 + 65,000 = 93,495. Keeping in mind that most of my estimates are significant only in the 10,000s range, I would guess 90,000 dead, a figure that might plausibly be as low as 80,000 or as high as 100,000. Given current levels of killing, we are likely to hit the 100,000 dead mark in Iraq very soon. This is a very bloody war. As for 600,000+ figures, the weight of evidence leans against these estimates. This doesn't mean they're wrong. Further research using different methodologies might find similar results. If this happens, I will abandon my estimates for the ones that provide the "best bet" as scientific estimates. For right now, I require extraordinary proof for such extraordinary claims.

posted by Jeffrey # 10/11/2006 01:23:00 AM