Fatally Flawed Numbers
There have been a number of brilliant Brits figuring out mathematics and statistical probabilities, and unlike the prestigious British medical journal Lancet, none of them was inclined to make the same mistake twice.
Richard Price made the mathematical deductions necessary to establish the scientific basis for estimating insurance risk and mortality. Thomas Bayes proved that the likelihood that something will happen can be plausibly estimated by how often it has occurred in the past (a theory that will become crucial to web engine searches). Thomas Malthus caused a storm when he mathematically proved that population increases in a geometric ratio and far exceeds food supplies, which at that time grew arithmetically. And Dr John Snow plotted the location of deaths from cholera in London on a map, discovered the cause of the cholera and ended the epidemic while establishing statistical mapping as an invaluable tool in medical epidemiology.
I mention these positive advances as a contrast to a rather lamentable exception.
In 2004, shortly before the American Presidential election, the Lancet, one of the oldest and most respected peer-reviewed medical journals in the world, published a report about the deaths of Iraqis due to the war. In a study subsequently declared deeply flawed, Lancet pronounced, “We estimate there were 98,000 extra deaths (95% CI 8000-194,000) during the post-war period.”
Fred Kaplan in Slate explains what the crucial numbers inside the parentheses mean. Astonishingly they mean that “the authors are 95 percent confident that the war-caused deaths totaled some number between 8,000 and 194,000. (The number cited in plain language — 98,000 — is roughly at the halfway point in this absurdly vast range.)”
This week we have a second report from Lancet. Apparently wishing to embody Bayesian probability theory by repeating the same errors they made in the past, the journal has for a second time published suspect findings just before an American election. This time their methodology was based on 547 actual deaths from which they extrapolated that 654,965 Iraqis had died in war-related deaths.
This unsupported number is a disservice to rational thought and to Lancet's reputation.
According to Kaplan, "There is one group out there counting civilian casualties in a way that's tangible, specific, and very useful—a team of mainly British researchers. . . called Iraq Body Count >" Their figures to date suggest that a minimum of 43,850 Iraqis and a maximum of 48,693 Iraqis have died due to the war and the targeting of civilians by terrorists.
These losses are terrible, but they do not add up to half a million people, and they are nowhere near the hundreds of thousands of Iraqis that Saddam Hussein murdered before the war.
That's more than you probably wanted to hear from me on this subject. What is it that bothers me? The idea that people at Lancet are not any better than I am at figures, or my fear they don't care about numbers and are willing to dress them up as something they are not?
October 16 Update. Instapundit reports that "IRAQ BODY COUNT, often criticized for offering inflated civilian death figures, is now criticizing the Lancet study for offering inflated civilian death figures.