Playing with the numbers

John Leo

7/25/2005 12:00:00 AM - John Leo
Isn't it awful, a friend said at dinner the other night, that 100,000 Iraqi civilians have died since the U.S. invasion? When I asked where the sta­tistic came from, he said maybe it was 8,000, but def­ initely somewhere between 8,000 and 100,000. That's a pretty broad spread, so I decided to do some checking.

The 100,000 estimate is from a survey of Iraqi households conducted last year by a team of scholars from Johns Hop­kins University and published in a British medical jour­nal, the Lancet. As luck would have it, the team was an­tiwar, and the study was released just before the presidential election. The study's coauthor called the 100,000 figure "a conservative estimate," the customary phrase attached to politically useful wild guesses. The study said, "We estimate there were 98,000 extra deaths (95% CI 8,000-194,000) during the postwar period."

Writing on Slate, Fred Kaplan translated that little tech­nical phrase between the parentheses: It means that the authors are 95 percent certain that war-caused deaths totaled somewhere between 8,000 and 194,000. Kaplan's conclusions: "The math is too vague to be useful."

Iraq Body Count and the Oxford Research Group, Britain-based antiwar organizations, released an analysis of Iraqi civilian fatalities last week, based on their collection of media reports ( It said 24,865 civilians had died in the first two years after the invasion, with U.S.-led forces accounting for 37 percent of the total, criminal violence 36 percent, and "antioccu­pation forces/insurgents" 9 per­cent. The Times of London dismissed the study as "an entirely arbitrary figure published by political agitators." But Michael O'Hanlon, who tracks statistics on Iraq at the Brookings Insti­tution, says the study "is proba­bly not far off, and it's certain­ly a more serious work than the Lancet report."

The modern numbers game of war dead began with the Gulf War. Greenpeace said 15,000 Iraqi civilians died. The Amer­ican Friends Service Commit­tee/Red Crescent claimed that 300,000 civilians died. Various media assessments hovered around 1,200. Later, Foreign Policy magazine put the civilian dead at 1,000. Unsurprisingly, the high estimates come from antiwar groups, often described in the media as neutral and nonpartisan. A New York Times article during the Afghan war ("Flaws in U.S. Air War Left Hundred of Civilians Dead") relied heavily on Global Ex­change, a hard-left, pro-Fidel Castro group blandly iden­tified by the Times as "an American organization that has sent survey teams into Afghan villages."

Today, yet another round of inflated estimates is breaking out, this one on the number of homeless veterans. A UPI story a few months back reported that nearly 300,000 veterans are homeless on any given night.

If so, as blogger Megan McArdle pointed out a few weeks ago on Asymmetrical Information, that would mean that every single homeless person in America must have served in the armed forces, since 300,000 is about the total num­ber of the homeless. The 2000 census, covering people living in shelters but not those living on the street, counted only 170,706 homeless people. The Department of Hous­ing and Urban Development asked cities and counties get­ting federal aid for the homeless to provide statistically valid counts. New York City reported 40,000 homeless, Los An­geles County 90,000,and Chicago 9,600.

The problem here is a familiar one. "Advocates for the homeless," as they are called in the usual press catchphrase, cannot resist passing on wildly inflated numbers. The pi­oneer here was the late Mitch Snyder, a prominent advo­cate, who admitted making up the "fact" that there were "many millions" of homeless in America to give the cause more leverage. The media accepted that estimate for years, though it was surely far higher than the actual number. Now the numbers foisted on the media have soared again. The Department of Veterans Affairs says that some 250,000 vets are living on the street on any given night. Since the department says that number accounts for something like a third of all homeless, this means they are working with a total estimate of more than 750,000 homeless.

This makes the department a piker compared with the Urban Institute and the National Sur­vey of Homeless Assistance Providers, which say, in a joint study, that between 2.3 million and 3.5 million people (and 529,000 to 840,000 veterans) are homeless at some time during the year. The lesson? Don't trust advocacy numbers.