It’s depressing to see a professor of demography pull this sort of stunt in the Washington Post:

Between March 21, 2003, when the first military death was recorded in Iraq, and March 31, 2006, there were 2,321 deaths among American troops in Iraq. Seventy-nine percent were a result of action by hostile forces. Troops spent a total of 592,002 “person-years” in Iraq during this period. The ratio of deaths to person-years, .00392, or 3.92 deaths per 1,000 person-years, is the death rate of military personnel in Iraq. … One meaningful comparison is to the civilian population of the United States. That rate was 8.42 per 1,000 in 2003, more than twice that for military personnel in Iraq.

What a joke. Note that the authors (Samuel H. Preston and Emily Buzzell) actually compare the crude death rate for the entire population of the United States to that of U.S. service personnel in Iraq. Who knew so many people died in America from every conceivable cause every day? There ought to be a law.

Preston and Buzzell go on to say,

The comparison is imperfect, of course, because a much higher fraction of the American population is elderly and subject to higher death rates from degenerative diseases. The death rate for U.S. men ages 18 to 39 in 2003 was 1.53 per 1,000—39 percent of that of troops in Iraq.

For “imperfect” here read “completely inappropriate,” or “dumber than a can of Cheez Wiz.” They continue:

But one can also find something equivalent to combat conditions on home soil. The death rate for African American men ages 20 to 34 in Philadelphia was 4.37 per 1,000 in 2002, 11 percent higher than among troops in Iraq. Slightly more than half the Philadelphia deaths were homicides.

Inevitably, people like Glenn Reynolds see numbers like this and—instead of thinking “Christ, inner city homicide rates are even remotely comparable to military deaths in Iraq!"—think, hey, “it’s hard to look at these numbers and see the catastrophe” that the mainstream media report on.

Look, it’s a well-known fact about the sociology of combat that even in a real, live, shooting war, only a comparatively small number of troops in an army ever see direct, front-line duty—if only because the number of people it takes to sustain those who do go out to the front line, or its equivalent, is very large. (Don’t get me wrong: many of those in support roles will face real dangers, too, and their lives will be very far from normal—it’s just that we’re talking about violent death rates here.) In fact, even amongst the front-line troops, exposure is more focused and limited than you might think. A similar thing is true of bombing campaigns in built-up areas, such as the recent one in the Lebanon. An awful lot of bombs can be dropped and an awful lot of buildings destroyed, and the deaths will be fewer than you might think. But that doesn’t mean it’s not dangerous, or completely dysfunctional. Who would think to say, “Hey, a bunch of buildings were destroyed by the bombing and hundreds died, but that’s fewer people per capita than will die of heart disease this quarter”?

This is why comparisons to death rates in civilian settings—even comparatively violent ones—are misguided. Anyone who thinks that someone walking around Philly is more likely to be violently attacked than a marine out on patrol in Baghdad is out of their mind. Moreover, troops on patrol are kitted out with protective gear, travel in well-organized groups, and have guns. And yet they still die in large numbers. Crude comparisons of death rates across very different settings mask big differentials in exposure to violent incidents, ignore fundamental differences in the structure of those incidents, and—in the case of military fatalities—ignore the huge improvements in field medicine that (according to data for 2004) allowed the ratio of wounded to killed soldiers in Iraq to be more than two and a half times what it was in Vietnam. Bear in mind, too, that all of what I’ve said so far ignores the elephant in the room, which is that the death rates in the article refer exclusively to U.S. forces on active duty in the whole of Iraq and not to regular Iraqi civilians. Contrary to what you may have heard, these people are not magically immune to the effects of car bombs, death squads, or suicide bombers.

While such comparisons are basically misguided, Preston and Buzzell would have been better off calculating the standardized death rate for police officers in the U.S. as a whole and comparing that number to U.S. military deaths across the whole of Iraq. (Alternatively, they could have looked at violent death rates for civilians in Baghdad and compared that number to Philadelphia’s inner city, but I doubt the data are available, precisely because Baghdad is such a disaster area.) Here’s a quick estimate for the case of law enforcement. The BLS says that there are about 624,000 “Police and sheriff’s patrol officers” in the United States, not counting detectives, supervisors or other staff. In 2004, OSHA reports that 121 officers died in the line of duty, with 66 dying in some kind of road accident and 48 dying in assaults or other violent acts. So this ratio of 121 to 624,000—0.000193—gives us a rate of 0.193 deaths per 1,000 for 2004. Being a police officer is not an easy job, yet (you should not by now be surprised to learn) the line-of-duty death rate is an order of magnitude lower than the overall U.S. death rate of 8.42 per thousand cited by Preston and Buzzell, to say nothing of the 3.92 per 1,000 they calculate for U.S. soldiers in Iraq.

All of my own caveats apply to using the number I just calculated to assess the seriousness of military death rates in Iraq. As I’ve said before, the acid test is quite straightforward. Would you—can you?—take a commercial flight to Baghdad tomorrow, get a taxi from the airport to the city, stay at a local hotel, see some sights and eat out at a decent restaurant without being in fear of your life? What about Philadelphia? (I’ll grant a cheesesteak exception on the fear-for-your-life part.) This test would have been passed by cities like Derry or Belfast for almost all of the period between 1970 to the present, so it’s not even a very high bar. I doubt that Preston and Buzzell are packing their bags for Baghdad, their own calculations on comparative death rates notwithstanding.

Update: Since I drafted this, Reynolds has gotten a telling-off from someone he’ll listen to—his correspondent makes more or less the same points I make here.