Saturday, August 26, 2006

Iraq: Playing with Numbers

Samuel H. Preston and Emily Buzzell have an article in The Washington Post on the statistical risks of our soldiers in Iraq; I have a number of problems with the article but here's the one paragraph I'm truly puzzled by:
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.

If you divide 592,002 "person-years" by roughly three years (we'll ignore the extra ten days), you get an average of 197,000 American troops in Iraq at any given time. I haven't kept track of the precise numbers, but I believe the numbers reported have varied most of the time between 120,000 and 170,000 with 140,000 troops in Iraq being the most common number quoted over the last three years. How do the authors get those extra 50,000 troops per year? Although the Iraq war has been far less deadly to our soldiers than most of our other wars, I suspect that the purpose of the article, in my opinion, is to obscure the risk to our soldiers and an extra 50,000 fictional troops would do that. It's possible that they're counting Air Force and Navy personnel who are stationed in the Persian Gulf and not in Iraq or it's possible the military handed them numbers that count from the moment our troops ship out from the US to the moment they return but the risk for our troops exists while they're in Iraq. I hope the authors clarify their methods. Or do they know something the rest of us don't?


Anonymous S.W. Anderson said...

My guess is they're counting not only Air Force and Navy personnel, but also forces in Afghanistan and elsewhere in the region. They're probably approaching it as a theater of operations.

But who knows? Given their approach to managing money and scientific reports, this could be their equivalent of:

"You put the lime in the coconut and stir it all together.

"You put the lime in the coconut and then you feel better."

11:36 PM  
Anonymous Anonymous said...

In response to Craig, if one soldier is in Iraq for two years, that counts as two person years. Therefore, you can't just divide person years by years to arrive at persons.

3:17 PM  
Anonymous Craig said...

anonymous, the question is what is the risk to our soldiers in Iraq. To get that risk, you have to determine how many of our troops are in Iraq at any given time. It doesn't make any difference whether one soldier has been there three weeks or three years if you're talking about the average risk for everyone.

Here's another way to look at this. The article mentioned people living in urban areas in America as a comparison. In such an area, the population rises and falls; even if the population is roughly stable at say a 100 thousand, there are people moving in to the area and people moving out. With such an area, it's statistically okay to assume an average population without tracking down every person and find out how long they lived in that area. With Iraq, it's mildly more difficult to come up with an average population of our troops because of the large numbers of soldiers coming and going and because of different troop levels but the authors chose a more technical though seemingly more accurate way of getting a total (person years) to use for calculating risk. But their technical method appears to have inflated the average number of troops in Iraq which throws off their calculation of risk (possibly because they were given flawed data in the first place).

Again, if the average number of troops in Iraq at any given time is 150 thousand, and that number stays stable, a quick way to calculate person years for three years is to multiply 150 thousand by 3 and you get 450,000 person years. That's not the number who have served in Iraq (it was well over a million the last time I checked), that's just the total troop commitment to Iraq in person years and you just divide by three to get the average population.

I'm suspicious that the authors couldn't give us or didn't bother to give us an average number of troops in Iraq as a comparison. Their numbers suggest a lower risk than I believe the reality is.

4:52 PM  
Anonymous Longevity Science said...

Thank you for your interesting post!
Once you have cited the work of Prof. Samuel Preston, perhaps you may find it interesting to take a look at discussion of his book "Fatal Years":
Longevity Science: Fatal Years

11:54 AM  

Post a Comment

<< Home