Saturday, September 25, 2004

Rumsfeld's Misuse of Statistics

Rumsfeld on Friday:

We had something like 200 or 300 or 400 people killed in many of the major cities of America last year. Is it perfectly peaceful? No. What's the difference? We just didn't see each homicide in every major city in the United States on television every night. It happens here in this city, in every major city in the world. Across Europe, across the Middle East, people are being killed. People do bad things to each other.

What's the difference? Let's see:

In 2000, the United States had a homicide rate of 6.1 per 100,000 people. (I chose 2000, to be nice to Rumsfeld, 2001 is higher). If this same homicide rate is applied to Iraq, then given a population of roughly 25 million, we'd expect Iraqis to have approximately 1,525 violent deaths in a one-year period. Though data on Iraqi civilian deaths is hard to obtain, one study that covered only Baghdad and three provinces out of twenty-one counted the number of violent deaths as at least 5,558 from May 2003 to May 2004. In other words, the risk of violent death for Iraqi civilians is almost four time as high as the risk of violent death in the United States, even when we use an estimate of Iraqi deaths that is obviously an enormous understatement. Some sources estimate the Iraqi civilian deaths during the same time period as 10,000, which would make Iraq more than six times as dangerous as the U.S..

These calculations are not meant to be taken as precise for obvious reasons (and for the fact that I can't verify that the U.S. homicide figure actually covers all violent deaths in the country), but they point out that Rumsfeld should eat his hat for saying such stupid things. If he has a hat.
Thanks for the original link to Atrios.