Of statistics and lies at Boston.com

Journalists often struggle with statistics. It's just a fact of life, since journalism demands a bit of talent for writing, a good bit of general knowledge, and a willingness to work for a bit more than nothing.

So let's just say that, self-evaluation aside, journalism is not done by the creme de la creme of society.

Add another example to the infinitude of past statistical gaffes, though in this case the error is merely implied.

Boston.com with the Boston Globe's Bryan Bender on the death tolls in Afghanistan and Iraq:

Compared with past conflicts, the percentage of nonhostile deaths is at a historic high. A full 20 percent - or 1 out of every 5 - of all US troop deaths in Iraq have been the result of nonhostile wounds. In Afghanistan - where far fewer US troops have been engaged in combat - the share of nonhostile deaths is nearly half of the total, although, with violence increasing, hostile deaths are on the increase.

The theme of the story is the supposed increase of accidental deaths, and the story implies that the accidental deaths are especially high lately. But that inference is as shaky as could be.

As the story points out, the stresses of deployment can increase the chances of accident, but the "historic high" from the quoted paragraph contributes virtually nothing to that thesis. The numbers contain two key variables, the number of non-hostile deaths and the number of hostile deaths. An especially low number of hostile deaths will make the number of non-hostile deaths rise by percentage.

So are the conflicts in Afghanistan and Iraq very low in hostile deaths compared to other past conflicts. Yes, as a matter of fact they are.

And are military deaths fairly commonplace even when no conflict is occurring? Yes, as a matter of fact they are.