Inside the Numbers:

Michel Martin (of ABC News, and part of the This Week roundtable) needs a lesson in analyzing data. On Sunday she lamented the high-percentage of friendly fire casualties in the current war, and noted similar numbers from the first Gulf War. The number cited was something like 25% deaths from friendly fire. I do not know if that number is accurate, although it sounds about right based on the overall numbers which have been reported.

The problem, of course, is that the reason that in prior wars that friendly fire deaths were a lower percentage of conflict-related deaths is because the absolute numbers of deaths due to the enemy were so high. The ratio of troops on the ground to battle deaths in this war has got to be one of the lowest in history. Further, there is simply a small “N” (i.e., sample size) problem here. With (as of the last report I heard) less than 100 deaths, any category in that number may seem over-represented, due to the low number of absolute deaths reported. For example, let’s say that 30% of the deaths were from grenades, where in the past the number of grenade-related deaths was only 10%. That doesn’t mean that we have become three-times more vulnerable to grenades (indeed, it could many things, or just be a statistical anomaly). Similarly, a high percentage of friendly fire deaths out of a relatively small number of overall deaths, means nothing.

I am not making light of the deaths in question, just making light of Ms. Martin’s analytical skills in this case. It is as if in the face of overwhelming success in the campaign, she had to find something negative to say.