Statistics 101

There's a lot of buzz today about violent deaths in Iraq trending down since last month. Encouraging as it is, I'm not about to get too excited about it because when this trend is put in context, its not as conclusive as we should all hope it to be.

As much as I would love to attribute this decline to Petraeus and his new counterinsurgent strategy, I don't really know that that's the case here. When viewed on a longer timeline, is becomes clear that this is the most recent dip in what has historically been a cycle.

Perhaps more misleading are reports that this month's Iraqi civilian deaths have been cut by more than half. Not misleading because the reports aren't true, mind you. They are. However, most people should recall one particular incident from August that makes that month a statistical outlier; a bombing in Qahtaniya that killed about 500 Iraqi civilians.

Again, not that the trend hasn't been down. Not that anyone is lying when they tell you casualties were lower in September than August, they're absolutely right. However, they're right in part because Iraq's most deadly bombing occurred in August.

So in the interest of more accurate data, let's remove that outlier by taking 500 civilian deaths off of the August tally. This gives us a total for August of about 1,475 recorded civilian deaths (1,975-500). September's total? 922 recorded civilian deaths. So the drop is much more like 37% instead of 53%, if we're interested in using these statistics to give us magnitude as well as direction.

No doubt about it, this is still a very positive figure and no doubt the people in charge - Gen. Petraeus and his staff - know much more than we do and know what the trends are better than we do.

Don't get me wrong, I'm delighted to hear that casualties are down. I'm positively thrilled. But I'm also reserving judgment until the data is more conclusive.