Slate’s Weisberg: Rounding as an Editorial Art Form

Slate’s Jacob Weisberg was on the Slate/NPR radio program Day to Day this morning, discussing President Bush’s appearance at the U.N. One particular statement caught me by surprise:
(transcript is, as usual, my own from repeated listening of the RealAudio, the relevant section begins at about 4 minutes in):
Weisberg: We’re taking casualties in Iraq every day; over the last few days, U.S. soldiers have been killed at the rate of two a day. I think there’s a general acknowledgement that the occupation is not going well — perhaps it could be going worse, but it’s not going well and there’s no end in sight.
(italics mine).
There’s only one problem with Weisberg’s flat statement on the casualty rate in Iraq: it just isn’t true.
The most easily accessible source I have found for data on casualties is the impressive website at simply titled Iraqi Coalition Casualty Count. As the site explains in their methodology page, the Department of Defense itself doesn’t make it terribly easy to find day-by-day data on casualties, although they do provide the raw data in the form of press releases and (extremely) high level summaries (PDF). Lunaville appears to be stepping up to the plate to try to provide the information in a searchable database.
That said, back to Weisberg’s statement. We’ll assume that when he stated that “we’re taking casualties in Iraq every day” he included wounded personnel as well as those killed — and by that definition he is almost certainly correct. But when he says “over the last few days, U.S. soldiers have been killed at the rate of two a day”, the facts don’t seem to support him.
Lunaville shows the following counts for U.S. soldiers killed over the past week:
9/22/2003 1
9/21/2003 0
9/20/2003 3
9/19/2003 0
9/18/2003 4
9/17/2003 0
9/16/2003 0
Depending on how you interpret Weisberg’s “last few days”, you get a different number for the average deaths per day. But there is no mathematical way you can make that number come out to even approach 2. If you look at the past three days, you get an average of 1.33; four days gives you 1.0; five days gives the maximum at 1.6, with six days at 1.33 again and the last week’s average being 1.14.
How about the whole month of September? Lunaville shows 21 U.S. soldiers killed in the past 22 days; an average of 0.95 a day. The month of August? 35 U.S. deaths in 31 days — an average of 1.13 a day.
All of these calculations are, of course, completely dependent on the source data. I am assuming Lunaville‘s counts are accurate, as they appear to be quite serious about their task and I’ve seen no evidence to contradict them. But be your own judge; that’s what the hyperlink is for.
But if you do accept the source numbers, there’s no way to avoid concluding that Weisberg’s statement that “over the last few days, U.S. soldiers have been killed at the rate of two a day” is simply wrong. If the question of just how bad things are in Iraq right now wasn’t so critical, I might be inclined to let a statement like this slide. But this issue is critical, and if journalists are going to declare Iraq to be “not going well”, we at least should demand that they get the numbers right.
What’s that you say? Weisberg was simply speaking during a radio interview, and that taking an exact spoken quote and dissecting it like this is a tad unfair? Well, I’m sure Weisberg himself might agree. Or maybe not.
Finally, in taking potshots at Mr. Weisberg, I don’t want to commit the offense of simply reducing these dreadful losses to mere numbers. Yeah, it’s not two a day, but it is still too damned many. And so I encourage you to go here, or here to see the faces of the soldiers who will not be coming home. What was lost with them cannot be replaced for their families and loved ones; and I can only offer my thoughts and sympathies to those that they have left behind.