Even with the power of propaganda and a befuddled media, it amazes me that we can be going into a major military policy debate with so little clarity on a few basic statistics upon which much of the debate is going to turn.

Are military deaths up or down? Are Iraqi civilian deaths up or down?

'Success' and 'failure' are quite subjective and they may be judged on various criteria. But these numbers, even if they don't tell the whole story, are concrete and readily ascertainable -- at least in the case of military deaths.

Here for instance is a post at DailyKos referencing an exchange between CNN's Wolf Blitzer and Rep. Charles Boustany (R-LA). The congressman rattles off a series of statistics from the approved Republican talking points about declining numbers of casualties. And Blitzer runs down the list demonstrating pretty clearly that Boustany's numbers -- and thus the numbers pretty much all president's supporters are dishing -- are simply false on their face.

So here we have that classic sort of modern media moment in which we have a debate wherein both sides arguments are fairly and equally represented -- one side with a series of bogus 'facts' and another with actual facts. Both sides get to make their case. And you can decide between them.

I actually got to thinking about this a couple days ago when I read this article on the McClatchy newswire. The headline reads

"U.S. combat deaths drop by half during 'surge'"

The article begins ...

U.S. combat deaths in Iraq have dropped by half in the three months since the buildup of 28,000 additional troops reached full strength, surprising analysts.

U.S. officials had predicted the increase would lead to higher American casualties as the troops "took the fight to the enemy." But that hasn't happened, even though U.S. forces have launched major offensives north and south of Baghdad.

This seemed a little different than what I'd heard. So I went and looked at icasualties.org, the site that keeps track of these numbers and, as far as I know, is widely respected in terms of its statistics.

And here are the numbers they show going back to January ...

August      82
July          79
June        101
May         126
April        104
March       81
February   81
January    83

Now, nowhere in the article did I see a specific explanation of the "drop by half". But I don't see any way to interpret these numbers that makes the McClatchy article even close to right.

If memory serves, the 'surge' came up to full strength in early June. And there clearly has been a decline in troop deaths, although a relatively modest one. And not close to half. What's more, if you look at the numbers over the course of the year, the big picture seems to be a relatively stable baseline of just over 80 fatalties a month with a bump up during the months when the surge was coming up to strength.

In other words, this 'drop by half' headline seems doubly wrong -- both in the big picture of the context of the drop and the small picture of the amount of the drop. In fact, it seems so far off that I'm still trying to think if there's some specific way they're interpreting these numbers that hasn't occurred to me. For instance, are they using a technical definition of combat personnel rather than all military personnel in country? Who can help me here?

Late Update: I'm not sure this is it. But I'm hearing some suggestion that the issue here may be what I speculated -- namely, that the issue is the definition of 'combat deaths'.

Here are the numbers of fatalities, again from icasualties, broken out by 'hostile' and 'non-hostile'.

   Hostile   Non-Hostile   Total
Jan     78     5     83
Feb     71     10     81
Mar     71     10     81
Apr     96     8     104
May    120     6     126
Jun     93     8     101
Jul      67     12     79
Aug     55     27     82

So, the claim still doesn't quite add up. But here at least you can see what they're pointing to. From a high point in May, the number of combat fatalities has dropped to about half what they were. It hard not to notice though that as the number of combat fatalities has dropped the number of non-hostile fatalities seems to be rising dramatically. August had 27, almost one per day. Since the beginning of the war only two months have had more -- May 03 and Jan 05. Has the counting methodology changed?

Even Later Update: These month by month numbers aside, the real story appears most clearly in these year by year comparisons provided by Kevin Drum. Short version: the run-up and run-down in casualty numbers seems far more likely to be caused by the repeated seasonal pattern than the 'surge'.