The release of the wikileaks data on the “secret story” of Afghanistan has led to a frenzy of the usual shallow reportage one comes to expect of the media, and nowhere more shallow than in the sensational finding that death rates of coalition soldiers have increased over the last 7 years. The Guardian even has a “datablog” on this topic, and  a rather sad film about the realities of soldiering in that country[1]. According to the datablogs there has been a “sharp rise” in deaths, and everywhere I look on the internet I just seem to find numbers of deaths.

I’ve been wondering about this for a while because the rhetoric around these numbers – in both opposing and supporting camps – seems to present these deaths as a sign of the catastrophic failure of the project, or the International Security Assistance Force (ISAF) or something. But I’m sure that over the years that “we” have been in Afghanistan, numbers of troops have also been increasing. And as the number of troops increases, so naturally should the number of deaths, yet something as simple as rates of death by available combat troops is not presented anywhere on the internet[2]. The same seems even more likely of IEDs, which have been a central focus of the recent discussion, and the missile attacks on helicopters. If you have 10 times as many troops in the field, you probably have more than 10 times as many flights, which means that there are more than 10 times as many chances for the enemy to finally successfully shoot you down.

A graph of counts of deaths is shown in Figure 1, and it clearly appears that the number of deaths in Afghanistan is sky-rocketing. If one makes the (apparently reasonable) assumption that the number of troops in Afghanistan was at its highest during the “war” then it seems like evidence of a “growing insurgency” and other cliches much-loved by our journalistic “friends.”

Figure 1: Coalition Deaths in Afghanistan, January 2004 - December 2009

However, this isn’t actually the case. It’s hard to find numbers of soldiers in Afghanistan on the internet, though Wikipedia has numbers past 2007. However, I found a report (pdf) from the Congressional Research Service which gives the number of soldiers in terms of “Boots on the Ground”[3].  This is an estimate of the number of soldiers actually at risk of combat, and is presented in the report as the average per month from Fiscal Year 2002 to 2012. This makes it very easy to link the death numbers being bandied about to the number of soldiers actually likely to get shot at, and to calculate a rate. I’ve done this, and the result is shown in Figure 2.

Figure 2: Rate of deaths per 1000 Boots On The Ground in Afghanistan, January 2004 - December 2009

It’s pretty obvious from Figure 2 that the death rate has not changed over 5 years in Afghanistan. What has happened is that the army has poured into the provinces and engaged the Taliban more extensively. It might be considered a failure of military policy to have allowed the death rate to stay static after 5 years but it’s not exactly a sign of a conflict out of control. The same conclusion should probably apply in spades to the IED numbers, and to the success rate for anti-helicopter attacks. In fact I wold argue that a static death rate is merely an example of a conflict not yet won, and one metric by which the US military should be able to assess the security of the provinces – it’s a useful tool, and provided it doesn’t go up they know they aren’t on the wrong track, though they could probably do better (and looking at the figures from the Guardian, in December 2009 a total of 87 civilians, 65 Afghan soldiers, 35 NATO soldiers, and 287 “Taliban” would agree with me if they were still alive).

Other things to note from this data are a clear sign of spikes in deaths at yearly intervals, corresponding roughly to the summer months of June – August[4]. That pattern may not have existed in the first 12 months, which could be due to small numbers or might represent a 12 month period with no insurgency. If they had had 50000 “Boots on the Ground” in January 2004 would things be different now? This is a very feeble possible data point in favour of the conclusion that they might have been.

This data is a really good example of how one has to consider the size of the risk pool in an analysis. I made this point about the HIP scheme, and it applies here too. There are many reasons to oppose the Afghanistan war, but there are enough casualties in this war without adding epidemiology to the list[5].

As a final note, I’m opposed to both America’s current wars, though I understand how it was hard for the US to take any other approach to Afghanistan in 2001. I think it’s a tragic waste of life, especially Afghan, with little benefit for anyone involved in (or subjected) to the thing, and ultimately won’t be successful. Watching that movie on the Guardian website, with some young man dying on screen, his face blurred out because of the damage done, a piece of meat in the hands of both his medical team and his geopolitical masters, makes me angry at the evil futility of modern politics – and of course I don’t even get to see the nastiness dished out to the Afghans (except when wikileaks work their sinister magic). But boutique wars and vengeance dealt cold from 11000′ can be argued against without reference to bad stats. Especially stats as bad as some of what has been presented about Afghanistan!

Oh, and finally, a few caveats:

  • The population I used was for American soldiers only, but I don’t get the impression that the balance of US to non-US soldiers has changed a lot since 2004, and obviously the conclusions here would be weakened if the relative ratios changed significantly recently
  • The death figures I used were ultimately the NATO official figures, not the additional ones from the wikileaks data. This is because the NATO official figures are usually higher than the wikileaks numbers I got from the Guardian. Interesting, that…?
  • This analysis could be made more sophisticated using monthly actual figures of boots on the ground, which are presented in a chart in the linked report, but I couldn’t be bothered because it wouldn’t change the outcome one whit

fn1: watching that certainly puts my nervousness about a kickboxing fight into perspective!

fn2: if you can find a rate, please show me!

fn3: This is apparently a technical term. Secure that shit, Hudson!

fn4: And who wouldn’t jump at the opportunity to spend 3 months lugging a backpack, a heavy weapon, and 5 litres of water into a combat zone in 55C heat? I really do think that this war asks too much of too many young people.

fn5: And lord knows epidemiology has suffered badly in Iraq because of the American right and their objection to all forms of reason or logic.