Thursday, September 13, 2007

Iraq, the US Surge, and Statistics

When General David Petraeus presented his report (pdf) to the US and the world this week, he made important claims about a decrease in violence and casualties. The first key figure from his presentation is reproduced below. He claims that firstly, there is a decreasing rate of attacks on US troops and related targets, and secondly, there is a decreasing rate of civilian casualties associated with the surge. But how reliable are the data used to make the claims and how well does it compare to other available sources of information?

To assess the first claim we looked at the data for US casualties compiled by Iraq Coalition Casualty Count. The two images below are graphs we produced of their fatality and wounded monitoring tables for the period September 2006 - 2007 and July 2006-2007 respectively. It can be seen that the pattern is somewhat different than the graphs of attack rates shown above by Petraeus. While the graph of US coalition fatalities does show a marked decline between May and July this does not correspond to the period of the surge. The 'surge' may be considered as either comprising February 2007 onwards, or, comprising the period of the surge offensives, which Petraeus presents as occurring from 16 June onwards. In either event, by August the downwards trend has stopped and fatality rates are no longer declining. Current fatality rates during September this year are 2.8 deaths/day, higher than they were in September 2006 when the death rate stood at 2.6/day.

Examining the available data for injuries reveals a similarly inconclusive picture. Between July 2006 and July 2007 (months for which complete data are available) it is impossible to identify any period of decline that relates to the surge.

It should also be noted that as the US were surging in Bagdhad, the British were themselves experiencing a dramatic upsurge in attacks and casualties in the south of Iraq, the two events being perhaps more closely linked than either government may wish to acknowledge.

In conclusion, we have to say that there is no evidence, as yet, to suggest a decline in US casualties relating to the surge. The decline in attacks reported by Petraeus is therefore puzzling. Assuming that the decline shown by his attack rate is accurate, the only conclusion that can be drawn is that attacks by the Iraqi resistance against US forces are becoming more effective and deadly. This is no good news for the US.

The second major claim made in the presentation by Petraeus is that civilian casualties have declined.
"Civilian deaths of all categories, less natural causes, have also declined considerably, by over 45% Iraq-wide since the height of the sectarian violence in December."
If true, this is obviously an encouraging statement for all concerned, but the first problem is the source of the data that he cites. There are many serious deficiencies in 'host and coalition' reporting capacity in this area. Indeed, the Iraqi government recently refused to share its figures with the UN after the UN produced much higher estimates of civilian deaths. As for the US military, its previous commander of the Iraq campaign General Tommy Franks, infamously claimed "We don't do body counts". While this assertion was wisely taken with more than a pinch of salt, the US military has no known capacity for reliably recording or reporting total civilian casualties. So why should this current and very convenient claim on a decline in casualties be taken seriously now?

Looking at other sources of data shows: some support from the media monitoring project, Iraq Body Count, for a small decline in casualties, but nothing like the 45% reduction claimed by Petraeus. Media monitoring data of Iraqi civilian and security force casualties maintained by Iraq Coalition Casualty Monitor shows no clear trend associated with the surge.

But maybe the Iraqi people should have the last say on the success or otherwise of the surge. The BBC summarise the results of a poll of Iraqis indicating that "About 70% of Iraqis believe security has deteriorated in the area covered by the US military "surge" of the past six months".

Update: An analysis of casulaties during the 'surge' up to February 2008 is now available