Yesterday I did a little number crunching. This morning, I decided to chart some of those numbers to make it easier to see trends. The first thing to note is that I did replace one data value – April 2004’s number of 126 deaths – with the grand mean of 54.15789. While it wasn’t technically an outlier, it was awfully close, and it obfuscated some of the trends on the chart.
So, lets look at the graph that “pro-surge” people are going to take the most comfort from. This is a chart of troop deaths vs month, with a line for each year (plus the monthly average).
One might be tempted to look at the dropoff in deaths after May as proof the surge is working, but I’m not sure I’d agree. For starters, as a whole, with the exception of August, 2007 is by far the most violent year in Iraq so far in terms of troop deaths. In general, I’d say that’s a big strike against the “things are going well” hypothesis. Second, you can see there is a “bump” in activity across all the years in the late spring/early summer time frame with activity trailing off as you move into June and July, so I’m not entirely sure how much of that summer downturn is surge, and how much of that is seasonal variability. (I’d do an actual statistical test, but (a) I’d have to find the CD for the software, and (b) I’d have to remember how to use the software, as it’s been two years since I did p-value statistical analysis of experimental data, and (c) I have gainful employment that prevents me from spending that amount of time right now.)
Lets look at a second chart: deaths vs calendar year (with each month as a data line). It’s a different view of the same data, but it more sharply controls for seasonal variability, and it paints a pretty ugly picture.
With the exception of March (the initial invasion month, so numbers are skewed upwards in 2003), it’s a pretty clear trend that every year each month has tended to get more deadly. 2006 was actually an improvement on 2005, but 2007 was a big step in the wrong direction.
Of course, the deaths of US soldiers are far from the only metric that can (or should!) be examined if one wants to determine if “the surge is working”. Has the surge resulted in a downturn in US deaths? Possibly. Has it made the situation in Iraq better as compared to last year? Absolutely not. Are we any closer to wrapping this ugly business up? Clearly not.