Looking through the book Managing Complexity: Insights, Concepts, Applications by Dirk Helbing (editor) I stumbled on this article, and it sounded interesting. Reading it, it investigates prior and new work regarding human conflicts - especially distributions of size of casualty numbers in different wars.
Citing older work the distribution is most of the time a power law distribution with varying power coefficients. New work done by the author is to look at the distribution of casualties of events belonging to one war - and they also follow powerlaws (data for Iraq and Columbia are investigated) - and compared for terroristic events.
The power coefficient is in these at roughly 2.5. Making a further analysis of this coefficient as it changes during the war time (by subpartitioning the data) one sees that for the Iraq war it goes from lower to 2.5 (from large armies to insurgents), for the Columbia war from higher to 2.5 (small disorganized to better organized insurgents). Terroristic casualties are following a power law with about 2.5.
One more thing is the analysis of time sequence of events, compared to randomized set of these events: this shows a difference, meaning that the time-sequences are non-random, so have at least some systematic order in it (unfortunately no more details are given on that).
The power-law behavior might not be too surprising as it is the result of random group acts in this case, without any given 'scale'.
The article is interesting but I find it a bit introductory (it starts by explaining normal vs. power-law distributions), still the application of scientific statistical methods to this area is an important contribution.