Friday, May 30, 2008

Statistical Analysis of Decapitation as a Counter-Terror Strategy

The most recent edition of The Journal of International Policy Solutions published a statistical analysis I wrote on the efficacy of killing or capturing the top leaders of a terrorist organizations. Entitled "Testing The Snake Head Strategy: Does Killing or Capturing its Leaders Reduce a Terrorist Group's Activity?" the article can be read in its entirety here a summary of the method and findings follows.

It is conventional wisdom that removing an organization's leaders is an effective counter-terror strategy, but the quantitative analysis is less clear on the issue. Most of the successes focus on specific instances, such as the collapse of Sendero Luminoso in Peru after its leaders were removed. There are also examples on the other side, such as Hezbollah's increased deadliness and effectiveness after Israel's 1992 assassination of Hezbollah Secretary-General Abbas Musawi.

This study was an attempt to shed some light on the issue, focusing strictly on removing top leaders (#1 or #2 - so OBL or Zawahiri would count, but Khalid Sheikh Mohammed did not.) Sixty cases of terrorist leaders being killed, dying, or being captured and imprisoned for lengthy periods were identified. Using the Terrorism Knowledge Base, which is now sadly not available to the public anymore, I gathered data on incidents and fatalities in the two and five year periods before and after the organization lost its leader. Because there are lots of reasons why a terrorist group's level of activity might change I tested these results against a comparison group of 21 terrorist groups that did not lose a leader.

The Results

The data-set was relatively small so most of the results were not statistically significant. There was a trend of lower numbers of incidents after a group lost its leader, a trend that increased when a group lost its leader more than once. On the other hand there was an indication that the number of fatalities by Islamist groups increase after they lose a leader. Building on that when an Islamist leader is killed, rather than arrested, the increase is even greater.

None of these findings are rock solid and there are many other factors that could effect a terrorist group's activity. In particular, the increase in killings by Islamist groups after their leader is killed could reflect that the killing occurred i n the midst of a large-scale war (such as Chechnya or Algeria which were included in this study.) Still it is a potential cause for concern. Iraq was not included in this study, in great part because the data was too complicated to work with.

This is another key point the paper discusses, the data problems and other challenges (for example what should be measured - 9/11 was nearly 3000 deaths, but 4 incidents) in doing any quantitative analysis of terrorist activity. The paper also reviews other statistical studies of decapitation as a counter-terror strategy. At this point, my overall conclusions are the old academic fallback, "More study is needed." The paper ends:
Tolstoy began Anna Karenina by noting, "Happy families are all alike, every unhappy family is unhappy in its own way." Like unhappy families, terrorist groups differ from each other and finding universally applicable rules to understand them may not be realistic. Regardless of the quantitative results, decapitation will remain a counter-terror strategy. If a quantitative test can help indicate where and when it is most likely to be effective or have deleterious results, it can help conserve scarce counter-terror resources and avoid exacerbating situations where the decapitation strategy may be counter-productive."


Cernig said...

I'd like to see some wonkish thought on targeted decapitation - i.e. where intel can identify more moderate leaders-in-waiting and ensure they're eventually the only leaders of a terror group. The concept is one of artificial accelerated evolution of the group towards moderate aims by simply allowing the more moderate to survive. Arguably, the British government tried this in N.I. and ended up with the likes of McGuinness and Adams.

Aaron Mannes said...

Thanks for your thoughtful comment. To some extent, targeting leaders relies on opportunity - so the state strikes when it can and does not have the range of choices it likes. The record of identifying moderate leaders is a dicey one (even assuming the cadre of radicals leaders can be successfully neutralized). First it can be tough to know how the organization will change when a leader is removed and also in many groups there are no moderates (say Hezbollah, Hamas or al-Qaeda.)

This would have to be qualitative research as the dataset would be too tiny to allow significant conclusions.

As it happens I've been thinking about the IRA lately and will hopefully post some thoughts on it.

Anonymous said...

Your link doesn't seem to go to the article mentioned.

Aaron Mannes said...

Sorry about that here is the updated link -

I confess the paper has some methodological issues, but the key point remains that how a group will react to a decapitation strike appears to depend heavily on the specific group.