Thursday, October 2, 2008

Fast And Frugal Conflict Early Warning In Sub-Saharan Africa (Original Research)

Now that AFRICOM has officially stood up, there is likely to be a much greater emphasis on understanding the culture, economics, politics and seemingly inevitable conflicts of sub-Saharan Africa. While a number of authoritative sources are busy cranking out some really good products on Africa, simple, effective conflict early warning systems seem to be in short supply.

With this as background, one of Mercyhurst's graduate students, Bradley Perry, recently took a stab at trying to come up with such a system in his thesis, "Fast and Frugal Conflict Early Warning in Sub-Saharan Africa: The Role of Intelligence Analysis."

Bradley was in a unique position to write this thesis. In the first place, he came to us a bit later in life than most grad students, having spent a number of years in Ghana. In addition, while technically here, he completed his strategic intelligence project (on local reactions to a planned expansion of a national park) from a tent in Malawi. He was in Kenya about the same time as the recent upheavals there and is now working for iJet (where he was a member of the iJet team that recently took a share of the prize at the ODNI's Open Source Conference).

Riffing on Gerd Gigerenzer's research (outlined in his fascinating book, Gut Feelings) regarding "fast and frugal" evaluative systems, Bradley went looking for "good enough" indicators of potential conflict that he could chain together to form a predictive model. He found three promising indicators in the literature, political freedom, ethnic homogeneity, and income inequality, and proceeded to build exactly what he wanted to build -- a fast and frugal model for conflict prediction for Sub-Saharan Africa.

He tested the model on previous conflicts and got reasonably good results. The model tended to overpredict conflict in some cases but never failed to predict confict where one ultimately occurred. He was also able to rank order the potential conflicts in Sub-Saharan Africa by likelihood. Using recent data and plugging it into his model, he believes that, from most likely to least likely, Swaziland, Somalia, Equatorial Guinea, Angola, Congo (Brazzaville), Congo (Kinshasa), Zimbabwe, Rwanda, Cameroon, Cote D'Ivoire, Eritrea, Chad, Guinea and Sudan will see violent conflict (See the map below from the thesis).

Beyond the model and the predictions it makes, the literature review on other early warning systems concerning Africa and on the validity of various indicators that predict conflict is definitely worth the read. It is an excellent work and, if interested, you can download the entire thesis here.

Related Posts:
Non-State Actors In Sub-Saharan Africa
Security Sector Reform In Sub-Saharan Africa


Anonymous said...

I posted a few comments on this thesis to my blog at

maria said...


I'm trying to download the thesis but the link must be broken or something. How can I get it?