Collier is talking about civil wars, but what about different kinds of conflicts like more localized land disputes? In fact, a very interesting book on land conflict that I'm currently reading argues for an almost polar opposite perspective. Land conflicts defy generalization, the authors claim, and can only be understood through the idiosyncracies of the particular context. The most general statement they allow is that “when multiple forms of inequality coincide, and when these in turn are aligned along divisions of identity, then the probability of violent conflict increases.” And the particular forms of inequality and identity that are relevant will vary from case to case. I think it's worth noting that this is a book by anthropologists, and to some extent that perspective serves to justify the discipline of anthropology- after all, if we can't generalize, the only people who can tell us what's really going on are going to be anthropologists.
I'm not sure if the anthropologists are right, but this suggests interesting question that I hope to look at in future research- how accurately can we predict the incidence of conflict in a particular area given data on the characteristics (resources, economic activity, ethnic makeup, etc.) of that area? If the answer is, "reasonably well," then the anthropologists are wrong- the best strategy to deal with land conflicts should be based on looking at the factors that tend to give rise to conflict, rather than detailed contextual nuances. On the other hand, if the answer is, "not very well," then we'll have to send in the anthropologists.