Within a large and growing literature on insurgencies, scholars have engaged in fierce debate about the determinants of conflict outcomes. Having noted that materialcapability is a poor predictor of conflict outcomes, intense disagreement has arisen over why this is the case. Some argue that insurgencies are defeated through military and
police means of punishment and prosecution. This is referred to as the combat model. Others argue that insurgencies are ultimately defeated through political means, and I refer
to this as the social model. Why each of these two processes is thought to be more effective is rarely well explained or specified by their proponents.
Because each of these model yields different and competing expectations for the outcomes of insurgent conflicts, I evaluate their relative merits in this study. To evaluate
these two competing schools of thought in the security studies literature, I present a conditional theory of insurgent outcomes that predicts when the combat and social
models will be relevant. In order to do this, I approach insurgencies using scholarship from the study of terrorism, deriving three archetypical motivational logics of insurgency
action: strategic, organizational, and extremist. From the scholarship on insurgencies, I also develop a typology of counterinsurgency strategies focusing on three broad but
distinct strategic approaches: policing, accommodation, and reciprocal punishment. The combination of a particular insurgent motivation with a particular counterinsurgent
strategy will result in an increase or decrease in insurgent violence.
I hypothesize that this conditional model will better predict insurgent outcomes than either the combat model or the social model alone. However, even if it does not, its
evaluation will still serve as a useful comparison of the relative merits of those two models. To test the model I code statements made by leaders of insurgencies to discern
their motivations and compare these codings to the organizational dynamics and actions of the insurgency to arrive at a categorization of the group according to one of the dynamics I described. I also code statements and policies enacted by the government that make up its counterinsurgency strategy in order to arrive a coding of its strategic behavior. Based on the combination of these two codings, I predict either insurgent success or failure in that given time period. Then, I track insurgent incidents and insurgent-caused fatalities as a means of evaluating this prediction. I make predictions and evaluate their results in four case studies: The Chechen insurgency against Russia (1994-present), the PIRA insurgency against Great Britain (1969-1999), the LTTE insurgency against Sri Lanka (1976-present), and the Algerian insurgency against France
(1954-1962).
I find that the conditional model of insurgent conflict correctly predicts the change in insurgent violence in nine of 14 strategic periods, representing 75 of the total
97 years under analysis. The social model correctly predicts 10 of the 14 strategic periods, representing 70 of the 97 years. The combat model correctly predicts four of the
14 strategic periods, representing 27 of the 97 years. Overall, I found that the conditional performs as well as the social model in predicting insurgent conflict outcomes, and much better than the combat model. The results of this study indicate that scholars of insurgency and violence in general should give greater consideration to both the strategic choices of states and the motivations of the insurgents themselves.