Summary: Kalyvas is looking, not at civil war, but at the incidence of violence against ‘‘non-combatants’’ within the context of civil war. He argues that this violence has its own logic which emerges from the interaction between the warring sides and the civilian population. Kalyvas contends that neither insurgents nor incumbents can rely on civilian support, so in order to secure it, some kind of violence is ultimately necessary. For Kalyvas, ‘‘indiscriminate violence’’ is counter-productive, as there is equal incentive to co-operate or not, thus over time political actors learn to replace indiscriminate with selective violence, targeting particular named individuals. Kalyvas articulates three different contexts (or zones, as he calls them). In the first, one side or the other has complete control, the result being full compliance and cooperation by the civilian population; denunciations of other individuals are implausible, so in these situations, selective violence is low. In the second, one side dominates the other, but there is contestation. In such an environment, selective violence should be high as there is the opportunity for individuals to denounce others who are suspected of helping the enemy. In the final zone, where both sides exercise partial control (i.e. the state rules in the day, but the insurgents rule at night), selective violence should be counter-intuitively low. The reason is that while each group wants to kill prominent supporters of the opposition, the civilians will not supply this information because there is fear of counter-denunciation.
Important Insight: Selective violence against non-combatants is a product of the balance of power between incumbents and insurgents. Only when one group dominates but the other poses a threat should we expect to see high incidences of selective violence. In all other cases, the motivation or the opportunity to denounce others is absent.
Methodology: Tests his theory with a case study of Greece in the 1940s