Pathways to civil war: A study of multiple paths toward civil war.

摘要

This dissertation is about conflict escalation to civil war, and examines why some political confrontations escalate and why principal conflict actors continue fighting rather than reaching a number of political arrangements at various points of the course of conflict. Unlike previous studies, this study treats the progression to civil war as one of complex alternate paths. In so doing, building on the perspective of asymmetric information (i.e. uncertainty) problems as a cause of war, this study claims that involving each conflict actor's cognitive variances about its opponent's willingness to resolve and military strength would bolster either side's costly military mobilization and boil over into civil war. Four extant hypotheses on conflict escalation and two specific propositions from a two-sided uncertainty perspective are tested with ordered and binary multiple logistic regression analyses against state-year aggregated data on government repression and armed resistance levels as well as civil war onset from 1976 to 2000. A comparative case illustration of the Lebanese civil war of 1975--1976 and the Northern Ireland conflict of 1970--1998 further illuminates the internal conflict dynamics toward or away from civil war, examining the emergence of principal and secondary armed actors in the course of conflict. Both the quantitative and qualitative studies provide evidence for the roles of uncertainty in either government leaders' or armed rebel leaders' decisions to fight or make certain concessions, while demonstrating that structural, institutional, demographic, and insurgent-favorable factors help explain the causes and persistence of 'initial' communal violence, armed resistance, and government repression. The study concludes with substantive policy implications for preventing conflict escalation and calls for stepping up efforts at establishing actor-based theoretical underpinnings to understand civil war as multi-interdependent reciprocal processes.