Doctor of Philosophy / Security Studies Interdepartmental Program / Emizet N. Kisangani / This study examines the effects of regime type on support to foreign insurgent groups. Theoretically, it relies on structural and normative characteristics of democracies by arguing that leaders in these regimes tend to encounter multiple disincentive mechanisms generally not found in non-democracies. Thus, leaders of democratic regimes are less likely to actively support foreign insurgent groups as a component of strategy below the threshold of military intervention. When they do choose to lend their support, they tend to choose either low-level types of support (simple material support) or high-level support (full military intervention). Leaders of non-democratic regimes, however, can employ the full spectrum of support types to seize strategic opportunities and tailor strategies that are more costly and more risky. The dissertation tests this theory by using both quantitative and qualitative research methods. The statistical analysis of a dyadic, cross-sectional, time-series dataset of 179 countries from 1975 to 2009 provides some support for the proposed hypotheses. Structured, focused comparison of three conflicts with multiple within-case observations (cases) also reveals modest support for the hypothesis that democracies are unlikely to support insurgent groups in general and have multiple disincentives toward providing mid-level types of support that expose the democratic leaders to additional costs and risks. Unexpectedly, the qualitative case studies reveal that in addition to the structural disincentives initially identified, leaders of democratic regimes may have a harder time managing the principal-agent relationship between the supporting state (principal) and the insurgent groups (agents). The need to maintain a large winning coalition to survive as a leader in a democracy presents multiple principal-agent problems and allows rebel leaders and rebel factions to resist integration, prevent the loss of autonomy, and facilitate the establishment of alternative avenues of resource mobilization. While previous literature in political science and international relations provides evidence that structural characteristics of democratic regimes make them good at winning interstate wars, this study provides initial evidence that those same structural characteristics make democracies’ success more elusive when applying unconventional strategies short of war.

Limited systematic research has investigated how conflict events shape the spatial-temporal variation of insurgent violence against civilians. Although previous research has investigated how degrees of territorial control relate to general levels of violence against civilians, it remains largely an open question how the dynamics within territorial control determine violence against civilians by insurgents. This study aims to address this gap by hypothesizing that (1) insurgents become more likely to commit fatal violence against civilians, and (2) kill more civilians in contested areas when they lose territorial control. The Armed Conflict Location and Event Dataset (ACLED) was used along with Peace Research Institute Oslo’s (PRIO) GRID Dataset to create a novel data frame of all territorially contested area-weeks on the African continent between 1997 and 2017 (n = 3035). Contrary to theoretical expectations, logistic regressions indicate a lower risk of insurgent violence against civilians in contested areas following an insurgent territorial loss than following a break-even. Zero-inflated negative binomial regressions moreover tentatively indicate that insurgents kill more civilians following territorial wins in the short-term, and following territorial loss in the long-term. These results suggest that proactive counterinsurgency campaigns are in the interest of civilians in civil war.