2026/06/11 by Leonardo Gentil-Fernandes
Social Sciences · #Alliance #Categorical variable #Civil Conflict #Conceptualization #Ideology #International Relations and Foreign Policy #Political Conflict and Governance #Politics #Similarity (geometry) #Terrorism, Counterterrorism, and Political Violence #Variation (astronomy)
paper · doi:10.1093/isq/sqag056
published in International Studies Quarterly 70(3) (Oxford University Press)
openalex publication_date 2026/06/11 · openalex created_date 2026/07/25 · openalex updated_date 2026/07/25
Abstract This research note distinguishes between rebel ideology and political goals to improve explanations of alliance formation during civil war. While existing studies emphasize ideological similarity as a driver of cooperation among violent nonstate actors, they rely on binary or categorical labels that obscure important variation in what rebel groups actually seek to achieve. I argue that political goals—defined as the substantive outcomes groups pursue in a post-conflict order—are multidimensional and continuous, and thus require a different measurement approach. I develop a Bayesian multidimensional ideal point model that estimates rebel group goals from expert-coded indicators of group characteristics. I then incorporate these estimates into a unified framework to predict alliance formation. The results show that goal proximity is a strong and consistent predictor of cooperation, outperforming commonly cited factors, such as shared foreign sponsorship and military capacity. This approach improves out-of-sample predictive accuracy and clarifies theoretical expectations about alignment. More broadly, it contributes to ongoing efforts to refine the measurement of latent actor preferences in conflict research.