(iPoster) Modeling Rebel Network Evolution: Power and Identity in Civil Wars

• In Event: Poster Sessions: Information Technology and Politics & Political Networks
  In Poster Session: Poster Session: Political Networks

Abstract

This study develops a computational network-science framework to model the relational dynamics of non-state armed groups, specifically examining how identity-based configurations (Balcells, Chen, and Pischedda 2022; Bapat and Bond 2012; Corradi 2023; Topal 2024) and power asymmetries (Christia 2012; Krause 2014) drive alliance formation. Utilizing longitudinal data from the Syrian conflict, I construct a multi-layered network of rebel factions by integrating datasets on formal alliances and defections (Gade, Hafez, and Gabbay 2019) alongside battlefield engagements (Sundberg and Melander 2013). The research implements a dual-track statistical approach designed to isolate the drivers of network evolution: Additive and Multiplicative Random Effects (AME) models are utilized to account for latent sender-receiver effects and reciprocal patterns, while Exponential Random Graph Models (ERGMs) assess the structural dependencies of the network, such as transitivity and popularity. By formalizing identity as a measure of ideological homophily and power as a function of material capabilities and nodal centrality, the models quantify the relative impact of these variables on the probability of cooperation and conflict.
This methodological framework is refined through a subnational comparative design of Hay’at Tahrir al-Sham (HTS), ISIS, and the Democratic Union Party (PYD), which allows for a controlled analysis of how specific network configurations evolve under varying power and identity constraints.

Author

• Betul Ozturan, Boston College