When Jihadist Factions Split: A Data-Driven Network Analysis

This article investigates group fragmentation in the al-Qaeda and Islamic State ecosystems, employing network analysis to examine the impact of specific network conditions on the probability of a faction splitting. Using new datasets of faction–faction (FF) and terrorist–terrorist (TT) relationships, the article tests 18 hypotheses exploring connections between factional splits and the number, polarity, and strength of FF and TT relationships, among other factors. The article offers three major findings. First, a greater number of relationships between factions is positively correlated with the probability of fragmentation. Second, having a small or moderate number of a faction’s members belonging to another faction increases the probability of a split, but more significant cross-factional membership decreases the probability. Third, both high-degree centrality of a faction’s leader and significant variations in the degree centrality within a faction’s leadership structure is correlated with increased probability of a split.