*Names in bold indicate Presenter
While the Obama Administration has ratcheted down the rhetorical commitment to a “global war on terrorism,” the US is still engaged in suppression of terrorism across the globe. The reality of limited resources makes it imperative to focus national assets on those groups with the most potential for destruction and death. In previous work (Park, Rethemeyer, Asal, and Ackerman forthcoming) we used a cross-sectional dataset containing data on 395 terrorist organizations to examine the factors that lead some organizations to ally. While this work provided important insights, the generality of our results was constrained by our inability to properly examine causation with longitudinal data, the limited organizational coverage of our first effort, and the paucity of organizational variables available in our first dataset. This paper addresses the limits of our first effort with a new dataset and new methods.
The paper proposed here examines organizational alliances using the Big Allied and Dangerous Dataset, Version 2 (BAAD2). BAAD2 is a panel dataset that includes 580 terrorist and insurgent organizations across 10 time periods (1998-2007). The dataset is unique in that it includes data on both compositional variables (such size, age, ideology, leadership structure, and sources material support) and network variables (including alliances between organizations and organizational relationships with states). Compiled through extensive hand-coding of tens of thousands of print sources, BAAD2 provides organizations data on nearly 90% of terrorist organizations that were recorded to have perpetrated at least one attack in the Global Terrorism Database (GTD) and virtually all insurgent organizations recoded as having perpetrated at least 25 battle death in the Uppsala Conflict Data Program (UCDP) dataset. Collectively, these organizations are responsible for more than 20,000 deaths and nearly 5,200 terrorist incidents. We will model this data using recently developed stochastic methods for modeling longitudinal social network data. Our model focuses on organizational factors (age, size, funding structure, participation in terrorist activity, ideology), country context (wealth, regime type, military size), geography (preference for partners that are near), and fundamental social processes in covert networks (preference for reciprocity and closure to foster trust).
The paper concludes with an analysis of the policy and strategy implications of our empirical analysis. Our previous work suggests that interrupting the process of organizational alliance formation seems quite important because alliances correlate with bad behavior.
- Devils_FINAL.pdf (2553.2KB)