Too central to fail? Terror networks and leadership decapitation

ABSTRACT

Leadership decapitation, as a means of hindering the operations and hastening the demise of terrorist organizations, has been the subject of a growing body of research. However, these studies have not examined how an organization’s position in a broader network impacts its ability to weather decapitation. We argue that highly networked organizations possess characteristics that make decapitation less effective. To test this argument, we combine data on leadership decapitation with network data on terrorist organizations and find that well-networked organizations are resilience to leadership decapitation. Our study has implications for our understanding of how terrorist organizations respond to counterterrorism efforts.

KEYWORDS:

• Leadership decapitation
• terrorist organizations
• networks
• leadership
• counterterrorism

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website.

Notes

¹ This nuanced effect of leadership decapitation was also shown far afield from the outcome of terrorist organization mortality rate in a study of how various types of strikes led to different impacts in the Israeli and Palestinian stock market (Zussman and Zussman 2006).

² Although the argument being made here is that removing a leader should impact a group’s capacity for violence, there is another argument to be made, which is that removing a leader impacts the quality of violence (Abrahms 2018). Under this logic, one of the negative repercussions of removing a leader could be that the leader may have been a restraining force on the types and ferocity of violence employed. This argument is not directly tested here, as the focus in this article is on organizational capacity and longevity.

³ Many of these documents are maintained on the website of the Combating Terrorism Center at https://www.ctc.usma.edu/programs-resources/harmony-program. Some of the documents that illustrate the organization’s bureaucratic structure are: “AL QA`IDA CONSTITUTIONAL CHARTER,” “AL-QA`IDA’S STRUCTURE AND BYLAWS,” “AL-QA`IDA GOALS AND STRUCTURE,” and “INTERIOR ORGANIZATION.”

⁴ To be clear, al-Muhajiroun does not appear in our dataset, but we felt the example and discussion highlighted by the group’s experience we were noting in the article.

⁵ The authors also considered using Jenna Jordan’s dataset that contains 298 incidents of leadership targeting from 1945–2004, but at the time of this publication, that dataset was not publically available.

⁶ The Phillips (2014) dataset was attractive because it featured time-series data that coded network variables for terrorist organizations.

⁷ We do run and discuss Cox proportional hazards models more fully in the online appendix. Although the results are generally similar to those we discuss here in the article, the interaction term between Allies’ Ties and Leadership Decapitation is just outside of conventional bounds of significance, but because of the violation of the proportionality assumption, we cannot be confident in these results. When we remove the regional variables, which tests show violate the proportionality assumption, the interaction variable is comparable in statistical and substantive significance to what we present in the main text.

⁸ We include the graphical representation across all values of Allies’ Ties in the online appendix.

⁹ The group represented at the far-right end of the histogram shown is Figure 1 with the large value of Allies’ Ties is al-Qa’ida. We ran an additional estimation which excluded al-Qa’ida and there were no statistical or substantive changes to the main results presented here.