class: center, middle .title[Bitter Friends] .subtitle[How Relationships between Violent Non-State Actors Form, Are Used, and Shape Behavior] .author[Laila A. Wahedi] .institution[Georgetown University] .date[April 19, 2017] .center[] .footnote[Created with [{Remark.js}](http://remarkjs.com/) using [{Markdown}](https://daringfireball.net/projects/markdown/) + [{MathJax}](https://www.mathjax.org/) + [{Liminal}](http://www.jmlilly.net/liminal.zip)] --- class: center, middle .title[Bloody Benefactors] .subtitle[A Model of Transregional Terrorist Sponsorship in Civil War] .center[] .footnote[ Laila A. Wahedi -- Follow along at Wahedi.US, under Current Presentation -- law98@georgetown.edu] --- # Asset Specificity:
Groups are good at different things
.left-column[ ### Local Groups * Local Infrastructure * Local Networks * Local legitimacy * Territory ] .right-column[ ### Transregional Groups * Global Infrastructure * Donor Networks * Training, intelligence * Global legitimacy ] .footnote[ Laila A. Wahedi -- Follow along at Wahedi.US, under Current Presentation -- law98@georgetown.edu] --- class: center, middle
Who Transregional Groups Want:
Ideologues
✘
Weak Groups
✘
Strong Groups
✔
.footnote[ Laila A. Wahedi -- Follow along at Wahedi.US, under Current Presentation -- law98@georgetown.edu] --- class: middle
Strong Local Groups
* Prefer autonomy * Have what they need .footnote[ Laila A. Wahedi -- Follow along at Wahedi.US, under Current Presentation -- law98@georgetown.edu] --- class: center, middle .title[Anti-Social Networks] .subtitle[The Effects of Violent Group Cooperative Network Structure on Capacity] .center[] .footnote[ Laila A. Wahedi -- Follow along at Wahedi.US, under Current Presentation -- law98@georgetown.edu] --- class: middle
Beyond Direct Partnerships
.footnote[ Laila A. Wahedi -- Follow along at Wahedi.US, under Current Presentation -- law98@georgetown.edu] --- class: Two-column layout
Effect of Local Structure
.left-column[
] .right-column[ * Contextualized Knowledge * Specialization * Leverage * Coordination * Focal Point * Security ] .footnote[ Laila A. Wahedi -- Follow along at Wahedi.US, under Current Presentation -- law98@georgetown.edu] --- class: Two-column layout
Findings
.left-column[
### Violence * More Violence * Positive but diminishing marginal returns to partnership ### Survival * Well connected groups survive longest ] .right-column[
### Violence * Less Violence * Negative diminishing marginal returns to partnership ### Survival * Well-connected groups die soonest ] .footnote[ Laila A. Wahedi -- Follow along at Wahedi.US, under Current Presentation -- law98@georgetown.edu] --- class: center, middle .title[Devils in the Details] .subtitle[Learning and Diffusion in Networks of Violent Groups] .center[] .footnote[ Laila A. Wahedi -- Follow along at Wahedi.US, under Current Presentation -- law98@georgetown.edu] --- class: middle:
Theory
## Groups must adapt to survive * Adopting new tactics helps them adapt ## Adopting and sharing tactics is risky * Sharing implies revealing secret operational procedures * Adopting risks operatives and reputation ## Partnerships help reduce the costs of sharing and adopting * Partners less likely to leak information * Helps groups learn to do it right the first time --- class: middle
Modeling Challenges
## Endogeneity * Propensity to adopt or learning? -- ## Time Polynomials: * Look at propensity to adopt at a given time * Given partner use in the recent past * Given no use previously * Straight forward interpretation -- ## Activity Periods: * Groups that don't exist cannot adopt a tactic * Need to go back in time to observe first adoption .footnote[ Laila A. Wahedi -- Follow along at Wahedi.US, under Current Presentation -- law98@georgetown.edu] --- class: middle
Multi-Level Logit
## Illicit Organizations Have Secrets * Random baseline for each group over time * Add covariates we have -- ## Different Tactical Difficulty * Randome baseline of adoption of each tactic type -- ## Interpretation * Probability of a
given group
adopting a
given tactic
in a month contingent on failure to adopt in the past, and whether a partner used that tactic. .footnote[ Laila A. Wahedi -- Follow along at Wahedi.US, under Current Presentation -- law98@georgetown.edu] --- class: middle
Evidence of Diffusion
## Groups are more likely to adopt tactics their partners have used ## Well-connected groups in centralized networks are most likely to adopt from partners ## Well-connected groups in decentralized networks are least likely to adopt from partners .footnote[ Laila A. Wahedi -- Follow along at Wahedi.US, under Current Presentation -- law98@georgetown.edu] --- class: middle
Note on Final Draft
### Graduate School asked me to remove the abstract sections from the papers