12.07.2015 Views

Sunbelt XXXI International Network for Social Network ... - INSNA

Sunbelt XXXI International Network for Social Network ... - INSNA

Sunbelt XXXI International Network for Social Network ... - INSNA

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Two‐mode Projection And Data LossEverett, M G.; Borgatti, S P.Mathematical and Statistical <strong>Network</strong> ModelsTwo‐mode <strong>Network</strong>s, Data, ProjectionFRI.PM2The standard projections take a two‐mode binary matrix A and construct AAT and/or ATA and then analyze these. If our matrix A is an actor by event matrixthen the <strong>for</strong>mer is an actor by actor matrix in which the entries are the number of events pairs of actors attended together, and the latter is an event by eventmatrix of the number of actors common to both events. The projections are actually similarity matrices derived from the rows and columns of the data matrixA. In the binary case these are counts of the number of times the rows (or columns) have a one in common. One of the criticisms of using projections is thatthere is a loss of structural in<strong>for</strong>mation and it is true that using either AAT or ATA alone does lose structural in<strong>for</strong>mation. However, it is not clear how muchin<strong>for</strong>mation is actually still embedded in the projections and to what extent data is actually lost. A closer examination of this issue suggests that relatively littlein<strong>for</strong>mation is lost and even less is lost if we consider both projections together. This suggests a different approach to analyzing two‐mode data namely toanalyze both projections and combine the solutions.Unconventional Weapons And Drug Smuggling: A Dual‐network Configurational Analysis Of Terrorist OrganizationsSchoon, Eric; Asal, Victor; Breiger, Ronald; Melamed, David; Milward, H. B.; Rethemeyer, R. K.Criminals, Gangs, Terrorists, and <strong>Network</strong>sOrganizations, QCA, Covert, Dark <strong>Network</strong>sSAT.AM1Terrorist organizations are violent and often motivated politically—but this does not make them drug smugglers. Drug smuggling is an activity that many see asimmoral and problematic <strong>for</strong> organizations trying to claim moral superiority over their opponents and who are likely to see themselves as “the good guys.”Indeed, most terrorist organizations do not engage in drug smuggling. In this paper we focus on chemical, biological, radiological , or nuclear weapons (CBRN)acquisition or use, drug smuggling, and terrorist action. We examine how these activities intersect across 395 terrorist groups. To begin, we assess how keyorganizational attributes influence the probability of these organizations engaging in drug smuggling using logistic regression. However, in this paper we argue<strong>for</strong> moving beyond these conventional analytic methods, by linking actors and attribute data to two‐mode network analysis, configurational methods (as inCharles Ragin’s QCA framework), and barycentric correspondence analysis (as recently proposed by Ronald Breiger). Using our methods we provide a visualmap of the intersection of all the variables, and simultaneously all the actors in our study, while identifying key conjunctures of variables that enable us topredict drug smuggling. A key finding is that, while the (degree) centrality of terrorist organizations within a network of inter‐organizational alliances clearly hasan impact on their involvement in drug smuggling, this connectivity must be combined with other variables, in the multiple paths that we identify that are mostlikely to lead to drug smuggling.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!