[log in to unmask]" type="cite">Dissertation Defense Announcement
To:  The George Mason University Community


Candidate: David L. Bolduc
Program:    PhD Biodefense

Date:   Tuesday April 26, 2011
Time:   1:30 p.m. 
Place:  George Mason University
 	     Research I Bldg., Room 161
	     Fairfax campus
  
Dissertation Chair: Dr. Robert L. Dudley
Committee members: Dr. Charles L. Bailey, Dr. Joseph A. Marr
Title: "Development of an Algorithm for Predicting 'Relative Risk'
            of Terrorist-CBRN"

The dissertation is on reserve in the Johnson Center Library, Fairfax campus.
The doctoral project will not be read at the meeting, but should be read in advance. 
All members of the George Mason University community are invited to attend.


ABSTRACT: 
In the past decade there has been a tremendous rise in the number of extremist groups with radical goals.  To achieve their objectives, some of these groups 
are resorting to terrorist tactics.  With the rapid expansion of scientific technology brought on through increasing globalization, the prospect of incorporating the
use of chemical, biological, radiological and nuclear (CBRN) weapons into their tactics is becoming increasingly feasible and likely.  As these radical groups
become more energized, they may resort to using CBRN.  In order to avert a disaster from a terrorist-CBRN (T-CBRN) event, it is important to study
the likelihood of these groups employing CBRN.  Analysts continue to search for sound methodologies in assessing the risk of such groups pursuing CBRN.
Psychological and mathematical modeling attempts have been made in discerning terrorist’s propensities for CBRN Asal (2010), Sullivan and Perry (2004), 
Post (2002), Tucker (2000) etc., but there still remains considerable gaps in these areas.  
  

 

Because of the deficiencies in analytical methodologies and controversial techniques used for modeling T-CBRN, we proposed the questions: can a more accurate predictive model of T-CBRN use be developed?  And, when compared against past event data, would such a model reliably predict the amount of “Relative Risk” of a terrorist that would elect to use CBRN?

 

 This study entailed the development of an algorithm linked with statistical software (ProStat Version 5.5, Poly Software International) for predicting the Relative Risk of a terrorist seeking CBRN.  This was achieved through four phases.  Phase I involved searching open literature and proposing independent-variables associated with T-CBRN.  Phase II entailed the construction of a “Random-Nations Matrix” representing the T-CBRN universe.  This matrix was then used for correlating variables suspected of being associated with T-CBRN.  Phase III involved the construction of a multivariate model from the variables which met our correlation criteria with T-CBRN.  The final phase, Phase IV, entailed the construction of an algorithm derived from the model design, for predicting the amount of Relative Risk of a terrorist seeking, acquiring and or using CBRN, and the Relative Risk of T-CBRN occurring in a specific country.   Probability was measured by the strength of the relationship indicated by the p-value.

 

 The study drew primarily from the following databases: the Memorial Institute for the Prevention of Terrorism (2007), the U.S. Department of State (2010), Globalsecurity.org (2010), the Global Terrorism Database (2010), the Central Intelligence Agency World Fact Book (2009-2010), the James Martin Center for Nonproliferation Studies/Monterey WMD Terrorism Database (2010), the Center for Defense Information (2009), the Freedom House Center for Systemic Peace (2009) and the McDonald’s Corporation’s list of worldwide McDonald’s restaurants (2010).

 

A Random Nations Matrix was developed for determining the two independent-variables that correlated with our dependent-variable “Terrorist-CBRN”.  This matrix was constructed from the random selection of 74 countries and/or locations from the CIA World Fact Book (2009-2010).  Thirty-nine of the same variables were measured for each of the seventy-four selected countries or locations compiled in the matrix.  The 39 variables selected for each country or location, were correlated with each other and the T-CBRN variable for determining correlative significance.  A multivariate-model was developed with 67.3 percent predictive power for T-CBRN using the independent-variables: “Political Violence Per Capita” and “Number of Terrorist-CBRN Technical Experts”.  Average lag-times between terrorist’s interest and use of CBRN were also calculated for the various classifications of terrorist-groups.

  
[log in to unmask]" type="cite">