A Department of Homeland Security Center of Excellence led by the University of Maryland

A consortium of researchers dedicated to improving the understanding of the human causes and consequences of terrorism

Breiger turns regression models 'inside out'


Breiger turns regression models 'inside out'

START Researcher evaluates CBRN activity

August 26, 2013Emily Smith

During a recent START Research Roundtable, Dr. Ronald Breiger explained how he and a team of START researchers are developing innovative analytic methods to improve situational awareness of Chemical Biological Radiological and Nuclear (CBRN) activities.

By using quantitative social science methods, Breiger and his team are able to analyze hundreds of organizations and events to determine the relative salience of potential causal factors by imputing likely behavioral trends. That is, they translate statistics into usable qualitative data about terrorism.

Breiger, a sociologist and professor at the University of Arizona, is working with START researchers Gary Ackerman, Brinton Milward, Victor Asal, Karl Rethemeyer and Lauren Pinson to draw data from START's database "Profiles of Incidents Involving CBRN by Non-State Actors" (POICN), in which the data describing incidents are coded with respect to degrees of credibility and consistency. Breiger restricted the data to that with the highest credibility.

In his talk, "New Analytic Methods for the Exploitation of Open-Source Structured Databases on the Pursuit of WMD Terrorism," Breiger discussed how he does not use standard statistical regression modeling in this research, but rather a reverse methodology.

"There is a specificity with respect to CBRN actors and events in our models that is difficult to find in standard regression models," Breiger said of his methodology. "The impetus of this work from a methodological point of view is to turn around the problem."

From their data, the researchers are able to conclude several tendencies of CBRN terrorists, and hope to implement their findings in future research. For example, a standard regression model suggests that the variable of "religious extremism" is negatively associated with possessing (rather than pursuing) CBRN agents.

Breiger finds that, while this is true in general, there is a notable cluster of events (centered in Southeast Asia, involving chemicals, transpiring in the most recent time period covered by the database,2007-2011) for which the opposite result holds: these events were more likely to involve religious extremists who possessed agents. In this way, the overall tendency?which the standard regression model has correctly found?misses a key set of cases that require a distinctive theory to account for their CBRN activities.

"Neither theorists nor practitioners like to be surprised by the less predominant, but nonetheless patterned, features of the human terrain," Breiger said, and suggested that the methods he and his colleagues are developing allow for a more complete and more specific situational awareness than is usually discovered by use of standard methods.

"These results, and others like them, are windows into situational awareness of terrorist use and pursuit of CBRN weapons. We see our research as opening some new windows into this subject matter, and to advancing our ability to predict outcomes," Breiger said. "In ways that we have just begun to tap, the POICN database will have a huge impact on future scientific analysis of CBRN as an arena for terrorism studies."

The project Breiger spoke about is funded by the Defense Threat Reduction Agency (DTRA) and supports both the enhancing of existing databases and the leveraging of those databases to extrapolate information on CBRN activities.

To read Breiger's START Researcher Spotlight, click here. To learn more about the project, click here.