View Full Version : Terrorist Target Selection -- Quantitative Studies?

01-16-2008, 08:44 PM
Hi all, I'm new to this very fascinating group. In a grad school course I am taking, I am to replicate a quantitative study that uses maximum likelihood methods (logistic regression, event count models, multi-equation models, or time series). Anybody familiar with a good study using these methods?

01-16-2008, 09:50 PM
Have you looked at Robert Pape's work?
"The Strategic Logic of Suicide Terrorism (http://www.danieldrezner.com/research/guest/Pape1.pdf)," American Political Science Review 97.3 (August 2003) p.343-361.

01-16-2008, 10:12 PM
I've been thinking we need to recreate one using the terrorist database. All you need are two or three attributes (time, location, type, organization if any claiming) and you could do all kinds of quantatitaive statistical analysis. You're going to be in the realm of multi-variate but with SAS, SPSS, etc.. that isn't to bad.

Ken White
01-16-2008, 10:15 PM
Why not go here and tell a little about yourself...

LINK (http://council.smallwarsjournal.com/showthread.php?t=1441&page=22).

01-16-2008, 10:56 PM
Try the RAND studies Reducing Terrorism Risk at Shopping Centers: An Analysis of Various Security Options (http://www.rand.org/pubs/technical_reports/2006/RAND_TR401.pdf).

....Our analysis is built largely around a modeling approach in which the estimated effectiveness of specific security options in reducing the risk of particular terrorist attack scenarios is weighed against the costs of implementing those options. The basic elements of the model are a set of attack scenarios, estimates of the relative likelihoods and the consequences of each scenario, a set of potential security options, the cost of each option, and the likely effectiveness of each option in each scenario....
....and Terrorism Risk Modeling for Intelligence Analysis and Infrastructure Protection (http://www.rand.org/pubs/technical_reports/2007/RAND_TR386.pdf).

...Since 2001, companies in the insurance industry have begun using terrorism risk models to help understand and manage their exposure to terrorism losses in different markets. As part of DHS’s learning process, OI&A asked the RAND Corporation to explore how these tools might be useful for its Homeland Infrastructure Threat and Risk Analysis Center (HITRAC). HITRAC provides intelligence analysis about terrorism threats in the United States and risk assessments to help protect critical infrastructure.

One such insurance-industry model is the Probabilistic Terrorism Model developed by RMS (RMS, undated). Founded at Stanford University in 1988, RMS provides the insurance and reinsurance industries with products and services for quantifying and managing catastrophe risks. The RMS model estimates the risk of macroterrorism, which RMS defines as attacks capable of causing (1) more than $1 billion in economic losses, (2) more than 100 fatalities or 500 injuries, or (3) massively symbolic damage. Starting with specific attack scenarios, the model assesses the threat of various types of attack on different targets, the vulnerability of those targets to those attacks, and the expected annual consequences of successful attacks in terms of casualties and property loss. The overall risk of any given attack scenario reflects all three of these factors....

Little Doughnut
01-28-2011, 12:35 AM
Did you ever succeed in replicating your study?

The Global Terrorism Database (http://www.start.umd.edu/gtd/) at U of Maryland might have been a good place to start collecting data.

I believe Jane's Terrorism and Insurgency Center also publishes data on target types. A subscription costs an arm and both legs though. Good news, they let you pick which arm you want to keep. Don't know if your university had access or not.

What study did you replicate?

I'm attempting to do a quant model showing that insurgencies quelled through power sharing agreements are more likely to become violent again than those solved by vanquishing the insurgents....

01-28-2011, 02:15 AM
from LD
I'm attempting to do a quant model showing that insurgencies quelled through power sharing agreements are more likely to become violent again than those solved by vanquishing the insurgents....

Not that I'm opposed to taking a position and then justifying it - that is my real world.

Looks like your model (if it shows what you posit) will be more along the Wilf (http://council.smallwarsjournal.com/member.php?u=1814) model - as opposed to the Bob's World (http://council.smallwarsjournal.com/member.php?u=2880) model.

In any event, welcome to SWC.



Little Doughnut
01-28-2011, 06:29 PM
We'll see if I can actually prove it.

As far as I can ascertain there is a high rate of recidivism when rebels are conscripted into an existing government rather than winning outright or being defeated outright. I'm still in the exploratory stage but will be happy to post up what I find once I'm finished.

Thanks for the welcome, one can only hang around for so long before getting involved.

01-28-2011, 11:03 PM
you'll find some "involvement" with this topic. ;)

Sounds like you are into some of the issues left unresolved in 2010 RAND, Ben Connable and Martin C. Libicki, How Insurgencies End (http://www.rand.org/pubs/monographs/MG965.html) - esp. with respect to conflicts arising from their Type III: Mixed (Stalemate/Negotiated Settlement) and Type IV: Inconclusive or Ongoing Outcome.



01-31-2011, 05:18 PM
You may or may not find this of interest - slightly different perspective:

McCormick, G.H. “Terrorist decision making.” Political Science 6, no. 1 (2003): 473. http://www.annualreviews.org/doi/full/10.1146/annurev.polisci.6.121901.085601?select23=Choose.