In the latest issue of Criminology, Torbjorn Skardhamar addresses some potentially significant issues with latent trajectory models that should have a direct impact on how terrorism studies assesses group behavior. Specifically, Skardhamar makes the argument that latent trajectories can and will be found in any data with multiple heterogeneous patterns contained therein. He tests this hypothesis using a simulation methodology that shows latent groups being identified within data containing known groups but also identifying latent groups in a randomly generated dataset, though there is no reason to believe any exist.
This should have a significant impact on the conduct of future terrorism research, and the interpretation of current research, particularly regarding the work done by researchers at the START Center at the University of Maryland, which used latent trajectory modeling to identify four attack trajectories within START’s open source terrorism database and examine trajectories across nations (Dugan, LaFree, and Miller, 2007; LaFree, Morris & Dugan, 2009; LaFree, Yang & Crenshaw, 2009). In general, the results are still quite valid, however, they must be interpreted with a great deal of caution as the trajectories identified cannot be taken as tests of any theory, or proof of their empirical existence.
The general problem identified is one that has, and seemingly will continue, to plague terrorism studies: We don’t have well specified theory. In my research I am as guilty as the next researcher, as I’m avowedly interested in what the data tells us, not what theory tells us the data should tell us. I rationalize it by reminding people that we still don’t know that much about terrorism, and so perhaps we’re not ready for advanced theory yet. However one looks at it, it is clear that even with our ever increasing statistical sophistication, we must be cautious about taking our findings as reality. Skardhamar’s article, really, is no more than a reminder of this, but it is one that should be well taken by terrorism studies.
Skardhamar, T. (2010). Distinguising facts and artifacts in group-based modeling. Criminology 48, 295-320.
Dugan, L., LaFree, G. & Miller, E., Cragin, K. (2008). Group-based trajectories of frequency and severity of attacks by terrorist organizations, 1970 to 1997. Paper presented at the American Society of Criminology Annual Meeting, St. Louis, MO.
LaFree, G. Morris, N., & Dugan, L. (2009). Comparing trajectories for total, attributed and fatal attacks, 1970-2006. The British Journal of Criminology, 1-28.
LaFree, G. Yang, S. & Crenshaw, M. (2009). Trajectories of terrorism. Criminology and Public Policy 8, 445-473.
Copyright © 2018 The Institute for the Study of Violent Groups.