Whenever we think about terrorism, destruction and disarray comes to mind. Attacks are surprising and are typically well-orchestrated. There’s no way that we could predict their attacks, right? Well, University of Warwick’s Dr. Weisi Guo put this claim to the test.
Data and Peacekeeping
In his paper, Guo detailed how current statistical analyses involving terrorist behaviour are focused on the large scale — looking at national data and historical records spanning a few centuries. According to him, tracking factors like the attack frequencies and time duration between attacks in this manner is crucial. This serves as a metric of the effectiveness of peacekeeping efforts in the side of the government.
For instance, a decreasing trend in attack frequency would suggest that negotiation efforts are effective. An opposite trend, on the other hand, serves as an indication that some adjustments might be needed to be done. Data allows us to make decisions that are evidence-driven.
Cities and Terrorism
While the insights provided by the current paradigm of analysis is helpful, Guo wanted to know what other insights the data can bring. He also wanted to determine if the data on hand can be harnessed to create predictive models of the attacks.
Such insights and models would prove to be an asset, especially for cities. Right now, the way terrorist attack data are analyzed misses the fact that cities take the brunt of these attacks. Majority of cities in the world are centres of government and economic activities of a country. They are ideal targets if you want to sow fear and chaos into society. If it is possible to develop a prediction system, cities would be empowered with a strengthened capacity for counter-terrorism.
Using the terrorism and conflict data from the Global Terrorism Database which records the locations of conflict incidents and their corresponding death tolls in, Guo studied the behaviour of the attacks.
In the study, he found out that the number of attacks received by a city isn’t correlated with the population size. In other words, a decrease or increase in population won’t influence the attacks received by a city. Furthermore, the size of the city wouldn’t translate to a lower or higher number of attacks, either.
Similarly, he found out that the death toll isn’t correlated with the population size either, which might seem counterintuitive. This is because one might be led to think that larger populations are expected to receive higher death tolls.
This means that cities of all sizes should be well-guarded against terrorist attacks.
Predictability in Chaos?
For the highlight of the study, Guo developed a predictive model guided by the data. The model is developed so that provided the number of attacks the city suffered along with some key assumptions on the behaviour of these attacks, the next attack could be predicted.
The model also monitors the information contained by the terrorist attacks. A higher amount of information contained translates to higher uncertainty of the attacks. A higher uncertainty, in turn, translates to a higher difficulty in accurately predicting the next attack.
It turns out that while the model Guo developed seemed to have soundly captured the behaviour of the attacks in a statistical sense, much information is lost when the model is used for prediction. This means that as of yet, the model isn’t a reliable way to predict attacks. Further research is necessary.
What’s more, upon analysis of the data which spans from 2002 to 2014, Guo observed that the uncertainty of the attacks is increasing. Needless to say, this is bad news.
The challenge of creating a predictive model for terrorist behaviour remains. While we aren’t successful yet, we are now aware that it is possible to find regularities in the seemingly unpredictable terrorism process. This should prod other researchers to begin exploring terrorist behaviour in our cities.