During a Research Roundtable at START this month, Global Terrorism Database (GTD) Project Manager Erin Miller detailed the patterns of terrorism worldwide using GTD data updated through 2011. With the new data, which was released in October, the GTD now includes comprehensive data on more than 104,000 domestic and international terrorist attacks between 1970 and 2011.
Miller and her team found that attacks in Iraq, India, Pakistan, Afghanistan and Russia accounted for 70 percent of terrorist attacks worldwide in 2011. Iraq, which had the most terrorist attacks in 2011 (1,300), also had the most over time from 1970 to 2011 (7,807). From 1970-2011, fatalities from terrorist attacks in Asia accounted for nearly 30 percent of the fatalities from terrorist attacks worldwide; fatalities in the Middle East and North Africa accounted for nearly 25 percent.
The new GTD data shows the prevalence of activity among al-Qaida-linked groups in 2011. Such groups are responsible for four of the five most lethal attacks in 2011:
- al-Qaida in the Arabian Peninsula (AQAP) Yemen: March 28?110 killed, 45 injured
- Tehrik-i-Taliban Pakistan (TTP) Pakistan: May 13?80 killed, 140 injured
- al-Shabaab: Somalia: Oct. 470 killed, 42 injured
- al-Qaida in Iraq?Iraq: March 2965 killed, 95 injured
Only 10 terrorist attacks occurred in the United States in 2011, accounting for less than 0.2 percent of terrorist attacks worldwide.
Moving data collection forward The recently released 2011 data includes several new variables and supplemental work on perpetrator groups and weapon types. The 2011 data also included decades of geocoded data for eight different regions across the world. This geocoded data allows researchers to track geographic patterns of attacks for key conflicts. GTD researchers are also expanding and refining data collection efforts going forward. Beginning in 2012, all GTD data is being collected in-house at START headquarters using newly developed tools, including machine-learning technology.
Developed exclusively for the GTD, the machine-learning model allows START researchers to more efficiently determine which open-source news items are relevant to terrorist attacks and the GTD. Along with tools that help eliminate repetitive source articles, this new model streamlines data collection. The process involves categorizing approximately 10,000 articles per week according to their likely relevance, helping GTD coders find the 10?12 percent of those articles germane to terrorist attacks more quickly.
Miller said the new data collection tools could ultimately be used for other data collection efforts as well.