The Terrorism and Extremist Violence in the United States (TEVUS) Database

Project Details


The Terrorism and Extremist Violence in the United States (TEVUS) Database integrates four open-source data sets to facilitate more robust and sophisticated analyses of the behaviors, operations, and activities of violent extremists within the United States. The overall goals of the project are to:

  • Collect and validate data related to terrorism and extremist violence in the United States;
  • Merge data from different data sets and consolidate this information into a functional, accessible, and relational database; and
  • Analyze these data to provide insights into the dynamics of terrorist activity and extremist crime in the United States.

During the past few years, this project has greatly enhanced the scope and quality of empirical data by expanding the U.S. Extremist Crime Database (ECDB), the American Terrorism Study (ATS), and the U.S. component of the Global Terrorism Database (GTD), as well as developing the Profiles of Perpetrators of Terrorism in the United States (PPT-US) data set.  These four data sets each have unique features:

  • The American Terrorism Study (ATS) is led by researchers at the Terrorism Research Center at the University of Arkansas. The ATS is an empirical relational database consisting of data on federal terrorism-related court cases, persons indicted in these court cases, and related officially designated terrorism incidents. Included in the TEVUS portal are data from court case, person, organization, affiliation, incident, and precursor activity (antecedent) tables in the ATS. Variables included cover demographic information, terrorist group to which the individual belongs, and temporal and geospatial data on incidents and antecedent activities.  
  • The Global Terrorism Database (GTD) is led by START researchers at the University of Maryland. The GTD is an open-source database that includes information on terrorist attacks around the world from 1970 through 2018. It is comprised of systematic data on domestic as well as international terrorist attacks that occurred during this time period and includes more than 190,000 cases. Only attacks that occurred in the United States are included in the TEVUS portal. 
  • The U.S. Extremist Crime Database (ECDB) is led by researchers at John Jay College of Criminal Justice, Michigan State University, Seattle University and Indiana University – Purdue University, Indianapolis. The ECDB is a relational database that includes information on all publicly known violent and financial crimes committed in the United States by extremists associated with al-Qa’ida and its associated movement (AQAM) - which for the purpose of this dataset also include crimes committed by extremists associated with the Islamic State of Iraq and the Levant (ISIL), the violent Far Right (FR), and the Animal and Earth Liberation Fronts (ELF and ALF). The ECDB includes information on the incidents themselves, as well as their perpetrators, related organizations, and victims. It currently covers the period between 1990 and 2018.
  • Profiles of Perpetrators of Terrorism in the United States (PPT-US) is led by START researchers at the University of Maryland. PPT-US is a group-level dataset, including information on the background, ideology, structure, goals, and activities of groups and organizations identified as perpetrators of attacks in the Global Terrorism Database (GTD). Only GTD perpetrator groups for which there is high confidence of responsibility for at least one violent attack are included in PPT-US. There are over 140 groups included in the dataset that carried out terrorist attacks in the US between 1970 and 2018. 

These four open source data collections have greatly enhanced the capacity of the research community to provide useful and reliable information to homeland security operators, analysts, policymakers, and the academic community.  By integrating these data, along with information on the characteristics of locations that do and do not experience terrorist activity, and developing a user-friendly web interface to access them, this project will allow a range of end-users to conduct analyses using what is by far the most comprehensive open source database on terrorism and extremist violence in United States.

Primary Findings:

This project includes several subprojects:

  • For Bombing and Arson Attacks by the Earth Liberation Front (ELF) and Animal Liberation Front (ALF), Steven M. Chermak, Joshua D. Freilich, Celinet Duran, and William S. Parkin analyzed data from the U.S. Extremist Crime Database from 1995 to 2010. During this time, 239 arson and bombing attacks were committed by ELF and ALF, with just over half attributed to ELF. Some offenders had ties to only ALF or ELF, but most were connected to both groups. One or more arrest was made in only about a third of all incidents. Many of these perpetrators were convicted of multiple crimes, and only 39% of incidents were not related to other incidents, demonstrating that a relatively small group of individuals can be responsible for a large number of offenses. Deliverables related to this subproject include:
  • In Characteristics of American Communities Where Terrorists Lived, Planned, and Conducted Their Attacks, Brent Smith, Kevin Fitzpatrick, Paxton Roberts, and Kelly Damphousse focused on identifying whether there were differences in the socioeconomic, housing, and sociodemographic characteristics of communities that were associated with terrorist residential and pre-incident activities compared with communities that were not. They found that census tracts in which perpetrators lived or conducted pre-incident activities were generally characterized by lower socioeconomic status, poorer housing conditions, and sociodemographic characteristics that were significantly different than tracts without residential or pre-incident activity. At the same time, this overall pattern varied by type of terrorist group. For example, the census tracts in which environmental and far-right perpetrators lived and conducted pre-incident activity tended to have lower percentages of foreign-born residents compared with those in which they did not, while the opposite was the case for international perpetrators. Deliverables related to this subproject include:
  • Community-level Indicators of Radicalization: A Data and Measurement Workshop, led by Shira Fishman, brought together subject matter experts on individual radicalization, the study of communities, and data measurement issues to discuss how analyses of archival or institutional data at the community level might provide new insights into violent radicalization. Participants identified a range of measures that could potentially be associated with the number of terrorist attacks or ideologically motivated crimes committed in a community, including measures related to economic distress, social capital, political inclusion/exclusion, social support, demographics, and ideology, and offered recommendations for conducting comparative analyses. Deliverables related to this subproject include:
  • In County-level Correlates of Terrorism in the United States, Gary LaFree and Bianca Bersani built on Hot Spots of Terrorism and Other Crimes and expanded their analysis to cover 1990-2010, looking at both the time period that led up to September 11th, 2001 and the years following it. Again, they examined the relationship between predictors of ordinary crime and county-level terrorism, using GTD data. As in their previous analyses, they found evidence of clustering, with a quarter of all attacks occurring in just 10 counties, but also with at least one attack occurring in all 48 continental states. They also found that terrorist attacks were more likely to occur in counties that had larger populations, greater residential instability, and greater language diversity. Unlike traditional crime, terrorist attacks were less likely to occur in counties that had higher levels of concentrated disadvantage.  Much like traditional crime, terrorism is not randomly distributed. Deliverables related to this subproject include:
  • Examining the Relationship between Population Characteristics, the Presence of Hate Groups and the Presence Violent Far-right Extremists at the County Level, led by Amy Adamczyk, Steven Chermak, and Joshua Freilich, sought to determine whether the presence of hate groups (including white-supremacist groups and black-supremacist groups) and other population characteristics in counties would be significantly related to the likelihood of far-right terrorists residing in those same counties. Analyses established an empirical and statistically significant link between the presence of white-supremacist hate groups and the residences of far-right terrorists, providing evidence to support a relationship that had often been assumed. Findings also revealed that far-right terrorists were more likely to live in counties characterized by larger populations and lower levels of trust in others. Deliverables related to this subproject include:
  • Financial Crime and Material Support Schemes led by Steven M. Chermak, Joshua D. Freilich, updated and analyzed the Extremist Crime Database. This subproject specifically focused on Financial Crime and Material Support Schemes committed or attempted by supporters of al-Qa’ida and affiliated movements (AQAM) 1990 to June 2014 and Far Right Extremists 1990 to 2013 in the United States. The project identified 609 financial schemes involving at least one far-right extremist occurring in the United States from 1990 to 2013. It was found that AQAM supporters committed 50 financial schemes and 100 material support schemes between 1990 and mid-2014.​​
  • Geospatial Patterns of Antecedent Behavior among Terrorists, led by Brent Smith, Kelly Damphousse, and Paxton Roberts, updated the American Terrorism Study. In their analyses, the researchers found that the locations of perpetrators’ pre-incident activities and residences may help predict their attack locations, particularly when ideology is taken into account.  For example, a greater percentage of the residences of al-Qa’ida-related (55%) and far-right (44%) perpetrators were within 30 miles of the incident location compared with far-left (29%) and environmental (25%) perpetrators’ residences. Al-Qa’ida-related perpetrators were also most active prior to an attack, conducting more antecedent acts per incident compared with far-right, far-left, and environmental perpetrators. Deliverables related to this subproject include:
  • Hot Spots of Terrorism and Other Crimes in the United States, 1970-2008, led by Gary LaFree and Bianca Bersani, identified 65 (out of 3143) counties in the United States that experienced more than six terrorist attacks between 1970 and 2008 and classified these counties as "terrorist hot spots." The counties included large urban centers (Manhattan, Los Angeles, and Miami-Dade) as well as more rural counties (Dakota County NE). A close look at the data revealed that there were different hot spots for far-right terrorism, far-left terrorism, nationalist terrorism, religious terrorism, and single-issue terrorism around the country. The research team also found significant correlations between the occurrence of terrorism and the occurrence of ordinary crime in counties, but the correlation was not perfect: terrorism also occurred in counties that experienced low rates of ordinary crime. Other county-level characteristics that were significantly and positively related to the occurrence of terrorism included residential instability and language diversity. Deliverables related to this subproject include:
  • Organizational Dynamics of FarRight Hate Groups in the United States, led by Steven M. Chermak, Joshua D. Freilich, and Michael Suttmoeller, analyzed 275 violent and non‐violent far-right hate groups. Groups whose members had committed at least one ideologically motivated violent crime were categorized as violent, and groups whose members had not were considered non-violent. The analysis revealed that 21% of the 275 far-right hate groups included in the study had members who had committed at least one violent criminal act. The researchers then examined a number of group characteristics related to organizational capacity, organizational constituency, strategic connectivity, and structural arrangements. They found that as groups increased in size or age, the likelihood of them being involved in violence increased. Violence was also related to geographic region; groups in the West and Northeast were significantly more likely to be involved in violence. On the other hand, they also found that groups that published ideological literature (e.g. newsletters, pamphlets) were significantly less likely to be involved in violence. Deliverables related to this subproject include:
  • For Profiles of Perpetrators of Terrorism in the United States, Erin Miller and Kathleen Smarick created a new database on terrorist groups. The database includes groups that committed at least one attack in the United States and is updated annually. Analysis of the data showed that terrorist groups that targeted the homeland varied in ideology, activity, and longevity. Nearly all of the groups emerged before 2000, and just over half of them carried out attacks for less than one year.  Groups with religious ideologies represented two out of five emergent groups between 2000 and 2011 but only 7% of groups between 1970 and 2011. Interestingly, about half of all groups also engaged in non-violent political activities. Deliverables related to this subproject include:
  • Terrorist Attacks in the United States, led by Gary LaFree, Laura Dugan, and Erin Miller, updated and analyzed the Global Terrorism Database, identifying more than 2,600 terrorist attacks that occurred in the United States between 1970 and 2013. These attacks varied in their lethality and targets. While taken together, the incidents resulted in more than 3,500 fatalities; 86% of those occurred on September 11, 2001, and nearly 80% of all attacks involved no fatalities or injuries. The most common targets were business-related (28%) and government-related (18%), and in many cases, buildings and infrastructure were attacked more frequently than individuals.
  • In Understanding Lone-actor Terrorism: A Comparative Analysis with Violent Hate Crimes and Group-based Terrorism, Victor Asal, Kathleen Deloughery, and Ryan D. King examined the characteristics of the 101 lone-actor terrorist attacks that occurred in the United States between 1992 and 2010 and compared them with the violent hate crimes and group-based terrorist attacks that occurred during the same period. They found that year-to-year changes in lone-actor terrorism were moderately correlated with group-based terrorism, indicating that the two seemed to ebb and flow together rather than one replacing the other. Interestingly, the locations where lone-actor terrorism occurs tended to share more demographic similarities with the locations of violent hate crime offending than with the locations of group-based terrorism. Therefore, it may be possible to learn more about where lone-actor terrorism occurs by examining patterns in violent hate crime, a type of violence that both academics and practitioners understand more fully. 

The TEVUS project at its core involves data collection and integration. As this project continues, it will enhance and expand the U.S. data on specific terrorist actors and events by allowing researchers to evaluate additional primary and secondary sources, systematically code information previously collected, update information on past incidents, broaden the scope of data collected, and identify relevant data that were previously missing. The four data sets included in the project contribute a large quantity of data on terrorism and extremist crime, each bringing distinct structures and methodologies. In several subprojects, additional data are also drawn from resources such as the Uniform Crime Reports, the U.S. Census, and data collected by other organizations.

The large amount of commonality combined with the unique methodology that each data set employs leads to a complex integration process. The data integration effort links the original data sets, as well as county-level Census data, into a larger normalized relational database. This database will support efficient analysis of the data and will be accessed through a web portal with querying and mapping capabilities. The portal will provide a seamless view of the data, and users will be able to search for information by four object types: event, perpetrator, group, or court case.

The subprojects associated with this effort involve a range of bivariate and multivariate statistical analyses that incorporate variables from the core data sets, as well as additional data collected to examine the geographic contexts of terrorism and extremist crime.


Project Period:

Selected Publications