United States Extremist Crime Database (ECDB), 1990-2010

Project Details


This project involves (1) systematic collection of open-source data on non-violent and violent criminal behavior associated with far right-wing extremist groups, including data on event, perpetrator, and victim, all integrated into a relational database; (2) statistical analyses of data, to include: descriptive analyses, bivariate analysis and multivariate regressions, and time-series analyses.

Primary Findings:

More than 650 cases concerning far-right extremists in financial crime investigations were assigned, reviewed, and coded. We have coded over 384 "financial schemes," i.e. illicit financial operations carried out over an extended period of time by one or more far-rightists in the period between 1990 and 2010. These financial schemes were committed by over 993 suspects, 80% (N= 770) of whom were identified as far-right supporters. The remaining 20% (N= 203) were non-extremist associates who collaborated with far-rightists to commit these schemes.

The most common scheme type was tax avoidance (63%), which consists of various anti-tax strategies ranging from the simple refusal to pay income taxes to more complex scams involving bogus financial instruments and offshore tax havens. The second most common scheme type was the filing of bogus liens (17%). The primary motivation for engaging in financial crimes was ideological (47%), although greed and other motives were also at play in 40% of cases. The shortest scheme lasted less than one year, while the longest one covered a period of almost 30 years from start to end. On average, a far-right financial scheme lasted for five years. The total financial losses linked to schemes coded so far amount to over $500 million. The most costly scheme in terms of financial losses was $80 million. Currently, we have coded around 90% of the cases we have identified. We plan to complete this coding during Spring 2012. Subsequently, the data will be validated and cleaned and we anticipate completing this by the end of 2012. Upon completion of data collection and validation, we will conduct longitudinal analysis to examine changes in financial extremist crime trends over time.

Another line of research will focus on exploring the relational ties between extremists and non-extremist associates to understand how financial extremist networks are formed and operate. In addition, these financial data will be used to investigate whether the criminality of political extremists "escalate" from non-violent financial offenses to violent acts. A number of far-right suspects who committed high profile ideologically motivated attacks during the last 20 years first committed financial crimes. We have also identified more than 375 far-right homicide incidents that occurred between 1990 and the present. We have coded 377 of these homicide events. These homicide incidents varied temporally and spatially. The year that the most homicides occurred was 2007 (n= 30) and the least occurred in 2010 (n= 9). The average number of homicides per year between 1990 and 2010 was 17. Over 35 of these homicide incidents occurred in prison or another custodial setting or institution. 38% of the 377 homicide events were ideologically motivated; 61% were non-ideologically motivated.

Of the ideologically motivated homicide incidents (N= 144) the most common type was anti-race/ethnicity, followed by anti-government. We identified over 50 law-enforcement officers (i.e., federal, state, local police, correctional officers, and private security guards) killed by far-right extremists. We also identified 29 far-right suspects that were killed by law enforcement (12 of these far-rightists were killed during one of the 375 far-right homicide incidents and the other 17 were killed in additional incidents where no far-right homicide occurred). In addition, we identified 41 far-right suspects who committed suicide (22 of these suspects were linked to one of the 375 homicide incidents, the other 19 committed suicide in additional incidents where no far-right homicide occurred). These far-right data have been cleaned and validated. We have published a number of articles using these data and plan to continue publishing with these data for the next few years.


The ECDB was developed in 3 stages:

I. Identifying Crimes: Five types of sources were used to identify crimes: 1. Existing databases (e.g., GTD; Monterey; RAND) 2. Official sources - Congressional hearings, ATF, IRS, DOJ (press releases & links to court documents) 3. Scholarly & journalist accounts - Systematic literature review (320 studies) 4. Watch-group reports (e.g., ADL, SPLC, etc) 5. Systematic media searches (e.g., 'skinhead & crime,' etc.)

II. Searching Crimes: Each identified criminal incident was assigned to a specific researcher who searched 26 web-based search engines to uncover all publicly available materials on it: (1) Lexis-Nexis; (2) Proquest; (3) Yahoo; (4) Google; (5) Copernic; (6) News Library; (7) Infotrac; (8) Google Scholar; (9) Amazon; (10) Google U.S. Government; (11) Federation of American Scientists; (12) Google Videos; (13) Center for the Study of Intelligence; (14) Surf Wax; (15) Dogpile; (16) Mamma; (17) Librarians' Internet Index; (18) Scirus; (19) All the Web; (20) Google News; (21) Google Blog; (22) Homeland Security Digital Library; (23) Vinelink; (24) The inmate locator; (25) Individual State Department of Corrections (DOCs); and (26) Blackbookonline.info Uncovered open source information includes: Media reports; Government documents; Court records (injunctions, affidavits, indictments, appellate decisions); Movement materials; Books; Scholarly accounts; Videos; Blogs; Opinion pieces

III. Coding Cases: In the final stage, the open source information was provided to a coder. Coders searched these incidents, suspects, and victims to verify that the original searches were complete and did not miss important information. If the original search materials were incomplete, the coder conducted "targeted follow up searches" to fill in missing values. The coders used the open source materials to code variables found in incident, scheme, suspect and victim forms that were connected to an online database. Often the search materials contained documents from different types of sources that occasionally contained conflicting information. Such discrepancies implicate reliability issues related to source type. In these situations greater weight was granted to the more "trusted" sources. Similar to Sageman (2004: 65) "in decreasing degrees of reliability... [the ECDB favors] court proceedings subject to cross examination, followed by reports of court proceedings, then corroborated information from people with direct access to information provided, uncorroborated statements from people with that access, and finally statements from people who had heard the information secondhand" (see also Damphousse, 2007). Similarly, court records were favored over media reports and media reports were favored over watch group reports.


Project Period:

More Information