U. S. Department of Justice Federal Bureau of Investigation Behavioral Analysis Unit A Study of Lone Offender Terrorism in the United States (1972 – 2015) National Center for the Analysis of Violent Crime 1 November 2019 Message from FBI Director Wray: As we’ve all witnessed, the threats we face from terrorism and targeted violence are rapidly evolving. We’ve seen a steady increase in the number of attacks and the array of attack methodologies, targets, and underlying motivations driving the attackers. The FBI is committed to using all appropriate tools and resources to prevent these acts from happening. We work side-by-side with our federal, state, local, and tribal law enforcement partners to address and mitigate threats of terrorism and targeted violence. These efforts are led largely by our formal task forces, such as Joint Terrorism Task Forces and Violent Crime Task Forces, and also through less formal, but equally important, liaison relationships between the FBI and law enforcement professionals at all levels of government. But prevention is more than just a law enforcement effort. Law enforcement is working diligently to improve its collaboration and coordination with other government entities, such as community mental health, social services, probation and parole, and educators, as well as private sector partners and stakeholders, to share information and ensure all entities are working together to help manage and mitigate threats. All citizens have a critical role in prevention. Prevention efforts are greatly enhanced by the early recognition and reporting of suspicious behaviors by those individuals around a person of concern, such as family members, peers, and community members. Bystanders need guidance to recognize concerning behaviors and overcome natural resistance to reporting. Just as important as early recognition by bystanders is the need to have well-trained, skilled, and competent receivers of that reporting – individuals who can assess potential threats and share information with other stakeholders so that they can gather additional information, further assess the threat, and take action to mitigate that threat. One method for coordinating these often complicated responsibilities is the use of multi-disciplinary threat assessment and management teams. The core concept of these teams brings stakeholders and subject matter experts together to accurately and holistically assess threats and to devise effective and appropriate threat management strategies. The FBI’s Behavioral Threat Assessment Center (BTAC), established in 2010, is a national-level multi-disciplinary and multi-agency threat assessment and management team. The BTAC provides operational support 2 to FBI field offices and our law enforcement partners on some of the most complex, concerning, and complicated threat cases related to terrorism and targeted violence. The BTAC also conducts extensive research on prior acts of terrorism and targeted violence to learn from past events, to enhance and improve prevention capabilities, and to train the community and other stakeholders involved in this space. The Lone Offender Terrorism Report (LOTR) is the latest example of this important research. The lessons learned from the LOTR are similar in many ways to past FBI research, as well as academic research, on the pre-attack behaviors, stressors, and risk factors exhibited by and experienced by previous attackers. While the attackers in this report were ideologically-motivated lone offenders, they were rarely completely isolated and alone, and they traveled down the same observable and discernable pathways to violence as other attackers. The lessons learned in this report reinforce the principle that, like other acts of targeted violence, lone offender terrorism may be preventable through early recognition and reporting of concerning behavior. This research report is unique from other research on this topic because of the richness and quality of the underlying data used to conduct the research (e.g., case files and other law enforcement records). In addition, the FBI, through the BTAC’s daily work in providing operational support on threat cases, is well positioned to analyze the findings through an operational lens and to make observations from the data, which can be used by other threat assessment professionals. With this focus, the authors of this report have attempted to look beyond the data to identify and articulate substantive suggestions and considerations for how readers might apply the lessons learned to enhance our collective prevention efforts. Christopher A. Wray Director Federal Bureau of Investigation 3 National Center For The Analysis of Violent Crime Federal Bureau of Investigation Behavioral Analysis Unit Behavioral Threat Assessment Center About the Author: 4 About the Author: The Federal Bureau of Investigation’s Behavioral Threat Assessment Center (BTAC) is a national-level, multi-agency, multidisciplinary Task Force focused on the prevention of terrorism and targeted violence through the application of behaviorally-based operational support, training and research. In this unique capacity, the BTAC provides investigative and operational support for the FBI’s most complex, concerning, and complicated international and domestic terrorism investigations. In addition, the BTAC provides threat assessment and threat management support to federal, state, local, tribal and campus law enforcement partners, as well as community stakeholders, working diligently across the United States on targeted violence prevention. Significant lines of effort on targeted violence prevention include persons/adults of concern, potential active shooters, school shootings/threats, stalking and workplace violence. Primary Authors: Lauren Richards, M.A., Oak Ridge Institute for Science and Education Research Fellow Peter Molinaro, Ph.D., University of Michigan John Wyman, Supervisory Special Agent-Unit Chief, Behavioral Analysis Unit, Federal Bureau of Investigation Sarah Craun, Ph.D., Research Coordinator, Behavioral Analysis Unit, Federal Bureau of Investigation 5 National Center for the Analysis of Violent Crime: The National Center for the Analysis of Violent Crime (NCAVC) coexists with the Federal Bureau of Investigation’s (FBI) Behavioral Analysis Unit (BAU) and are both components of the FBI’s Critical Incident Response Group (CIRG) located at the FBI Academy in Quantico, Virginia. The primary mission of NCAVC/BAU is to provide behavioral-based operational support to federal, state, local, and international law enforcement agencies involved in the investigation of unusual or repetitive violent crimes, communicated threats, terrorism, and other matters of interest to law enforcement. The NCAVC/BAU is comprised of several units, each specializing in matters such as threat assessment, crimes against adults, crimes against children, cyber and counterintelligence, violent criminal apprehension program, and research. NCAVC/BAU staff members conduct detailed analysis of crimes from behavioral, forensic, and investigative perspectives. This analysis process provides law enforcement agencies with a better understanding of offender motivations and behaviors. The analysis is a tool that provides investigators with descriptive and behavioral characteristics of the most probable offender and advice regarding investigative techniques to help identify the offender. The NCAVC/BAU also conducts research from a law enforcement perspective. NCAVC/BAU research is designed to gain insight into criminal thought processes, motivations, and behaviors. Research findings are refined into innovative, investigative techniques that improve law enforcement’s effectiveness against violent criminals and are shared with law enforcement and other disciplines through publications, presentations, and training. 6 Acknowledgments The authors would like to acknowledge the efforts of our state and local law enforcement partners and BAU Coordinators for the time and effort that was taken to collect and provide case information. The authors would also like to thank the many researchers and BAU members, both former and current, who supported this study, including retired Unit Chief (UC) Shawn VanSlyke, Dr. Robert Fein, Bryan Vossekuil, Supervisory Special Agent (SSA) Amanda Moran, Intelligence Analyst (IA) Courtney Brawley, SSA Joseph Bertoldi, IA Andrea Fancher, SSA Stephanie Yanta, Emma Waldner, Alexa Lambros, Staff Operations Specialist Jennifer Flynn, Dr. Paul Gill, retired SSA Andre Simons, SSA Terri Patterson, SSA Karie Gibson, SSA Kevin Hughey, and IA Karen McCaulley. 7 Introduction Operationally defining the lone offender terrorist is challenging, as the nature of terrorism is both highly political and contextually dependent. Over the years, there has been disagreement over the classification of violent acts under the label of “terrorism.” While an argument can be made that the infliction of fear upon a targeted group or the public is sufficient to classify an act as terrorism, the FBI definition of terrorism requires a purported motivation that goes beyond exclusively personal motivations and attempts to influence change in furtherance of extremist ideologies of a social, political, religious, racial or environmental nature. The focus of the current study examined offenders who carried out their attacks independent of any direction from a terrorist group or organization. In 2009, the FBI’s Critical Incident Response Group (CIRG) Behavioral Analysis Unit (BAU) published a guide titled Threat Management for the Lone Offender, which offered preliminary operational guidance for managing potential threats based upon case observations. The guide introduced the early conceptualization of this Lone Offender Terrorism Study and began to compile data from an initial sample of twenty-one offenders. The initial coding protocol was modeled after the first operational exploration of 83 known attacks and approaches against public officials in the US from 1949 to 1996. 1 The current study was built upon these foundations, compiling information from federal and state investigative case files, increasing the sample size from the initial sample size of 21 to 52 cases, and revising and expanding the existing coding protocol. Lone offender terrorism continues to pose a threat as violent ideological groups and terrorist organizations place emphasis on inspiring lone offender-style attacks. Additionally, individuals who view violence as an accessible and justified method for advancing their own ideological goals can independently mobilize toward violent action. While attacks directly coordinated by terrorist organizations are generally more lethal globally, the United States departs from this trend: lone offender attacks in the United States are more deadly, possibly due to strong U.S. counterterrorism capacity for disrupting Fein, R. & Vossekuil, B. (1999). Assassination in the United States: An operational study of recent assassins, attackers, and near lethal approachers. Journal of Forensic Sciences, 44, 321-33. doi:10.1520/JFS14457J; Fein, R. & Vossekuil, B. (1999). Preventing Assassination: Secret Service Exceptional Case Study Project. https://www.ncjrs.gov/pdffiles1/Photocopy/167224NCJRS.pdf 1 8 attacks from cells and organizations.2 Despite the overall low incidence rate of terrorism, counterterrorism and threat assessment professionals are responsible for navigating a diverse and amorphous landscape of threats from individuals who seek targeted violence as a solution for issues that often blend ideological and personal motivators. Lone offender terrorism is not unique to a particular religion, culture, or political affiliation. This study on lone offender terrorism attacks included offenders who carried out violent attacks in furtherance of any claimed ideology or cause, as long as the offender was primarily radicalized within the United States and carried out the attack against targets within the United States. There is a certain degree of overlap between some ideological movements, and it was not uncommon for offenders in the study to blend elements of multiple ideologies. The current report provides an overview of the data, exploring the various topics encompassed within the coding protocol, including bystander observations, offenders’ backgrounds, family and social networks, behavioral characteristics, radicalization, and attack planning. Predicting lone offender terrorism incidents is not possible, but prior research and operational experience support the conclusion that acts of targeted violence, including lone offender terrorist attacks, may be preventable through early recognition and reporting of concerning behavior. The report aims to inform broader goals of enhancing bystander education and awareness, as well to aid the prevention efforts of law enforcement and multi-disciplinary threat assessment teams working to counter targeted violence threats every day. To work toward this objective, the authors examine relevant contextual factors and analyze the statements and behaviors offenders exhibited before carrying out their attacks. Phillips, B. J. (2017). Deadlier in the US? On lone wolves, terrorist groups, and attack lethality. Terrorism and Political Violence, 29(3), 533-549. doi:10.1080/09546553.2015.1054927 2 9 Method To be included in the current study, offenders must have attempted3 or completed an act of lethal violence4 in furtherance of an identified social, political, or ideological goal. While offenders may have affiliated or associated with a terrorist organization/ideological movement or may have received assistance from others at some stage during the planning or implementation of their attacks, they must have been both the primary architect and the primary actor in the attack action. The attack must have occurred within the US and the offender must have radicalized,5 at least primarily, within the United States. Offenders were identified by reviewing FBI case files and partner law enforcement records, in addition to searching academic literature, open source media, and the National Consortium for the Study of Terrorism and Responses to Terrorism (START) Global Terrorism Database (GTD).6 Approximately 240 cases were initially identified as potentially meeting inclusion criteria. Upon reviewing the elements of each case, 93 cases ultimately met all inclusion criteria, but 41 cases were excluded due to either a lack of sufficient information7 or the presence of pending adjudications or ongoing investigations. Fifty-two cases were coded for use in the study analysis, comprising a sample of lone offender terrorism cases from 1972 to 2015.8 Using a 244-question protocol, researchers double-coded9 cases using information obtained primarily from closed investigative files and supplemented by open source information. Planning and attackrelated variables were coded as they related to the offenders’ index attack, meaning the first offense that met all of the inclusion An attempt was included if the offender, in an effort to complete their attack goal, completed a substantive overt action that could have resulted in fatalities if not for intervention or happenstance. 4 “Lethal violence,” as used in this study, specifically referred to acts of violence that were intended to kill people. 5 The FBI defines radicalization as the process by which individuals come to believe that engagement in or facilitation of non-state violence to achieve social and political change is necessary and justified. 6 National Consortium for the Study of Terrorism and Responses to Terrorism (START). (2015). Global Terrorism Database [Data file]. Retrieved from https://www.start.umd.edu/gtd 7 Cases were excluded if there was not enough information to code at least 60 percent of the coding protocol variables as present or absent. 8 The data reported should not be used to make conclusions regarding trends over time. 9 To double-code cases, two research team members independently coded case files using the research protocol. The responses from each coder were then compared and any differences were discussed in order to reach agreement. If agreement could not be reached, differences were reconciled by a third team member. 3 10 requirements detailed above.10 Researchers coded for the presence or absence of behaviors only when objective case material indicated such evidence – otherwise, the variable of interest was coded as “unknown/unclear.” Therefore, the data presented in this report are conservative estimates, and the actual rate of occurrence of certain behaviors and characteristics may be higher than what is reported. The current report has removed all identifying offender information to allow for the widest dissemination possible. Fourteen offenders engaged in additional acts of terrorism. Preliminary data was recorded for those additional acts as part of a Serial Addendum, which was attached to the main coding protocol. The results of the Serial Addendum are discussed briefly in this report. Unless otherwise specified, any mention of offenders’ attacks throughout the report refers to offenders’ index attacks. 10 11 Part One: The Offenders Demographics The 52 offenders in the study varied widely in characteristics such as age, race, relationship status, and educational background. Consistent with conclusions reached by both academic and government researchers studying targeted violence, the study found no evidence to support the existence of any meaningful demographic profile of a lone offender terrorist.11,12 Gender Perhaps the only demographic trend consistently seen in studies of targeted violence is the overwhelming representation of male perpetrators. Although women can and have engaged in acts of targeted violence, all 52 offenders in the lone offender terrorism study sample were men.13 Age The youngest offender was 15 years old and the oldest was 88, with the distribution of ages displayed in Figure 1.14 The average age at the time of the attack was 37.7 years old. Radical Islamic violent extremists, who averaged 26.3 years of age at the time of their attack, were significantly younger than the other ideological groups (p<0.05).15 11Silver, J., Simons, A., & Craun, S. (2018). A study of the pre-attack behaviors of active shooters in the United States between 2000–2013. Federal Bureau of Investigation, U.S. Department of Justice, Washington, D.C. 20535. 12Gill, P. (2015). Lone-actor terrorists: A behavioural analysis. Routledge. 13 The few female lone offenders identified in the collection phase of the study were not included in the current data set due to either a lack of file information or pending investigations/adjudications. 14 Except for the youngest offender, all offenders were age 18 or older. 15 Age comparisons were run for all ideological groups, but only statistically significant results were included in the report. It is possible that additional significant differences would be found with the addition of new cases. 12 Figure 1 Offender Age at Time of Attack* (N=52) 25% 21% 20% 17% 15% 15% 12% 10% 10% 8% 6% 6% 50-54 55-59 4% 5% 2% 0% 15-17 18-24 25-29 30-34 35-39 40-44 45-49 60+ *Does not sum to 100 due to rounding. Offender Age at Time of Attack (N=52) Age Range 15-17 years old 2% 18-24 years old 17% 25-29 years old 12% 30-34 years old 21% 35-39 years old 10% 40-44 years old 15% 45-49 years old 4% 50-54 years old 6% 55-59 years old 6% 60 years old and up 8% 13 Citizenship, Ethnicity, and Race Most offenders were born in the US (n=47, 90%). Four offenders (8%) were naturalized citizens and one offender (2%) was a legal permanent resident. Most offenders were white/Caucasian (n=34, 65%), while the remaining 18 offenders (35%) were divided among five different racial groups, as shown in Figure 2.16 Figure 2 Race/Ethnicity (N=52) 80% 65% 60% 40% 20% 13% 8% 8% 4% 2% Asian Hispanic 0% White Middle Eastern Black Bi-racial Race/Ethnicity (N=52) Offender Race White 65% Middle Eastern 13% Black 8% Bi-racial 8% Asian 4% Hispanic 2% 16In cases where race was ambiguous, coders considered any information available about how the offenders self-identified. 14 Relationship Status As shown in Figure 3, most offenders were not in a relationship or married at the time of their attack (n=38, 73%). Figure 3 Relationship Status (N=52) 60% 48% 40% 23% 21% 20% 4% 2% 2% 0% Relationship Status (N=52) Offender Relationship Status Single 48% Divorced/Separated 23% Married 21% Partnered 4% Widowed 2% Unknown 2% Seventeen offenders (33%) had children, but only six offenders were known to have had their children under their care during the year prior to the attack. The remaining 11 offenders either had no or minimal contact with their minor-aged children or had children who were adults at the time of the attack. 15 Education Most offenders completed an associate degree or at least some college education (n=39, 75%). Slightly more than a third (n=19, 37%) obtained a bachelor's degree or higher. Six offenders (12%) were students at the time of their attack. Figure 4 Level of Education Completed (N=52) LESS THAN HS 8% HS/GED 15% ASSOCIATE/SOME COLLEGE 38% BACHELOR'S 23% MASTER'S 10% DOCTORAL 4% UNKNOWN 2% 0% 10% 20% 30% 40% 50% Level of Education Completed (N=52) Offender Level of Education Less than High School 8% High School Degree/GED 15% Associate Degree/Some College 38% Bachelor’s Degree 23% Master’s Degree 10% Doctoral Degree 4% Unknown 2% 16 Employment More than half of the offenders (n=28, 54%) were neither working nor attending school at the time of their attack. Three offenders (6%) were retired and two offenders (4%) were not working due to a disability. Of those employed (n=16, 31%), half were employed fulltime and half were employed part-time.17 Nineteen offenders (37%) were financially self-sufficient and 15 offenders (29%) were supported by family (i.e., parents, spouses, or other family members). Four offenders (8%) were supported primarily through federal or state aid and three (6%) were supported by other means. The primary financial source was unclear for the remaining 11 offenders (21%). Religious Affiliation Of the 26 offenders who identified as religious, 13 affiliated as Christian (50%), nine as Muslim (35%), one as Jewish (4%), and three as belonging to another religion (12%). In nine cases (17%), there was some indication the offender held spiritual or religious beliefs, but the offender either did not affiliate with any organized religion or their affiliation was unclear. Seventeen offenders (33%) either stated they were not religious or no information was present to suggest the offender held any religious affiliation. More than a third of offenders engaged in religious seeking at some point before their attack (n=20, 38%), either exploring one new religion or trying multiple religions or religious denominations. Military Service Nineteen offenders (37%) had served in the military,18 of which 10 served in the Army (53%), six in the Navy (32%), two in the National Guard or Reserves (11%), and one in the Coast Guard (5%). Five offenders (26%) deployed to a combat zone while in service. There was wide variation among offenders in length of time served and in the quality of their performance during their military career. 17 18 Employment status was unclear or unknown in two cases (4%). Percentage is out of 51 adult subjects. 17 Another five offenders (10%) took steps toward joining the military but were either rejected during the application process or did not follow through upon realizing they would not meet qualifications or existing policies.19 Military Discharge (n=19) Discharge Status n % Honorable 8 42% Medical/Psychiatric (Honorable) 3 16% General 3 16% Other than Honorable 1 5% Retired 1 5% Active service 1 5% Unclear/ Unknown 2 11% The existence of military experience was not connected to attack lethality (p=ns). While more than a third of the offenders in the study had prior military service,20 their military experiences differed considerably in scope (e.g., service locations, training, job types, and performance levels). Demographics: Operational Considerations When available, researchers compiled data pulled from official educational, employment, and military records, as well as information from interviews of individuals who knew offenders at various points in offenders’ lives. Official records provided useful information regarding dates, locations, and offenders’ knowledge or skills. Prior interactions with peers, subordinates, and superiors informed researchers about Of the two offenders rejected from the military, one admitted to drug use and no further information was available for the second. Of the three offenders who did not follow through with processing, one was due to continued drug use, one scored poorly on a screening assessment, and one did not meet the physical or educational requirements needed to enter the specialty in which he was interested. 20 Data from the U.S. Census Bureau and U.S. Department of Defense indicate that the number of American men who are active military or veterans have decreased over the past four decades. While rates in 1972 hovered around 43 percent of American men, by 2015, fewer than 15 percent of American men were active service or veterans. 19 18 offenders’ patterns of behavior, work performance, and the quality of their interpersonal relationships. Using multiple sources when available allowed researchers to compare pieces of information that were complementary or contradictory, allowing for a more thorough analysis. For instance, military records provided useful information about offenders’ service dates, job roles, and training - sometimes contradicting offenders’ statements about their service (e.g., whether they served in a special forces unit or a combat zone). However, supplemental information was invaluable in gaining a more complete picture of offenders’ histories. In one case, an offender had received a general discharge due to misconduct, supported by documentation of multiple alcohol incidents and the offender being absent without leave on at least one occasion. Although military documents provided a wealth of information, interviews with military peers and supervisors suggested that one of the driving factors behind the offender’s discharge was his continued racist activities while he was in service. Additionally, discharge status was not always fully indicative of performance, as seen in one case where an offender’s service was characterized as honorable, but the offender’s military record included documentation of a history of substance use and the revocation of his security clearance. Therefore, researchers considered the totality of the information pulled from official and unofficial records when answering questions about offenders’ life histories and prior behaviors. Fewer than a third of the offenders were actively employed at the time of their attack and fewer still were employed full time.21 In total, 28 offenders (54%) were neither employed nor attending school, underscoring the fact that many offenders had a certain degree of free time to focus on their grievances and ideologies, and to engage in logistical planning and preparation for their attack. These findings are consistent with BAU operational experience and direct interactions with previously-radicalized subjects, who have repeatedly cited the abundance of time and the lack of other responsibilities as a contributing factor to their radicalization. 21As measured by the U.S. Bureau of Labor Statistics (2019), rates of employment among civilian noninstitutionalized persons ages 16 and older ranged between 56.1 and 64.4 percent during the time span of 1972 to 2015. Americans not employed were either unemployed and seeking work or were otherwise not considered part of the labor force (e.g., retired, disabled, or not actively looking for work). 19 For investigators assessing potential threat cases, issues with employment can raise questions about whether individuals struggle to maintain minimum standards of employment, comply with authorities, and work successfully with others. Lack of gainful employment may also be an indication of potential stressors, such as financial strain, problematic interpersonal relationships at work, inappropriate behaviors, or the existence of grievances regarding prior terminations. Prior Criminal Behavior and Aggression Thirty-five offenders (70%) were arrested at least once as an adult before their attack, with the average number of adult arrests at 2.7.22 Slightly more than half of the offenders (n=26, 52%) were arrested more than once as an adult.23 Figure 5 Total Prior Adult Arrests (n=50) 35% 30% 30% 25% 18% 20% 20% 16% 15% 10% 8% 8% 5-9 10+ 5% 0% 0 1 2 3-4 NUMBER OF ARRESTS Percentage is out of 50 offenders. This excluded the one offender who was not an adult (i.e., under age 18) and the one offender whose criminal history was unknown. 23 For American men, the lifetime prevalence of ever being arrested is estimated to be 43 percent (Barnes, Jorgensen, Beaver, Boutwell, & Wright, 2015). 22 20 Total Prior Adult Arrests (n = 50) Prior Arrests 0 30% 1 18% 2 20% 3 -4 16% 5-9 8% 10 and up 8% Out of the 34 cases (65%) in which sufficient information was available regarding offenders’ adolescent years, nine offenders (26%) were arrested at least once before the age of eighteen. Seven out of those nine offenders were arrested again as an adult. At least 15 offenders (29%) were previously arrested for one or more violent offense.24,25 Approximately a third of offenders (n=17, 33%) were arrested for one or more alcohol or drug offense. Out of those offenders with prior arrests (n=37), approximately half (n=19, 51%) served time in a correctional facility. Because formal arrest histories may not capture all violent or aggressive behavior, the researchers examined additional indicators of violence and aggression. Based upon interviews with those who knew the offenders and other pieces of case evidence, many offenders (n=43, 83%) had previously exhibited hostile, explosive, and/or aggressive behavior (e.g., threatening statements or a volatile temper). Civil protection or no contact orders were filed against nine offenders (17%) at one point before their attacks. In total, at least 30 offenders (58%) were known to have previously carried out physical violence.26 Criminal Issues: Operational Considerations Many offenders were arrested at least once before their attack and more than a third served time in prison. Previous publications have Details of known prior arrests were sometimes incomplete or unclear. Therefore, the numbers related to arrest characteristics (i.e., violent/non-violent arrests or drug/alcohol-related arrests) are likely underestimates. 25 As this paragraph combines juvenile and adult arrest information, percentages are out of 51 cases. 26 This number included both offenders who were never arrested for violence and offenders who had prior arrests for violent offenses but carried out additional acts of violence. Twenty-six offenders had engaged in battery or other physical violence that did not result in an arrest. 24 21 emphasized that while prior arrests do not directly predict an individual’s risk for targeted violence, arrest records are a valuable source of information for threat assessors and can potentially shed light on prior violent behavior.27 It is important for threat assessors to examine all available law enforcement records, particularly those memorializing interactions with a person of concern. While the majority of the offenders had not been previously arrested for a violent offense, most had previously exhibited behavior that was at the very least hostile or aggressive. More than half had previously engaged in physical battery or violence, either reflected in the offender’s arrest history or in accounts given by individuals who knew the offender. Although prior violence was not present in all cases, operational experience has noted that the demonstrated willingness to use violence to resolve problems can suggest an enhanced risk of future violence. Mental Health and Substance Use A quarter of the offenders (n=13, 25%) were formally diagnosed with one or more psychiatric disorders at some point before their attack. Mood disorders such as depression and bipolar disorder (n=9) were the most commonly diagnosed, followed by psychotic disorders such as schizophrenia (n=4). Other diagnosed disorders included substance use disorder (n=3), personality disorder (n=1), adjustment disorder (n=1), and attention disorder (n=1). Six offenders were diagnosed with more than one disorder. Case materials typically do not contain medical or mental health records. Therefore, researchers were unable to obtain conclusive data about how many offenders were receiving treatment, the types of treatment received, or offenders’ adherence to any prescribed medication regimens. Of the 13 offenders who had received a mental health diagnosis, at least five offenders (38%) were known to have been receiving psychiatric counseling and/or medication within the year before their attack. Amman, M., Bowlin, M., Buckles, L., Burton, K., Brunell, K., Gibson, K., Griffin, S., Kennedy, K., & Robins, C. (2017). Making prevention a reality: Identifying, assessing, and managing the threat of targeted attacks. Washington DC: Federal Bureau of Investigation, National Center for the Analysis of Violent Crime, Behavioral Threat Assessment Center. 27 22 Seven additional offenders (13%) were diagnosed with one or more psychiatric disorders after their attack, upon evaluation by one or multiple mental health professionals. Post-attack diagnoses included psychotic disorders (n=6), mood disorders (n=2), personality disorders (n=2), and adjustment disorder (n=1). Mental Health Stressors In addition to the 20 offenders (38%) who were ultimately diagnosed with a psychiatric disorder, there were 18 cases (35%) in which offenders were suspected by others (e.g., friends, family, associates, or mental health professionals) of having one or more undiagnosed mental disorders.28 While it cannot be concluded that an offender likely had a suspected disorder, the data suggests that people in the offenders’ networks may have noted behaviors or symptoms that could have been indicative of mental health stressors. Out of the 20 offenders diagnosed with a disorder, there were 13 cases in which other individuals or the offenders themselves noted symptoms of a separate disorder. 28 23 Researchers measured additional variables related to specific perception issues. Coding was based upon observable information in the case files (e.g., offender behaviors, writings and online postings, and/or statements from bystanders) and focused on issues that were present before the commission of the offenders’ attacks. The presence of individual perception issues did not indicate the offender had a disorder.  Paranoia (n=32, 62%) More than half of the offenders demonstrated a pervasive distrust and suspiciousness of others, or a belief that others were plotting to harm them. Paranoia was either connected to fear or suspicion of specific individuals or was general and vague, such as fears linked to beliefs in conspiracies about government surveillance or manipulation by secret organizations. (Although the percentage of offenders experiencing paranoia appears high, it should be considered that paranoia may be expected to some degree for an individual who is planning to carry out violence.)  Delusions (n=23, 44%) Almost half of the offenders held a false belief or set of false beliefs that were irrational and persistent.  Grandiosity (n=18, 35%) Slightly more than a third of the offenders possessed inflated and unrealistic self-perceptions, or beliefs that they were individually superior to others. Most of the offenders in the study did not exhibit signs of grandiosity.  Hallucinations (n=5, 10%) Prior to their attacks, few offenders reported or showed signs of experiencing auditory hallucinations, described as hearing voices that were not actually present and were distinct from their own thoughts. 24 Suicidality Twenty-one (40%) offenders expressed suicidal ideation29 at some point before their attack, three of whom were known to have engaged in a prior suicide attempt. Seven of the 21 offenders (33%) with a history of suicidality ultimately died during the commission of their attack.30 Thirteen of the 20 offenders (65%) who died during the commission of their attack31 had no known history of suicidality. The presence of prior suicidal ideation alone was not indicative of whether an offender would ultimately commit suicide as part of their attack (p=ns). However, the study did not measure when suicidal ideation occurred in relation to the attack, or whether suicide was specifically part of an offender’s attack planning, Drug Use At least 26 offenders (50%) exhibited some evidence of prior drug use. Of those offenders who had used marijuana, about half (n=12, 52%) had used additional illegal drugs. Prior illegal drug use occurred both among offenders with a psychiatric diagnosis (n=11, 55%) and those without a diagnosis (n=15, 48%).32 While few offenders were formally diagnosed with a substance use disorder, issues with drug and alcohol use were common among offenders. In at least 22 cases (42%), one or more individuals who knew the offender had expressed concern about the offender’s alcohol and/or drug use. This finding was unsurprising, given the prevalence of prior drug- or alcohol-related arrests (n=17, 33%). In 20 cases (39%), alcohol alone was known to have harmed at least one area of the offender’s life (e.g., job loss, relationship or family conflict, or legal charges). Suicidal ideation included articulated thoughts, plans, or intentions of committing suicide. Four died by suicide and three died by law enforcement response. 31 Four died by suicide and nine died by law enforcement response. 32 Percentages are based upon the number of offenders who had a known history of drug use (n=20) and the number of offenders who did not have any known history of drug use (n=31), respectively. The presence of prior drug use was unknown/unclear in one case. 29 30 25 Mental Health and Substance Use: Operational Considerations Although many offenders appeared to experience mental health stressors, most offenders were not diagnosed with a psychiatric disorder. Research in the violence risk assessment field has shown that a psychiatric diagnosis is not independently predictive of an individual’s risk for future violence.33,34 Instead, severe mental health issues in combination with co-occurring risk factors, such as substance abuse, can impact risk. Drug Use Prevalence by Type Drug n % of prior users (n=26) % of all offenders (N=52) Marijuana 23 88% 44% Cocaine/Methamphetamine 10 38% 19% LSD/Psilocybin 8 31% 15% Prescription Abuse 4 15% 8% MDMA/Ecstasy 2 8% 4% Other 4 15% 8% Unclear Type 1 4% 2% Among the 20 offenders (38%) who were ultimately diagnosed with a psychiatric disorder, mood and psychotic disorders were the most common diagnoses. The rate of offenders diagnosed with mood disorders is similar to the estimated percentage of American men who Monahan, J., Steadman, H.J., Silver, E., Applebaum, P.S., Clark Robbins, P., Mulvey, E.P., Roth, L.H., Grisso, T., & Banks, S. (2001). Rethinking risk assessment: The MacArthur study of mental disorder and violence. New York, NY: Oxford University Press. 34 Elbogen, E.B., & Johnson, S.C. (2009). The intricate link between violence and mental disorder: results from the National Epidemiologic Survey on Alcohol and Related Conditions. Archives of General Psychiatry, 66(2), 152-161. doi:10.1001/archgenpsychiatry.2008.537 33 26 experience a mood disorder at some point over their lifetime (17.5%).35 However, while representing a small subset of the overall sample, the percentage of offenders diagnosed with a psychotic disorder (19%) appears higher than American men’s lifetime prevalence estimates of schizophrenia (<1.0%) or psychotic symptoms (10.6%).36,37 Additionally, the rate of prior suicidal ideation (35%) among offenders was higher than general population lifetime prevalence estimates (15.6%).38,39 These higher rates indicate areas for future research. While an official diagnosis provides some indication of a subject’s psychological wellness and state of mind, threat assessors should focus specifically on subjects’ behaviors, functioning, and overall stressors. By doing so, threat assessors can consider the impact of specific symptoms on a subject’s energy, capability, and coherence.40 Not all individuals with a given diagnosis will show the same symptoms or level of functioning. Symptoms may fluctuate over time and can be influenced by internal or environmental circumstances that are particular to the individual. Additionally, not all individuals who exhibit symptoms or experience mental health stressors meet the criteria for a disorder, as seen with the percentage of offenders who were suspected of having a disorder but were not evaluated or diagnosed. Awareness of a subject’s state of mind, coping mechanisms, and how they handle confrontation allows for more strategic planning if the need arises to intervene or address concerns with the subject. Harvard Medical School. (2007). Lifetime prevalence DSM-IV/WMH-CIDI disorders by sex and cohort. [Table 1]. National Comorbidity Survey (NCS). Retrieved July 30, 2019 from https://www.hcp.med.harvard.edu/ncs/index.php. 36 Wu, E. Q., Shi, L., Birnbaum, H., Hudson, T., & Kessler, R. (2006). Annual prevalence of diagnosed schizophrenia in the USA: a claims data analysis approach. Psychological medicine, 36(11), 1535-1540. doi:10.1017/S0033291706008191 37 Cohen, C. I., & Marino, L. (2013). Racial and ethnic differences in the prevalence of psychotic symptoms in the general population. Psychiatric Services, 64(11), 1103-1109. Retrieved from https://ps.psychiatryonline.org/doi/pdf/10.1176/appi.ps.201200348 38 Nock, M.K., Borges, G., Bromet, E.J., Alonso, J., Angermeyer, M., Beautrais, A., Bruffaerts, R., Chiu, W.T., De Girolamo, G., Gluzman, S., & De Graaf, R. (2008). Cross-national prevalence and risk factors for suicidal ideation, plans and attempts. The British Journal of Psychiatry, 192(2), 98-105. doi:10.1192/bjp.bp.107.040113 39 The estimated lifetime prevalence of suicide attempts among Americans is five percent (Nock et al., 2008). Three offenders (6%) were known to have had prior suicide attempts. 40 Amman, M., Bowlin, M., Buckles, L., Burton, K., Brunell, K., Gibson, K., Griffin, S., Kennedy, K., & Robins, C. (2017). Making prevention a reality: Identifying, assessing, and managing the threat of targeted attacks. Washington DC: Federal Bureau of Investigation, National Center for the Analysis of Violent Crime, Behavioral Threat Assessment Center. 35 27 Prior drug use rates among offenders are roughly equal to estimates of drug use in the general population.41 However, more than a third of offenders experienced employment, interpersonal, or legal issues stemming from problems with substance use. In considering substance use, threat assessors should look for information about how drug use impacts a subject’s behavior and note any significant changes in usage. While decreased drug use may indicate positive changes in a subject’s life situation, it does not necessarily eliminate concern. Threat assessors have noted some cases in which subjects planning an attack have decreased or stopped drug use in order to maintain a clearer mind for attack preparation and action. In cases where a person of concern is receiving mental health treatment, law enforcement should share information regarding concerns with the treating mental health practitioner. In many cases, such information sharing only flows in one direction. However, raising concerns of which the mental health practitioner may be unaware can provide the practitioner with more context when treating the individual. Communications may help illuminate potential threats when the practitioner considers the shared information in conjunction with other behaviors or statements shared by the individual during treatment. Additionally, law enforcement and other threat assessors need to be familiar with the Health Insurance Portability and Accountability Act (HIPAA), which provides exceptions to patient confidentiality in situations where there is an identified threat of violence. Pre-establishing communication between law enforcement and mental health practitioners can enable a faster response in emergency situations. Operational experience indicates that such proactive communication and team-building between law enforcement and community-based mental health practitioners has greatly enhanced threat mitigation efforts. Center for Behavioral Health Statistics and Quality. (2018). 2017 National Survey on Drug Use and Health: Detailed Tables. Substance Abuse and Mental Health Services Administration, Rockville, MD. Retrieved August 7, 2019 from https://www.samhsa.gov/data/sites/default/files/cbhsqreports/NSDUHDetailedTabs2017/NSDUHDetailedTabs2017.pdf 41 28 Radicalization Radicalization is the process through which an individual transitions from a nonviolent belief system to a belief system to actively advocate, facilitate, or use unlawful violence is necessary and justified to affect societal or political change. The process by which an individual radicalizes can be difficult to discern fully, as radicalization is a nonlinear process that is shaped in many ways by an individual’s private thoughts, experiences, and opinions. Therefore, radicalization is highly personal and specific to each individual, and can vary greatly in both how it occurs and in how it manifests to outside observers. Multiple pathways can lead to radicalization, and radicalization itself does not always result in mobilization to violence. However, elements of radicalization could often be observed in the statements offenders made (either in conversations with others or in writings and internet postings), the materials and information they consumed, and the situations or relationships they sought out in person and online. The following sub-sections highlight the offenders’ ideologies and grievances, along with a brief discussion of the timeframes in which they radicalized. Ideologies Like other perpetrators of targeted violence, the offenders in this study accepted the use of violence as a means to achieve a goal. As stated earlier, what separates the lone offenders terrorists are the claims that their violence is done in service of larger ideological goals such as inciting social or political change. Although some offenders did not conform to a precise set of ideological beliefs, most offenders were categorized 42 as primarily carrying out their attack for the following ideological causes: anti-government violent extremism (n=13, 25%), racially-motivated violent extremism advocating for the superiority of the white race (n=10, 19%), radical Islamist violent extremism (n=10, 19%), pro-life violent extremism (n=5, 10%), racially-motivated violent extremism using force or violence in response to real or perceived racism and injustice in American society (n=2, 4%), and environmental violent extremism (n=2, 4%). The Primary ideology was categorized based upon the target selected and the offender’s rationale for selecting the target. Definitions for the listed ideologies adhere to the FBI’s categorizations at the time of the report publication. 42 29 remaining 10 offenders (19%) were placed in an “other”/idiosyncratic category.43 Figure 6 Primary Ideology (N=52) Anti-Government Violent Extremism 25% Racially-Motivated Violent Extremism 23% Radical Islamist Violent Extremism 19% Pro-Life Violent Extremism 10% Environmental Violent Extremism 4% Other Violent Extremism 19% 0% 5% 10% 15% 20% 25% 30% Primary Ideology (N = 52) Ideologies Anti-Government Violent Extremism 25% Racially-Motivated Violent Extremism 23% Radical Islamist Violent Extremism 19% Pro-Life Violent Extremism 10% Environmental Violent Extremism Other Violent Extremism 4% 19% The sample of lone offender terrorists used for this report should not be extrapolated to draw conclusions about US domestic terrorism trends. 43 30 The offenders in the anti-government violent extremism category (n=13, 25%) adhered to a range of ideologies, from those that endorsed the belief that the current government was corrupt to those that were against the idea of government itself. Many of the anti-government extremists in the study shared the conviction that part or all of the United States government had been corrupted. Offenders sometimes expressed belief in a global conspiracy, or “New World Order,” which purportedly seeks to seize property and rights away from citizens. Also included were sovereign citizen extremists, who saw themselves as independent from the United States and its jurisdiction, viewing government representatives, courts, taxing entities, and most law enforcement officers as having no legitimate authority. These beliefs were then used to support the use of force or violence to overthrow, challenge, or “defend” against the U.S. government and its representatives. Racially/ethnically motivated violent extremism (n=12, 23%) encompasses threats involving the use or threat of force or violence, in violation of federal law, in furtherance of political or social agendas which are deemed to derive from bias, often related to race, held by the actor against others, including a given population group. One significant sub-group of racially motivated violent extremists (n=10, 19%) use force or violence in violation of criminal law in response to their belief in the superiority of the white race and justify violence through a variety of political, cultural, and religious beliefs. Some believe the U.S. Government is conspiring with minority populations to bring about the demise of the white race and respond to this perceived threat through attacks on minorities and perceived race traitors, plots to overthrow the government, or efforts to establish a separate white homeland. Another sub-group of racially motivated violent extremists (n=2, 4%) use force or violence in violation of criminal law in response to real or perceived racism and injustice in American society; some do so in furtherance of establishing a separate black homeland or autonomous black social institutions, communities, or governing organizations within the United States. A desire for physical separation is typically based on a religious or political belief system, which is sometimes formed around or includes supremacy or superiority. Offenders in the radical Islamist violent extremism category (n=10, 19%) adhered to ideologies that justified the use of force or violence to 31 perceived threats to Islamic nations, societies, or values. The study included offenders who sought to engage in violent jihad. These offenders often voiced support for, were inspired by, or claimed loose affiliation with foreign terrorist organizations (FTOs) such as alQaeda (AQ) or the Islamic State of Iraq and al-Sham (ISIS), but were not following any direction or instruction from an FTO to carry out their attack. Also included were offenders who may not have explicitly advocated for violent jihad or claimed affiliation to an FTO, but sought to use violence as retaliation for US military action against Muslim nations or groups overseas. Abortion extremism is separated into two categories, pro-life and prochoice. The pro-life violent extremist offenders (n=5, 10%) claim moral legitimacy to save the life of the unborn and justify murder, arsons, bombings, blockades, and threats against reproductive health care facilities to achieve this goal. Pro-choice violent extremists (n=0, 0%) believe it is their moral duty to protect those who provide or receive abortion services and have engaged in violence, harassment, and threats against members of the pro-life movement. Offenders in the environmental violent extremism category (n=2, 4%) viewed violence as a legitimate option to halt or bring attention to real or perceived threats to the environment. The “other” violent extremism category (n=10, 19%) encompassed offenders who possessed ideologies that were idiosyncratic and did not fit into a clearly identified ideological movement. Ideologies either covered numerous topics or were highly specific to a single issue or grievance. For example, two offenders (4%) were motivated by a selfprofessed anti-liberal ideology and targeted organizations and individuals they perceived as promoting liberal causes or politics. Another two offenders (4%) were motivated by a violent anti-law enforcement message that appeared unrelated to any broader ideology (e.g., the anti-government sentiment held by sovereign citizens). 32 Offenders were sorted into groups based upon the primary or overarching ideology that framed their index attack. However, most offenders (n=39, 75%) either supported more than one violent ideology or supported a violent ideology that involved multiple themes or components. For example, at least two offenders who VIOLENT IDEOLOGICAL attacked the Jewish THEMES (N=52) community and were n % categorized primarily as ANTI-US 26 50% racially motivated violent GOVERNMENT extremists who believed in the superiority of the RELIGION 22 42% white race also believed RACE/ETHNICITY 16 31% violence against abortion providers was justified OTHER / 14 27% because they saw abortion IDIOSYNCRATIC as a threat to the white PRO-LIFE 11 21% race. At least three offenders explicitly ANTI-LAW 10 19% discussed targeting anyone ENFORCEMENT they considered immoral SEXUAL 7 14% and any individual or ORIENTATION institution that sanctioned immorality. All three ANTI-CAPITALISM 5 10% supported the use of ENVIRONMENTAL/ 3 6% violence against racial ANIMAL RIGHTS and/or religious minorities, abortion providers, homosexuals, and the US government. Although not always a focal theme of offenders’ ideologies, the view that intended victims were immoral was seen in half of the cases studied (n=27, 52%). To capture the multi-faceted nature of offenders’ violent belief systems, the above table reflects the occurrence of various themes in which the offender expressed support for violence in defense of or against the listed category. Although not included below, there were additional cases in which offenders held extreme views about a particular theme, but it was unclear whether the extreme views were violent in nature. Figure 7 displays the overlap between the three most commonlyoccurring violent ideological themes. Overlap existed when there were blurred lines between constructs, such as when an offender’s racial or ethnic identity was intertwined with their religious ideology. Other times, an offender either held multiple extreme beliefs or 33 perceived multiple sources as contributing to their grievances. For instance, an offender who endorsed white supremacy may have also believed the government was either engaging in a conspiracy with minorities or actively working to suppress their views. Figure 7. Overlap in the Three Most Common Ideological Themes (n=40) Overlap In Offender Ideologal Themes (n=40) Government Only 14 offenders Religion Only 4 offenders Race Only 2 offenders Government & Religion 6 offenders Government & Race 2 offenders Religion & Race 8 offenders Government, Religion & Race 4 offenders 34 Grievances Although the offenders carried out violence in service of a stated ideological goal, these broader claimed purposes rarely existed in isolation from personal motives. Many of the offenders had personal experiences, motives, and life stressors that co-mingled with their violent ideologies and their reasons for carrying out an attack. Most offenders (n=36, 69%) had an identifiable primary grievance,44,45 defined as a real or perceived injustice or feeling of being wronged. Grievances could be directly personal, as seen in cases where offenders perceived themselves to be the target of discrimination, unfair practices, or targeting by other individuals, groups, or institutions. In other instances, offenders were focused on perceived injustices they believed were being carried out against other groups or against society. Offenders’ grievances sometimes incorporated beliefs in conspiracy theories.46 Grievances did not have to be logical or rational, and sometimes only made sense to the offender. Primary grievances could be specific and directly connected to a target, as seen with the anti-abortion offenders in the study who attacked reproductive health care centers or doctors who performed abortions. However, grievances could also be much broader and indirectly related to target choice. For example, one offender with a history of engaging in sprawling rants about politics and the government ultimately shot one civilian and took a second civilian hostage in order to gain media attention. The ultimate target choice was neither part of the offender’s grievances nor symbolic of his Grievances are the cause of distress or resentment, but can sometimes be nebulous. For coding purposes, a grievance was recorded when an expressed injustice or complaint appeared to be particularly important to the offender. Coders assessed the totality of information pulled from offenders’ statements, writings, and the amount of time they spent focused on the issue or engaged in related activities (e.g., political protest). 45 The 69 percent is likely an underestimate, as primary grievance was unknown/unclear in 15 cases (29%). Many of the unknown/unclear cases were coded as such when it was either unclear whether the offender had a primary grievance (i.e., multiple grievances existed, with no individual grievance appearing to be more prominent than the others) or when the offender had vague complaints that resembled a grievance, but there was not enough information to clarify the complaint or to conclude how serious the complaint was to the offender. 46 Although the study did not directly measure whether offenders’ primary grievances or ideologies were centered around conspiracies, the study did find that at least 24 offenders (46%) discussed or consumed information about conspiracy theories (i.e., beliefs that an event or series of events/circumstances are covertly planned, manipulated, or guided by influential groups or organizations). 44 35 ideological message. Instead, the target was an accessible way for the offender to gain a platform to air his grievances. Radicalization Timeframe It is difficult to determine when exactly an individual first begins to adhere to an ideology and when they first begin to accept the use of violence in furtherance of that ideology. Researchers examined offenders’ statements and writings, as well as statements made by those who knew the offenders, to gather any available information relating to a timeline of radicalization and mobilization.47 In the 36 cases (69%) where time estimations could be made, 34 offenders (94%) were involved with their ideology for more than a year before they carried out their attack. Only two offenders first became involved with their ideology within the year leading up to their attack. Unless clearly articulated by the offender, it was even more difficult to estimate when the offender first decided to carry out their attack. In the 21 cases (40%) where such information was available, 17 offenders (81%) developed their initial idea to attack within the year. Of those 17 offenders, six (35%) developed their initial idea within a month prior to their attack. Radicalization: Operational Considerations Bystander observations may not provide full context when trying to determine an individual’s potential for violence. It is extremely difficult to identify a specific timeline of when an individual radicalizes in relation to a decision to personally engage in violent action. The data indicates that most offenders adopted their ideological beliefs years before ultimately carrying out their attack, with cases of more rapid radicalization appearing to be the exception. Categorizing individual lone offender terrorists by their ideologies alone can be somewhat reductive, considering the amount of variance between individuals who share ideological beliefs and the overlap that exists between personal grievances and ideologies. While some offenders adopted belief systems that mirrored existing ideological movements, others only adhered to pieces of an established movement or combined elements of different ideologies. Other offenders had ideologies that revolved around a specific grievance or set of The FBI defines mobilization as the process by which radicalized individuals take action to prepare for or engage in violence or material support for violence to advance their cause. 47 36 grievances instead of a more broadly-known belief system or movement. There were cases in which the source or rationale behind the offender’s ideology could be difficult to identify or appeared illogical. Violent action is rarely taken for a single reason and is more often the result of a combination of psychological and social issues impacting a subject. Instead of focusing solely on ideology as the driving force behind a trajectory toward targeted violence, threat assessors should examine all potential warning signs, risk factors, and triggers that indicate a subject may engage in violence. As such, threat assessors should work to gather and document available information concerning all aspects of a subject’s life, allowing for a thorough and structured analysis by a team trained in disciplines beyond just law enforcement. This combination of perspectives helps to avoid the potential collection or assessment biases inherent to any single theoretical approach or model. Models or actuarial tools can be useful for triage or case prioritization, but structured professional judgment is needed to evaluate each situation on a case-by-case basis. Propaganda /Media Consumption Ideological media of various forms appeared to be an influential part of offenders’ lives. Researchers noted any evidence of physical and electronic media, as well as references to media contained within offenders’ statements and writings. At least 40 offenders (77%) consumed radical ideological material or propaganda.48 Thirty-one offenders (60%) viewed printed materials, such as books, pamphlets, magazines, newsletters, or other types of publications. Nine offenders (17%) consumed propaganda through mediums such as CDs, tapes, movies, radio shows, and mail order speeches. While the study began to look at some online behaviors the researchers’ protocol did not contain a specific question about offenders’ access to the internet. However, the first instance of an offender using the internet to view propaganda was an offender who carried out an attack in 1999. Of the 40 attacks that occurred during or after 1999, 24 offenders (60%) were known to have viewed propaganda through social media sites and other online platforms such as blogs, forums, and websites. Propaganda was defined as information or ideas disseminated to promote a particular political cause or point of view and to influence the opinions, emotions, and/or behaviors of a targeted audience. 48 37 The expansion of the internet may have allowed offenders to obtain easy access to radical content and to individuals willing to distribute radical materials. However, the consumption of propaganda Online Platform occurred across all decades within Launch Year the study. Additionally, some printed materials continued to hold popularity. Of the 24 offenders Stormfront: 1996 known to have viewed online AOL IM: 1997 propaganda, 16 (67%) viewed propaganda through other Myspace: 2003 mediums as well. Facebook: 2004 While not always directly linked YouTube: 2005 with propaganda materials, Twitter: 2006 previous terrorist attacks or acts of violence appeared to interest or inspire a proportion of lone offenders. At least 16 offenders (31%) researched and/or referenced prior attacks carried out by other lone offenders or terrorist groups. Prior research into other attacks was unknown in 23 cases (44%). As many of the frequently mentioned prior attacks were highly publicized when they occurred, the actual percentage of offenders who were influenced by prior attacks (either through self-research or exposure to media) may be higher than the 31 percent reported in this study. Beyond prior attacks serving as a source of ideological inspiration or providing information useful to attack planning and preparation, the publicity that surrounded attacks may have been particularly salient to some offenders. Twenty-two offenders (42%) selected their target and/or chose to carry out an attack at least partially on the basis that they wanted to attract media attention. Propaganda/Media: Operational Considerations While material or propaganda that advocates or supports the use of violence to further an ideology can be clearly assessed as concerning, there may be less clarity when encountering material that may be biased or skewed but does not explicitly endorse violence. When encountering such material, it could be suggested that instead of centering focus upon the content itself, the material should be viewed through the lens of the individual viewing the content. Additionally, the content should be placed into context with other statements and behaviors exhibited by that individual. Within that individual’s frame 38 of reference, how do they interpret and respond to the content? Do they align non-violent radical ideological materials with additional materials that spread violent rhetoric? Materials that do not directly advocate violence may still be used by an individual to justify or support their own violent belief system or as a foundation for future violent beliefs. Operational experience has found that individuals can also be inspired by influential or charismatic leaders. Offenders sometimes expressed admiration for leaders of violent ideological groups or movements, and sometimes referenced past lone offender terrorists. Extensive media coverage of prior attacks appeared to influence some offenders, as almost half selected their target at least partially based upon the belief that their attack would attract media attention, either for themselves (e.g., personal notoriety or a platform to air personal grievances) or to gain attention for the causes they claimed to represent. As technology advances, increased media capabilities allow offenders to directly communicate with intended audiences. However, in a time where information is constantly flowing, operational experience has noted instances in which offenders felt the necessity to obtain a certain number of casualties or choose particularly vulnerable victims in order to gain the desired amount of media attention. In addition to providing an accessible outlet for offenders’ messaging, online platforms can provide a sense of community and engagement for individuals who otherwise feel marginalized or not wellrepresented offline or among more mainstream online platforms. Threat assessors should stay familiar and up to date with current social media and messaging platforms and applications. Public Narratives Fifty offenders (96%) produced writing or videos intended to be viewed by others. Of those 50, 24 offenders (48%) only produced content before their attack,49 four offenders (8%) only wrote publicly after their attack, and 22 offenders (44%) produced content both before and after their attack. Out of those who produced public narratives, 40 offenders wrote letters (80%) that described their grievances, ideologies, or the intent behind their actions. Letters were either sent to individuals, news Most of the offenders who only produced content before their attack died during or soon after the commission of their attack (n=17, 71%). 49 39 media, potential targets, or left somewhere where they would be found after the offender engaged in their attack. Twenty-four offenders (48%) posted videos or writings (e.g., blogs, essays, or manifestos) online to social media sites, forums, or self-made websites. Thirteen offenders (26%) wrote books (often self-published online) or position papers that were put into print. Books usually focused on ideologies, conspiracies, or offenders’ own personal narratives and memoirs. At least 16 offenders (32%) produced content through other mediums such as audio or video tapes, music, unpublished books, or documents and journals intended to be found after their attack. The various themes present in offenders’ public writings or videos are highlighted below. Writing Themes (n=50) Theme n % Reflection on ideology 43 86% Proselytizing 36 72% Personal narrative and/or experiences 31 62% Claiming responsibility or explaining reasoning for attack 27 54% Warning regarding planned or imminent attack 10 20% At least 13 offenders (26%) sent material directly to news media. Materials included letters to the editor that discussed offenders’ ideologies, grievances, or another topic of concern; criticisms of news statements or articles; advertisements directed to likeminded individuals; and manifestos or explanations sent soon before or after an attack. Public Writing: Operational Considerations Ideological messaging is salient to the construct of terrorism, as a lack of social or political motive precluded other targeted violence offenders from being entered into the study. Therefore, messaging was frequently a key component of offenders’ attack planning and target selection. Some offenders verbalized their messaging, but most offenders also created writings or videos in order to publicly 40 memorialize their grievances, ideologies, and the rationales for their attacks. Manifestos and letters relating specifically to the attack were often posted or sent shortly before the attack, brought to the attack site, or sent after the attack had occurred. While some offenders’ pre-attack writings and postings addressed grievances and ideologies without explicitly discussing violence, other offenders posted public writings and videos that endorsed violence in the weeks, months, and years before their attack. 41 Part Two: The Attacks Targets Of the 52 lone offender terrorists, 33 carried out attacks that resulted in fatalities.50 A total of 258 victims were killed and 982 victims were injured. Primary targets included:51          Federal government facilities/personnel (n=9, 17%) Law enforcement personnel (n=8, 15%)52 Religious centers/personnel (n=6, 12%) Medical facilities/personnel (n=5, 10%) Private individuals (n=4, 8%) Educational facilities/students/faculty (n=4, 8%) Other businesses (n=3, 6%) Specific organizations or groups (n=3, 6%) Other targets (n= 10, 19%) Additional Acts of Terrorism The analyses presented in this report are primarily based upon each offender’s index attack, meaning the first act of terrorism that met the project’s inclusion criteria, which stated that the action was an attempted or completed act of lethal violence and that the offender was the primary actor and primary architect driving that action. Fourteen offenders (27%) committed at least one additional act of terrorism.53 Among this subset, 61 additional terrorism acts were Offender deaths were not counted in this analysis. Does not sum to 100 due to rounding. 52 An additional 10 offenders attacked law enforcement or armed security in reaction to first responders to the scene. 53 An additional act of terrorism was considered any attempted or completed violent criminal act in furtherance of a social, political, and/or ideological goal. 50 51 42 carried out before54 and/or after the offenders’ index attacks. In total, 30 additional victims were killed and 66 were injured. Of the 14 offenders:  13 offenders (93%) were ultimately arrested. One offender (7%) was killed by law enforcement who responded to the offender’s final attack site.  Seven offenders (50%) attempted or committed one additional terrorism attack. The other seven offenders attempted or committed between three and 18 additional attacks.  Five offenders (36%) targeted sites or individuals for ideological reasons that were different than the ideology motivating their index attack.  Four offenders (29%) carried out additional shootings and/or bombings that accounted for the total number of additional victims killed or injured. Planning and Preparation Attack planning is a process that represents a progression from the initial idea of carrying out an attack to the actual decision to engage in violent action. Planning for an attack involves multiple smaller decisions (e.g., selecting a target and choosing an attack method) and is more difficult to observe than preparation, which involves the physical steps and actions taken to follow through on planning and carrying out an attack. In terms of target research and selection:  38 offenders (73%) selected their target because it was instrumental to their goal or ideology (e.g., targeting a clinic to stop abortions or targeting a racial or religious group seen as a threat).  Of those 38 offenders, 18 (47%) also selected targets based upon some degree of symbolic value.55 Incidents that occurred before the offender’s index attack were not counted as the index because they did not meet all criteria mentioned above. 55 Instrumental and symbolic goals were not always mutually exclusive. For instance, an offender who attacked a military location viewed the target as symbolic of the broader US military actions that were at the center of his grievance, but also viewed the attack as a way to directly harm military assets. 54 43  22 offenders (42%) selected their target and/or chose to carry out an attack at least partially on the basis that they wanted to attract media attention.  21 offenders (40%) considered other targets before selecting their primary target.56  16 offenders (31%) had The following planning and preparation activities were multiple targets planned for only seen in a small number their index attack. of cases:  Of those 16 offenders,  Falsifying identifying 10 (63%) ultimately documents (n=7, attacked multiple 14%) targets as part of their  Taking index attack.57 photos/videos of offense-related  13 offenders (25%) targeted information during at least one specific planning (n=6, 12%) individual (known to or by  Using protective the offender) in their attack.  In eight of these 13 cases (62%), other individuals besides the specific target(s) were killed and/or injured.58  tactical gear during the attack (n=6, 12%) Altering/disguising appearance to avoid detection (n=4, 8%) During the time in which offenders were planning and preparing for their offense(s), nine offenders (17%) were stopped, detained, and/or examined by law enforcement. Three offenders had contact with law enforcement the morning of their offense (two for traffic infractions and one for protest actions that had resulted in a complaint call). Ultimately, five offenders (10%) engaged in an attack that was reactive, spontaneous, or opportunistic in nature. These cases were included because the offenders attacked targets that were central to their grievances or ideologies, consistent with prior statements This is likely an underestimate, as consideration of other targets was unknown in 19 cases (37%). Multiple targets were counted when an offender moved immediately from their first attack site to a second attack site or left their first attack site in search of additional targets. Serial attacks involved a “cooling off” period after or before the index attack and were counted separately. 58 These eight cases included situations in which other individuals were harmed intentionally (e.g., the offender shot at additional random targets or the offender opened fire on first responders) or unintentionally (e.g., nearby individuals were caught in the gunfire or security personnel responded to a suspicious package). 56 57 44 endorsing violence. Four of these offenders opened fire on law enforcement officers after contact with the officers occurred, while one offender selected and attacked his target after driving around and scoping out other possible targets. Target Security In most cases, offenders attacked targets that had no or minimal security (n=41, 79%). No security usually referred to a target located in an open space or a building without protective measures. However, in some cases where target security was present, the offender carried out their attack outside the target location and therefore did not attempt to overcome security measures. Minimal security included security measures that may provide a degree of deterrence but did not present a substantial physical barrier between the offender and the target (e.g., surveillance/video cameras). In the 36 cases where information about offenders’ decision-making processes were available, 18 offenders (50%) selected their targets based at least partially on the target’s ease of access. Figure 8 100% 80% 60% Highest Level of Target Security* (n=47) 52% 40% 27% 15% 20% 0% No Security Minimal Security Physical Security * Chart does not include four cases (8%) in which the targets approached the offender or one case (2%) in which the target’s security level was unclear. Percentages add to over 100 due to cases in which multiple locations were targeted or more than one level of security was present. 45 Highest Level of Target Security (n=47) Level of Target Security No Security 52% Minimal Security 27% Physical Security 15% There were eight cases (15%) in which offenders targeted a location with physical security measures, such as security checkpoints/gates, security personnel, or electronically-secured doors. Eighteen (35%) offenders either surveilled their target in advance or otherwise frequented the target location.59 Surveillance was measured in cases where clear evidence or disclosure on the part of the offender indicated the offender had previously visited their target for the purpose of collecting target information that would assist in the planning and preparation for their attack. In other cases, offenders were already familiar with the targets or target areas if they frequented the locations as part of their normal routines (e.g., the target was located at or near the offender’s school, work, or home) or visited the locations for another purpose before planning their attacks. Only three offenders (6%) were known to have tested security measures before carrying out their attacks. Four offenders (8%) were known to have physically followed targeted individuals. Offender Planning: Operational Considerations In assessing the process by which lone offender terrorists planned and selected their targets, it is crucial to consider the offenders’ ultimate attack objectives. Some offenders based their targeting on locations, attacking anyone present at the site. Other offenders were more focused on specific individuals or types of individuals, and either avoided additional victims or did not care if they harmed individuals viewed as collateral casualties. Some offenders chose targets that were instrumental to their goals, expecting that violence would directly solve all or part of the perceived issue. Other targets were more symbolic in nature, involving locations or victims perceived as representative of a larger target or issue. Existing research highlights the rational processes used when an individual plans an act of terrorism and the cost-benefit analysis In 19 cases (37%), it was unknown or unclear whether the offender had surveilled or frequented the target location before their attack. 59 46 involved in selecting a target.60 Both research and operational experience emphasize how the presence of clearly visible security measures can sometimes serve as an effective deterrent, causing would-be offenders to perceive an increased or unacceptable level of risk associated with a hardened target (i.e., a target with physical security measures). For offenders who intend to leave the scene of their attack unidentified, visible security cameras and a high presence of personnel or potential witnesses may be a deterrence if the offender evaluates those risks when selecting their target. Yet such security may have little impact on offenders who plan their attacks with the intention of dying in the attack or who see post-attack arrest as an opportunity to publicize their message. Target hardening may dissuade an offender from selecting a target if they perceive those security measures as likely to prevent them from achieving their ultimate goal. However, an offender may still attack the perimeter of a secure target if such action is perceived as sufficient. Thus, the process of target selection involves individualized decision-making as to what level of security is acceptable in order to achieve the desired outcome. Keeping in mind that this study only examined those offenders who ultimately carried out an attack, we cannot conclude how many would-be offenders may have been deterred by target hardening. Attack Method Attacks using firearms were the most common type of attack among offenders (n=35, 67%), followed by attacks that used explosives (n=14, 27%). The chart below shows offenders’ primary attack methods.61,62 Gill, P., Marchment, Z., Corner, E., & Bouhana, N. (2018). Terrorist decision making in the context of risk, attack planning, and attack commission. Studies in Conflict & Terrorism, 1-16. doi:10.1080/1057610X.2018.1445501 61 In addition to their primary attack weapon, one offender used an explosive device, one offender used a firearm, one offender used a bladed instrument, and two offenders used arson or an incendiary device. 62 Even if not detonated successfully, explosives were still included as present and used if the device was brought to the scene and the offender either attempted to detonate the explosive or the explosive had the potential to detonate without initiation action on the part of the offender (e.g., the device detonated on a timer or due to unwitting action by a victim). 60 47 Figure 9 Primary Attack Method (N=52) 65% 25% EXPLOSIVE(S) FIREARM(S) 4% 6% BLADED INSTRUMENT AIRPLANE/VEHICLE Primary Attack Method (N=52) Attack Method Explosive(s) 25% Firearm(s) 65% Bladed Instrument 4% Airplane/Vehicle 6% Out of the 14 serial offenders who carried out additional acts of terrorism, nine offenders used explosives (64%), five used firearms (36%), and four used arson/incendiary attacks (29%).63 Firearms Of the 35 offenders who used a firearm in their index attack, 27 offenders (77%) used a handgun, 14 offenders (40%) used a rifle, and seven offenders (20%) used a shotgun. Seventeen shooters (49%) brought multiple firearms to their attack site. Slightly more than a third (n=12, 34%) had formal training or experience (e.g. military or law enforcement) in using firearms. An additional 18 (51%) had informal experience with firearms (e.g., practicing at a gun range). Percentage adds to above 100 percent because some serial offenders who carried out multiple attacks used more than one attack method. 63 48 Out of the 26 cases in which information on firearm acquisition was known:64     18 (69%) legally purchased their weapon(s) 5 (19%) illegally purchased their weapon(s) 4 (15%) borrowed or were given their weapon(s) 3 (12%) stole their weapon(s) Overall, most offenders had previously owned firearms (n=45, 87%). Two offenders had attempted to obtain a firearm for their attack, but were unable and ultimately chose a different attack method. Explosives Of the 14 offenders who used explosives in their index attack, 11 offenders (79%) personally brought explosives to the target location. Four offenders (29%) used a third party to deliver explosives: two offenders delivered explosive devices through the mail, one offender used a family member to deliver the explosive device, and one offender brought the explosive device to a location where it was unwittingly transported by a bystander.65 At least 31 offenders (60%) attempted to learn how to assemble explosives and/or tried to obtain explosive materials. This rate is higher than the 16 offenders (31%) who ultimately used explosives in their index and/or serial attack(s). Researchers observed that some offenders explicitly stated that they chose not to explore or pursue building explosives because they felt they lacked the capability. Attack Method: Operational Considerations One explosive attack resulted in more fatalities than all the other attacks combined. If omitting this outlier from the analysis, the results indicate shooting attacks had a higher number of fatalities compared to all other attacks (p<0.05).66 This is likely explained by attacks that did not result in fatalities: five shooting attacks (n=34, Information refers specifically to the firearms used in the offenders’ attacks. One offender was coded as both personally delivering and using a third party to deliver the explosive device. 66 Shooting attacks averaged 2.47 deaths (sd=2.59), while all other attacks averaged 0.38 deaths (sd=0.88). 64 65 49 15%)67 resulted in no victim fatalities, while 14 attacks that used other methods (n=17, 82%) resulted in no victim fatalities. The finding is consistent with prior observations regarding targeted violence offenders’ use of firearms and lethality, particularly in cases where an offender targets multiple individuals.68 Offender capability (i.e., the level of knowledge/skills possessed or needed to use the weapons chosen and the offender’s ability to access the materials or weapons) is an important component for threat assessors to consider. When investigating a person of concern, opportunities for observation often exist while offenders attempt to develop the capabilities needed to carry out their desired attacks. Operational experience, supported by this research, indicates offenders consider or pursue multiple attack methods when planning and preparing for their attacks. For example, not all offenders who attempted to learn about or assemble explosives ultimately used explosives in their attack. Similar to target selection, offenders choose their attack method(s) based upon a combination of factors, such as capability, opportunity, risk of discovery, and desired outcome. Therefore, threat assessors should consider multiple behaviors within context over time. Offender Outcome Thirty-two offenders (62%) were ultimately arrested after carrying out their index attack. Of those who were arrested, 12 offenders (38%) were detained at their attack site: six offenders were either shot by law enforcement or physically subdued by victims and/or witnesses, four offenders surrendered or were arrested without further incident, and two offenders surrendered after a hostage or standoff situation. Of the 20 offenders who left the site of their attack, five offenders (25%) were arrested the same day. The remaining 15 offenders were arrested days later (n=5; 25%), weeks or months later (n=5; 25%), or Not included as a shooting attack was an attack in which the offender’s secondary weapon was a firearm, which he used to threaten victims, but did not discharge during his attack. 68 Amman, M., Bowlin, M., Buckles, L., Burton, K., Brunell, K., Gibson, K., Griffin, S., Kennedy, K., & Robins, C. (2017). Making prevention a reality: Identifying, assessing, and managing the threat of targeted attacks. Washington DC: Federal Bureau of Investigation, National Center for the Analysis of Violent Crime, Behavioral Threat Assessment Center. 67 50 years later (n=5; 25%). Three offenders turned themselves in and two offenders were pursued from their attack site and arrested. Figure 10 Offender Outcome (N=52) 62% 23% 15% COMMITTED SUICIDE AT ATTACK SITE KILLED BY RESPONDING LAW ENFORCEMENT ARRESTED Offender Outcome (N = 52) Outcome Committed Suicide at Attack Site 15% Killed by Responding Law Enforcement 23% Arrested 62% Twenty offenders (38%) died during or in the immediate aftermath of their attack.69 Twelve offenders (23%) were killed by responding law enforcement and eight offenders (15%) committed suicide, either upon confrontation with responding law enforcement (n=6) or by the nature of their attack method (n=2). The 38 percent includes one serial offender who was killed by law enforcement at his final attack site. Not included are the two offenders who were arrested, but later committed suicide while incarcerated. 69 51 Part Three: The Bystander Lone offender terrorists were rarely completely isolated from other individuals. Offenders interacted with family members, peers, and strangers across a variety of social contexts, both online and offline. While the term bystanders traditionally refers to individuals who witness a specific event, counterterrorism and threat assessment professionals sometimes use an expanded definition of bystanders,70 referring to individuals who may witness a range of concerning preattack behaviors or statements. Bystander Categories     Family: Individuals such as parents, spouses, stepparents, siblings, aunts, and uncles. Peers: Individuals such as coworkers, classmates, close friends, and acquaintances. Authority Figures: Individuals such as religious leaders, professors, employers, mental health professionals, and law enforcement officers. Strangers: Individuals such as persons from commercial establishments who sold precursor materials or weapons, or individuals who witnessed offender behavior in online or offline public spaces. Individuals who knowingly provided assistance to the offender during the attack planning, preparation, or action were not considered bystanders. Unfortunately, bystanders may fail to observe the full context of offenders’ behaviors and may not realize the full extent of a potential offender’s threat level. The FBI defines bystanders as individuals whose relationship and level of interaction with a person of concern enables them to witness or become aware of activities or behavior that may indicate radicalization or mobilization to violence. 70 52 Family and Social Relationships Although the study was unable to assess with precision how frequently offenders interacted with others, at least 37 offenders (71%) either described or were described by others as having friendships or regular interactions with one or more friends or peers. At least 30 offenders (58%) interacted with friends or peers online. In more than half of the cases studied (n=29, 56%), offenders had contact with at least one other family member or associate who was sympathetic to radical or violent ideological views.  50% of offenders (n=26) had at least one family member or associate who appeared to be sympathetic to a nonviolent extremist ideology. While not always apparent if individuals around the offender clearly supported violence in service of an ideology, it was not unusual to see support for ideologically radical ideas and movements. For instance, family members or associates may have supported anti-government or racist movements, while not outwardly voicing or evidencing support for the use of violence. The ideology of family members or associates may or may not have been the same ideology held by the offender.  35% of offenders (n=18) had at least one family member or associate who adhered to a violent ideology. More than a third of offenders associated with individuals who believed the use of violence in furtherance of an ideology was justified. These individuals were sometimes supportive of or affiliated with violent ideological groups or organizations. Living Situation Approximately half of the offenders (n= 27, 52%) lived with at least one other person at the time of their attack, usually with a spouse, parent(s), and/or other family member(s). 53 Figure 11 100% Offenders' Co-habitants* (N=52) 80% 60% 40% 39% 39% 19% 20% 10% 4% 0% No One Spouse/Partner Family Member(s) Roommate(s) Other** *Total percentage adds to more than 100 due to cases in which the offender lived with multiple people. **“Other” included transient co-habitant relationships and other undefined relationships (e.g., a family member’s significant other). Offenders’ Co-habitants (N=52) Co-habitants No One 39% Spouse/Partner 19% Family Member(s) 39% Roommate 10% Other 4% 54 Family and Social Relationships: Operational Considerations Most offenders had some level of contact with family members, peers, or associates, even if the frequency and quality of those relationships fluctuated over time. While those who lived or otherwise interacted with the offender were not always fully aware of offenders’ ideologies or grievances, it is notable that other individuals were potentially in a position to observe offenders’ behaviors and general functioning before their attacks. Considering the number cases in which one or more family members or peers were sympathetic to extremist views, it is important to consider what influence or impact, if any, such views may have had on offenders. Operational observations have noted that the presence of a support structure (e.g., a significant other or close family member) usually mitigates risk. However, such relationships may in fact be risk-enhancing if the significant other or family member supports or sympathizes with a radical ideology, particularly one that justifies the use of violence. Bystanders are often best-positioned to spot indicators of radicalization, often prior to law enforcement’s awareness or authority to investigate. However, threat assessors should consider the impact a bystander’s own ideological viewpoints may have on their ability to identify concerning behavior and their willingness to directly or indirectly intervene. Group Affiliations Twenty-three offenders (44%) had previously claimed group membership or affiliation with an ideological movement, or associated with its group members. Within the year before the attack, 13 offenders (25%) were involved or associated with ideological groups that actively or historically supported or engaged in violence. An additional 10 offenders (19%) had some level of involvement with ideological groups that were either non-violent or had no clear links to violence. Rejection Many offenders had previously experienced rejection by other individuals, groups, or intuitions. Researchers recorded any rejections 55 that were noted across the offenders’ lifespans. Therefore, rejections were not necessarily proximate to offenders' attacks. 37 offenders (71%) were at some point rejected by at least one individual. This included rejection by family members, romantic partners, friends, or other social acquaintances. 21 offenders (40%) were at some point marginalized or rejected by a group or organization. This included rejection by social groups, professions, or religious institutions for a variety of reasons. Only four offenders (8%) were known to have been rejected by an extreme ideological group. Rejection: Operational Considerations While rejection itself may not be unusual, more widespread rejection across multiple areas of an offender’s life may give insight into the way an offender interacts with others in their family and social network. Additionally, the way in which an offender perceives and reacts to rejection can be informative – either providing an indicator of resilience or stress. Even more noteworthy are situations in which an offender seems to fixate on or internalize one or multiple rejections as part of a larger grievance. Offender Isolation: A Closer Look Most individuals have at least some level of interaction with others and are not completely isolated from outside influence. This lack of isolation becomes more apparent in the age of modern technology, where individuals can explore any specific interest online and can find likeminded individuals with relative ease. To address the discussion over degrees of influence and what constitutes a “lone” offender, researchers Borum, Fein, and Vossekuil (2012) make a case for a dimensional approach to understanding the phenomenon of “loneness.” The current project followed this lead by conceptualizing isolation as existing along a continuum. Researchers examined offenders’ isolation in their personal lives to develop a scale that further describes the lone offender terrorists studied. The Personal Isolation scale describes the presence of other individuals in an offender’s family and social network, and any 56 interactions or affiliations the offender had with social, political, or ideological groups. Coders used information about contacts and communications that occurred in both online and offline settings.71 Personal Isolation Personal Isolation scores were derived from six considerations72 addressing each offender’s interpersonal relationships and group affiliations. On a scale from zero to six, with a higher score indicating a higher degree of isolation, the average Personal Isolation score was 2.65. The distribution of scores in Figure 12 indicates that most of the lone offenders in the study sample were not fully isolated from other individuals or groups in the year before their attacks.73 Figure 12 Personal Isolation Scores (N=52)* 23% 21% 17% 14% 10% 0 10% 1 2 3 Isolation 4 5 6% 6 Mean=2.65, SD=1.68 *Does not sum to 100 due to rounding. The scales are not validated, but provide another way to describe the study sample of lone offender terrorists and provide future opportunities to explore comparisons with other types of terrorists or targeted violence offenders. 72 Each consideration was based upon variables pulled directly from the coding protocol, with one point assigned to each consideration. Items are detailed in Appendix A. 73 While the score distribution provides some insight into our subject pool, it should be noted that without using the scale to score a control group, we do not know how or if these scores would differ from the general population or other types of offenders. 71 57 Personal Isolation Scores (N=52) Isolation Scores 0 10% 1 21% 2 17% 3 23% 4 14% 5 10% 6 6% Five offenders ranked at the lowest end of the scale, meaning they were not considered isolated on any of the scale items. These five offenders had some level of prior affiliation, involvement, or contact with groups that shared their ideological beliefs (either violent or nonviolent), and at least four out of those five offenders initially adopted their extremist ideas due to influence from other individuals.74 All five offenders were married or in a relationship, had a close or somewhat close relationship with at least one family member, and had one or more individuals in their social network. Only three offenders received the highest isolated Personal Isolation score. Despite being highly isolated around the time of their attacks, all three offenders had a long history of concerning behavior Personal Isolation Items: and hostile relationships with family members, in addition  Prior organizational affiliation to issues or conflicts with  Living situation former friends, coworkers, or  Relationship status neighbors. These case  Family relationships examples suggest that even  Social networks the most “lone” offenders may  Self-radicalization have had past interactions with others where concerning behavior or statements were One offender had contact with others who also espoused extremist ideas. Based upon available information, it was unclear whether the offender’s initial adoption of his extremist ideas was due to selfgenerated beliefs or influence from others. 74 58 observed, potentially providing opportunities for earlier identification and intervention. Personal Isolation: Operational Considerations The Personal Isolation scorings are based upon offenders’ interactions in the year before their index attacks, offering a snapshot in time. As personal and social factors are dynamic, even offenders who appear socially isolated at the time of their attack often have past associates and family members who can provide historical information. Associates from beyond an offender’s current or most recent social circle can also provide further context critical to understanding an offender’s motivations, behaviors, and life stressors. In scoring social and family “closeness,” researchers made judgments based upon the amount of contact other family members and peers had with the offender and how much they knew about the offender.75 While the scoring on scale items did not directly provide information about whether offenders’ relationships and contacts were positive, the scoring did indicate which offenders were less isolated in their daily personal lives. Additionally, individuals who knew the offender for any extended period may have been more likely to recognize changes from the offender’s normal behavior or social interactions. What Did Bystanders Observe Bystanders were exposed to a range of concerning behaviors and statements made by offenders before their attacks, not all of which were directly related to an offender’s ideology or attack planning. For instance, family members may have been concerned about an offender’s alcohol abuse but may have had limited knowledge of the offender’s violent extremist ideology. Awareness of Stressors Stressors are circumstances, situations, or events that put a strain upon an individual. The researchers examined various Individual scores were based upon a five-point Likert scale. Scores that differed by two points or less were averaged. Any larger differences were discussed further in order to reach agreement upon a score. 75 59 categorizations of stress and noted cases in which others identified potential stressors experienced by the offenders before their attacks.76 Figure 14 Stressors Noted by Others* (Year Before the Attack) (N=52) 100% 80% 60% 48% 44% 40% 27% 20% 29% 44% 35% 19% 0% *Additional variables included loss of interpersonal relationship (10%), academic strain (6%), and other (e.g., isolation/rejection or conflict with peers/neighbors/property managers) (15%). Stressors Noted by Others in the Year Before Attack (N = 52) Stressor Financial 48% Physical Injury/Illness 19% Partner/Marital Stressor 27% Criminal/Legal 29% Family 44% Employment 35% Mental Health 44% As the data was constrained to stressors that other individuals were aware of, the reported data are likely underestimates of all stressors experienced by the offenders. 76 60 In most cases (n=47, 90%), bystanders were aware of at least one stressor the offender experienced. Usually, two or more stressors were noted (n=38, 73%). Awareness of Concerning Behaviors In every case, concern was expressed over at least one of the offenders’ behaviors at some point in time before their attack. In most cases (n=38, 73%), concerning behavioral changes were observed in the year before their attack. A change in behavior was noted when an existing behavior escalated in frequency, intensity, or concern, or when a new behavior emerged. The overall occurrences of concerning behaviors may have been considerably higher than what is captured in the current study. Coding depended on concrete behaviors that were both observable to bystanders and recorded in case records. For instance, if an offender had violent thoughts, but never expressed those thoughts to others, the concerning thinking was not reflected in Figure 15. Similarly, issues with mood, substance use, or sleep may not have been observable or noted by others. In cases where behavioral changes were observed, it is also possible that behaviors may not have presented in a way that caused others to be concerned. Additionally, not all bystander observations are captured in official records. 61 Figure 15 Concerns Expressed Over Behavior* (N=52) 85% Interpersonal Interactions 52% 83% Anger/Aggression 21% 75% Mood 40% 71% Thinking/Communication 38% 42% Alcohol/Drug Use 4% 40% Work/Academic Performance 10% 33% Impulsivity/Recklessness 8% 17% Weight/Eating 4% 17% Sleep 13% 15% Hygiene/Appearance 12% 0% 20% 40% 60% 80% 100% Concern over Behavior (Any Point Prior to Attack) Concern over Behavior Emerged or Escalated in Year Prior *Concerns over sexual behaviors, physical health, and “other” behaviors (e.g., spending patterns) not included on the chart were observed at lower rates. 62 Concerns Expressed Over Behavior Any Point Prior to the Attack (N=52) Concerns Interpersonal Interactions 85% Anger/Aggression 83% Mood 75% Thinking/Communication 71% Alcohol/Drug Use 42% Work/Academic Performance 40% Impulsivity/Recklessness 33% Weight/Eating 17% Sleep 17% Hygiene/Appearance 15% Concerns Expressed Over Behavior Emerged or Escalated in Prior Year (N=52) Concerns Interpersonal Interactions 52% Anger/Aggression 21% Mood 40% Thinking/Communication 38% Alcohol/Drug Use 4% Work/Academic Performance 10% Impulsivity/Recklessness 8% Weight/Eating 4% Sleep 13% Hygiene/Appearance 12% 63 In most cases (n=43, 83%), bystanders were concerned at some point with offenders’ anger or aggression. This finding is unsurprising when considering that more than half of the offenders had previously carried out battery or physical violence (n=30, 58%).77 Bystanders were also frequently troubled by offenders’ interpersonal interactions (n=44, 85%). Some interactions included the aggressive behavior described above, such as conflicts, fights, and other confrontations. However, other behaviors observed by family or peers included social withdrawal, or trouble obtaining or maintaining friendships and relationships. While some offenders had a long history of such issues, there were cases in which family and peers noted a concerning change when the offender became increasingly socially withdrawn. Issues with interpersonal interactions sometimes overlapped with concerns over offenders’ thinking and communication (n=37, 71%), which included a wide range of issues. Concerns sometimes related to statements indicative of a potential mental health issue, such as depression or suicidal ideation. Similarly, some concerns centered on the thought processes of the offender’s communication, such as illogical thinking or incoherent speech. Other times, bystanders were worried about the content or intensity of the offender’s ideas or beliefs, such as statements that were viewed as extreme, threatening, or supportive of violence. Using a separate item on the research protocol, researchers found that in at least 41 cases (79%), offenders expressed the belief that violence was justifiable in service of a cause or objective Based upon observable information in the case files (e.g., offender behaviors, writings and online postings, or statements from bystanders) many offenders exhibited fixations or obsessions (n=36, 69%), seen when offenders spent an unusual amount of time persistently seeking, sharing or discussing information relating to a specific topic, grievance, or ideology. prior to their attacks. As this rate is higher than the number of cases where bystanders expressed concern over the offender’s thinking or communication, it is possible that such statements did not elicit concern if viewed or heard by a limited number of people. Another Offenders who had been arrested for a violent offense and/or carried out violence not reflected in their arrest histories. 77 64 possible explanation is that statements supportive of violence may not have been perceived as alarming if observed by those who did not take the offender seriously or those who were sympathetic to the offender’s ideas. Also common were concerns over offenders’ moods (n=39, 75%). Of these cases, there were 10 cases in which the offender was diagnosed with a mood disorder and 16 cases in which bystanders suspected the offender of having a mood disorder. Mood concerns co-occurred with concerns over offenders’ substance use in 15 cases (38%). In 21 cases (40%), bystanders were worried about offenders’ work and/or academic performance. Performance issues included a lack of productivity or delinquent activity, as well as the consequences of performance issues, such as poor evaluations, suspensions, or terminations. Also included were situations in which offenders quit school or work, or had general difficulties obtaining or maintaining employment. In a third of the cases (n=17, 33%), bystanders were troubled by offenders’ impulsivity or recklessness, which included behaviors such as engaging in risky activities, sudden travel, or erratic spending. Although not always noted by bystanders as significant at the time, at least 17 offenders (33%) gave away possessions before their attacks, presumably in preparation for death or arrest. Stressors and Concerning Behaviors: Operational Considerations Factors such as the level of contact bystanders had with the offender, the type of relationship they shared, and the frequency or intensity of the offender’s behavior influenced the ways in which bystanders contextualized observations. Bystanders were sometimes unconcerned with behaviors that occurred only occasionally or in certain situations. At the same time, longstanding issues could also contribute to low levels of concern over behavioral issues. For instance, there were cases in which offenders expressed thoughts endorsing the use of violence; however, bystanders did not appear to take them seriously because the offenders had frequently expressed those views in the past, but had never acted on them. This sentiment 65 is aptly articulated by one bystander’s description of an offender: “[he] said a lot of things he would never do and did a lot of things he never said.” It is possible that in cases where an individual has a longstanding history of espousing extreme views, bystanders may have become desensitized to some extent. Instead of focusing on specific stressors, threat assessors should consider an individual’s circumstances, perceptions, and ability to cope with stressful situations. While coders captured bystander concerns that were specific, some bystanders only noted a vague oddness or general change in baseline behavior. Therefore, it is important to ask bystanders open-ended initial and follow-up questions about concerns and observations, instead of only inquiring about specific behaviors.78 Awareness of Offender’s Ideology and/or Attack Plans In almost all cases (n=48, 92%), at least one person knew of the offender’s non-violent ideology or grievance. In more than half of cases (n=32, 62%), at least one person knew the offender was supportive of violence in furtherance of an ideology. In 13 cases (25%), at least one other individual became aware of the offenders’ research, planning, or preparation for their attack. Beyond the awareness of general planning or preparation, there were nine cases (18%) in which at least one other individual became aware of the offenders’ specific attack plans. Figure 16 displays bystanders’ levels of awareness, with bystanders separated into categories based upon their relationship to the offenders. Considerable empirical evidence has supported the use of open-ended questioning in forensic contexts (e.g., Fisher & Schreiber, 2017). 78 66 Figure 16 Bystanders' Awareness of Offenders' Ideologies 100% 78% 80% 83% 78% 74% 69% 28% 60% 58% 56% 46% 42% 40% 35% 8% 25% 23% 23% 11% 19% 20% 10% 0% 4% 0% 6% 2% 2% 0% Non-Violent Ideology / Grievance Spouse/Partner* Violent Ideology Family* Research & Planning Specific Attack Plans Mentor* Peer Stranger No One *Percentages were out of cases where such a relationship was present before the attack. (Spouse/partner, n=13; family, n=50, mentor, n=9.) 67 Bystanders’ Awareness of Offenders’ Ideologies - Spouse (n=13) Non-Violent Ideology/Grievance 69% Violent Ideology 46% Research & Planning 23% Specific Attack Plans 23% Bystanders’ Awareness of Offenders’ Ideologies - Family (n=50) Non-Violent Ideology/Grievance 78% Violent Ideology 28% Research & Planning 6% Specific Attack Plans 4% Bystanders’ Awareness of Offenders’ Ideologies - Mentor (n=9) Non-Violent Ideology/Grievance 78% Violent Ideology 56% Research & Planning 0% Specific Attack Plans 11% Bystanders’ Awareness of Offenders’ Ideologies - Peer (N=52) Non-Violent Ideology/Grievance 83% Violent Ideology 42% Research & Planning 19% Specific Attack Plans 10% 68 Bystanders’ Awareness of Offenders’ Ideologies – Stranger (N=52) Non-Violent Ideology/Grievance 35% Violent Ideology 8% Research & Planning 2% Specific Attack Plans 0% Bystanders’ Awareness of Offenders’ Ideologies – No One (N=52) Non-Violent Ideology/Grievance 2% Violent Ideology 25% Research & Planning 58% Specific Attack Plans 74% For awareness of offenders’ research and planning, the study defined awareness as when a bystander noted an offender was making statements or engaging in activities that aroused their suspicion or appeared to be linked to criminal or violent intent. Offenders engaged in planning or preparation activities that may have been observed but were appreciated only in hindsight. Certain statements or activities may have appeared innocuous in the absence of any concerning context. For example, an offender may have asked questions about a public site, but may not have done so in a suspicious manner, or an offender may have purchased a firearm, but may not have done so in a way that indicated any intent to use the firearm for a criminal purpose. Concerning Statements Offenders were often vocal about their ideological beliefs prior to their attacks. More than half had used a public platform (e.g., protests, radio shows, presentations, workshops, written materials, online forums, or blog posts) to promote their grievances or violent extremist beliefs (n=27, 52%), and many engaged in ideologically-driven speeches or tirades that reportedly intruded into regular 69 conversations (n=29, 56%). Such speeches or tangents may not have been violent or concerning in nature but were often described by others as being a nuisance. Only six offenders (12%) communicated a direct threat, defined within this study as an unambiguous statement of intent to do harm, to their target before committing their attack. Many offenders had previously expressed the belief that violence was justifiable in service of a cause or objective (n=41, 79%). Researchers observed that such statements may have elicited concern from bystanders if the statements were considered unusual or particularly violent. However, violent or extreme statements may have attracted less notice if the offender’s audience shared some of the same beliefs or if those who observed the statements did not take the offender seriously. Awareness of Offender Ideology, Planning, and Concerning Statements: Operational Considerations In many cases, individuals were aware of offenders’ grievances and ideological beliefs. This is unsurprising considering the public platforms offenders often used to discuss and share their opinions and beliefs. When threat assessors become aware of concerns over an individual’s extreme views, they are tasked with assessing additional statements or behaviors that may be indicative of a potential threat of violence. In considering potential threat cases, threat assessors should look beyond the presence or absence of direct threats. There are cases in which communicated threats are followed by acts of violence. However, operational experience has found that those who intend to engage in violence will often avoid making specific or direct threats, as doing so would compromise their ability to carry out their plans. Threat assessment research has also concluded that acts of premeditated violence are not always preceded by direct threats.79 Meloy, J.R., Sheridan, L., & Hoffman, J. (Eds.) (2008). Stalking, threatening and attacking public figures: A psychological and behavioral analysis. New York, NY: Oxford University Press.); Fein, R., Vossekuil, B., & Holden, G. (1995). Threat assessment: An approach to prevent targeted violence. Washington, DC: U.S. Department of Justice, Office of Justice Programs, National Institute of Justice, Publication NCJ 155000. 79 70 Actions Taken by Other Individuals In at least 36 cases (69%), one or more individuals took some action to address one or multiple concerning behaviors they had observed. Often, this took the form of expressing concern directly to the offender and/or expressing concern to friends and family members. Depending on the case, these concerns may have been completely unrelated to the offender’s ideology or any attack planning and preparation. For example, family members may have confronted the offender about declining school performance or aggressive interactions. Figure 17 How Bystanders Addressed Concerns* (N=52) 54% 40% 33% 19% To Offender To Family To Friends Directly 2% To Anonymous Authority Tip Figure 4% 23% Other No Action Taken / No One Concerned *Percentages sum to more than 100 percent, as bystanders may have addressed concerns to multiple people or multiple bystanders may have expressed concerns about one offender. How Bystanders Addressed Concerns (N=52) To Offender Directly 54% To Family 40% To Friends 19% To Authority Figure 33% Anonymous Tip 2% Other 4% No Action Taken/No One Concerned 71 23% There were 17 cases (33%) in which bystanders expressed concern to one or more community authority figures at some point during the offender’s life. Authority figures included law enforcement (n=10, 19%), a medical or mental health professional (n=9, 17%), an employer (n=3, 6%), or a religious leader/official (n=2, 4%). Examples of issues reported to authority figures included disruptive outbursts, drug use, mental health issues, threats of violence against family members, and indirect threats against the subject of a grievance. Types of action taken included a restraining order, removal from a situation, monitoring, or substance and mental health treatment. Reporting concerns over violent ideology or attack planning There were 48 cases (92%) in which at least one bystander was aware of the offender’s grievance or ideology. In a quarter of those cases (n=12, 25%), bystanders reported concerns to local, state, or federal law enforcement over the offender’s extreme beliefs and statements, threats of violence, or other attack preparation behaviors. Information was brought forward by family members, peers, judges/lawyers, and/or a psychiatrist. Reports were made a year or more before the attack (n=5), within the year before the attack (n=5), or after the attack (n=2).80 Law enforcement response varied depending on the type of concerns reported, but actions included home visits, arrests or other legal proceedings, and facilitating contact with psychiatric services. Actions Taken: Operational Considerations Threat assessors and researchers can explore methods to increase reporting and improve the ways in which concerning behaviors are addressed by assessing the types of behaviors and statements that cause concern and the types of measures or interventions that bystanders take. While some similarities were seen across offenders, each offender had risk factors and stressors unique to their own circumstances. Therefore, threat assessments often require tailoring to the individual by assessing potential support systems, methods of redirection, and available resources. Prior operational research has found that individuals in the best position to observe and recognize indications of radicalization and violent ideation are family members and close friends. However, In two serial offending cases, family members notified authorities after the attacks had occurred, but before the offender was identified. 80 72 family and close friends can also be the most reluctant to report such observations. Researchers observed that cases in which family members reported concerns to authority figures were usually situations in which offenders directed threats of violence toward family members or family members were concerned about the offender’s safety. Additional reports occurred when family members or close contacts became aware of violent planning or action and did not sympathize with the offender’s ideology or violence in furtherance of that cause. Community authority figures who are in a position to respond to potential threats can work to develop trusting and substantive relationships with community members, educators, and religious leaders to sensitize them to potential signs of radicalization and mobilization. Additionally, threat assessors and community authority figures can work to empower bystanders, provide comfortable avenues of reporting, and assist in obtaining resources. Frequently, bystanders appear uncertain of how seriously they should take any specific statement or behavior and are unsure of what action to take. Bystanders close to the offender often addressed concerns primarily with the offender directly or first discussed their observations with other individuals. Efforts to promote communication and relationships between law enforcement and community leaders can be time and resource intensive, but the potential rewards for early recognition and reporting are incredibly valuable to prevention efforts, lengthening the potential time for prevention efforts before a potential attack. 73 Concluding Remarks When discussing lone offender terrorism, it is crucial to acknowledge that no one characteristic or factor - or specific combination of factors cause an individual to engage in targeted violence. Instead, the decision and ability to use violence in furtherance of an ideological goal are influenced by a complex blend of personal motivators, external influences, internal stressors, capability levels, and opportunities. Most offenders exhibited a range of concerning behaviors or statements prior to engaging in violent action. Additionally, most offenders were not truly isolated and had family, peers, or online contacts who were in a position to notice troubling behavior. While these bystanders may not always have been able to fully contextualize their observations, more than half of those who observed concerning behaviors made some effort to intervene or voice their concerns. Research and operational experience emphasize the importance of educating potential bystanders of warning signs and providing individuals with tools or mechanisms that enable them to report or otherwise address concerns. Law enforcement is one component of a larger multidisciplinary approach to assessing and mitigating potential threats. Community partners and private, non-governmental organizations (NGOs) have adopted a significant role in the management and mitigation of ideological and grievance-based threats. These organizations seek out and leverage the training and experience of practitioners, academics, and law enforcement. Yet, lingering challenges still exist, such as high case load and the ability to transfer threat cases to new geographical locations or entities. The integration of operational experience, research, and ongoing collaborations between law enforcement and community partners will allow for stronger, coordinated threat management strategies that work to disengage individuals from violent trajectories and reduce the risk of future lethal attacks. 74 Limitations The authors acknowledge several limitations of the current research. In terms of data collection, coding relied partially on case information gleaned from post-attack interviews with bystanders who knew or came into contact with offenders. Researchers corroborated information where possible and coded variables as unknown or unclear when source credibility was questionable or when provided information was too vague. However, bystanders may have been impacted by recall biases when providing information. Additionally, there were cases in which interviewees may have sympathized to some extent with offenders’ ideologies or may have been otherwise motivated to avoid connection to the offender or their crimes. Therefore, any lack of transparency could have resulted in the underreporting of certain variables, particularly those related to ideological associates, group contacts/affiliations, and bystander awareness before the attack. While the report provides many future avenues for exploration, it should be emphasized that the current data reports observations taken from one sample. Rates of characteristics, behaviors, or other variables were not compared to those exhibited by individuals who may also adhere to extreme views, but never mobilize toward violence. Therefore, the data cannot be used to provide indicators that would predict who will ultimately attempt or succeed in carrying out an attack. Additionally, the sample of lone offender terrorists was not compared to group-based terrorism actors or other types of targeted violence offenders, who may display different characteristics or preattack behaviors. Due to the low base rate of lone offender terrorism and the amount of variability between offenders, the addition of more cases could change the results of the current analysis. The small sample also means that care should be taken before making broad generalizations from the data, as any extreme example could skew the data in a particular direction. Any case examples inserted throughout the report are taken out of their full context in order to protect case anonymity. 75 References American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Arlington, VA: American Psychiatric Publishing. Amman, M., Bowlin, M., Buckles, L., Burton, K., Brunell, K., Gibson, K., Griffin, S., Kennedy, K., & Robins, C. (2017). Making prevention a reality: Identifying, assessing, and managing the threat of targeted attacks. Washington DC: Federal Bureau of Investigation, National Center for the Analysis of Violent Crime, Behavioral Threat Assessment Center. Barnes, J. C., Jorgensen, C., Beaver, K. M., Boutwell, B. B., & Wright, J. P. (2015). Arrest prevalence in a national sample of adults: The role of sex and race/ethnicity. American Journal of Criminal Justice, 40(3), 457-465. doi:10.1007/s12103-014-92733 Borum, R., Fein, R., & Vossekuil, B. (2012). A dimensional approach to analyzing lone offender terrorism. Aggression and Violent Behavior, 17. doi:10.1016/j.avb.2012.04.003 Cohen, C. I., & Marino, L. (2013). Racial and ethnic differences in the prevalence of psychotic symptoms in the general population. Psychiatric Services, 64(11), 1103-1109. Retrieved from https://ps.psychiatryonline.org/doi/pdf/10.1176/appi.ps.2012003 48. Center for Behavioral Health Statistics and Quality. (2018). 2017 National Survey on Drug Use and Health: Detailed Tables. Substance Abuse and Mental Health Services Administration, Rockville, MD. Retrieved August 7, 2019 from https://www.samhsa.gov/data/sites/default/files/cbhsqreports/NSDUHDetailedTabs2017/NSDUHDetailedTabs2017.p df Elbogen, E.B. & Johnson, S.C. (2009). The intricate link between violence and mental disorder: results from the National Epidemiologic Survey on Alcohol and Related Conditions. Archives of General Psychiatry, 66(2), 152-161. doi:10.1001/archgenpsychiatry.2008.537 76 Fein, R.A. & Vossekuil, B. (1999). Assassination in the United States: An operational study of recent assassins, attackers, and nearlethal approachers. Journal of Forensic Sciences, 44(2), 321-33. doi:10.1520/JFS14457J Fein, R., Vossekuil, B., & Holden, G. (1995). Threat assessment: An approach to prevent targeted violence. Washington, DC: U.S. Department of Justice, Office of Justice Programs, National Institute of Justice, Publication NCJ 155000. Fisher, R. P., & Schreiber, N. (2017). Interview protocols to improve eyewitness memory. In The Handbook of Eyewitness Psychology: Volume I (pp. 53-80). Psychology Press. Gill, P. (2015). Lone-actor terrorists: A behavioural analysis. Routledge. Gill, P., Marchment, Z., Corner, E., & Bouhana, N. (2018). Terrorist decision making in the context of risk, attack planning, and attack commission. Studies in Conflict & Terrorism, 1-16. doi:10.1080/1057610X.2018.1445501 Harvard Medical School. (2007). Lifetime prevalence DSM-IV/WMHCIDI disorders by sex and cohort. [Table 1]. National Comorbidity Survey (NCS). Retrieved July 30, 2019 from https://www.hcp.med.harvard.edu/ncs/index.php. Meloy, J.R., Sheridan, L., & Hoffman, J. (Eds.) (2008). Stalking, threatening and attacking public figures: A psychological and behavioral analysis. New York, NY: Oxford University Press. Monahan, J., Steadman, H.J., Silver, E., Applebaum, P.S., Clark Robbins, P., Mulvey, E.P., Roth, L.H., Grisso, T., & Banks, S. (2001). Rethinking risk assessment: The MacArthur study of mental disorder and violence. New York, NY: Oxford University Press. National Consortium for the Study of Terrorism and Responses to Terrorism (START). (2017). Global Terrorism Database [Data file]. Retrieved from https://www.start.umd.edu/gtd Nock, M.K., Borges, G., Bromet, E.J., Alonso, J., Angermeyer, M., Beautrais, A., Bruffaerts, R., Chiu, W.T., De Girolamo, G., Gluzman, S., & De Graaf, R. (2008). Cross-national prevalence 77 and risk factors for suicidal ideation, plans and attempts. The British Journal of Psychiatry, 192(2), 98-105. doi:10.1192/bjp.bp.107.040113 Oliveros, J.C. (2007-2015). Venny. An interactive tool for comparing lists with Venn's diagrams. Retrieved August 7, 2019 from http://bioinfogp.cnb.csic.es/tools/venny/index.html Phillips, B. J. (2017). Deadlier in the US? On lone wolves, terrorist groups, and attack lethality. Terrorism and Political Violence, 29(3), 533-549. doi:10.1080/09546553.2015.1054927 Silver, J., Simons, A., & Craun, S. (2018). A study of the pre-attack behaviors of active shooters in the United States between 2000–2013. Federal Bureau of Investigation, U.S. Department of Justice, Washington, D.C. 20535. Retrieved August 7, 2019 from https://www.fbi.gov/file-repository/pre-attack-behaviorsof-active-shooters-in-us-2000-2013.pdf/view U.S. Bureau of Labor Statistics. (2019, January 18). Employment status of the civilian noninstitutional population, 1948 to date. [Table.] Labor Force Statistics from the Current Population Survey. Household Data Annual Averages. United States Department of Labor. Retrieved August 7, 2019 from https://www.bls.gov/cps/cpsaat01.htm U.S. Census Bureau. (2015). Veteran status, American Community Survey 1-year estimates. [Table S2101]. Retrieved August 7, 2019 from https://factfinder.census.gov/faces/tableservices/jsf/pages/produ ctview.xhtml?pid=ACS_05_EST_S2101&prodType=table U.S. Bureau of the Census. (1973). Statistical abstract of the United States: 1973. (94th edition.) Washington, D.C. Retrieved August 7, 2019 from https://www.census.gov/library/publications/1973/compendia/st atab/94ed.html Wu, E. Q., Shi, L., Birnbaum, H., Hudson, T., & Kessler, R. (2006). Annual prevalence of diagnosed schizophrenia in the USA: a claims data analysis approach. Psychological medicine, 36(11), 1535-1540. doi:10.1017/S0033291706008191 78 Appendix Isolation Scales Personal Isolation The Personal Isolation score is based on a six-point scale, with a higher score indicating a higher degree of personal isolation. 1. Did the offender lack any type of organizational affiliation? Item 1 included affiliation to either a non-violent or violent ideological group. 2. Did the offender live alone? 3. Was the offender single? Item 3 included offenders who were divorced, separated or widowed – essentially, any offender who was not married or in a relationship at the time of the index attack. 4. Did the offender lack any close family relationships? In scoring family closeness, coders were asked to list each family member present in the offender’s network and score the level of closeness (defined as the quantity and quality of contact) between the offender and each family member using a five-point Likert scale. Item 4 included offenders who either had no family relationships or didn’t have any family relationships scoring above a three. 5. Did the offender lack a close social network? In scoring social network closeness, coders were asked to score whether “there were people in the offender’s social network that liked and were socially connected to the offender” using a fivepoint Likert scale. Item 5 included offenders who either did not have anyone in their social network or had a social network closeness score below a three. 79 6. Was the offender self-radicalized? Item 6 examined the mechanism by which offenders’ initial involvement in their violent extremist ideology occurred – whether offenders’ initial involvement occurred due to contact with other individuals who espoused an extremist ideology, or whether offenders independently developed their extremist ideologies. 80 National Center For The Analysis of Violent Crime Federal Bureau of Investigation Behavioral Analysis Unit 81