Similar crimes, similar behaviors? Comparing lone-actor terrorists and public mass murderers

1 INTRODUCTION

The study of both public mass murderers and lone-actor terrorists independently emerged as two distinct fields of inquiry in the past decade. The initial distinction centered around the primacy given to offender motivation in shaping inclusion criteria. Whilst early studies of lone-actor terrorists necessitated some form of overarching political, religious, or social goal [1], similar studies of mass murderers focused on mental health histories and precipitating events [2]. As the empirical study of both criminal types developed, both are in a state of conceptual and definitional flux. Mass murder studies differ on questions surrounding offender motive [3, 4], type of weapon used [3], number of wounded [5], and number killed [5-10]. Lone-actor terrorist researchers disagree on inclusion boundaries concerning the degree of loneness during an individual's radicalization [11], ideology [12, 13], the target [14, 15], outside direction [14, 16], and the presence of mental illness [17].

Amongst these conceptual debates, both literatures are questioning whether the distinction between both offender types is so clear cut. This is true for both single case study approaches and those with a larger sample size. For example, Stern [18] questions the political motivations of the terrorist Mir Aimal Kasi. On the other hand, White [19] analyzes the social and political themes within the mass murderer Elliot Rodgers’ manifesto. Gill's study [20] of 111 lone-actor terrorists suggests personal grievances and stressors often play a key role in the individual's radicalization and attack planning. On the other hand, Malkki [21] found the presence of political drivers and references to social factors in a majority of 28 rampage school shooter cases. Borum et al. [22] find this unclear distinction unsurprising: “The fundamental reason motivational analysis is so vexing is that humans are complicated. Their motivations—especially for horrific acts—are rarely pure and singular.”

Comparative approaches now coalesce around the idea that little separates these offender types. Capellan compared 242 non-ideological active shooters with 40 ideological active shooters. He concluded they “may be outcomes of the same social and psychological processes…part of a larger phenomenon of lone-actor grievance-fueled violence” [23 p407]. Capellan and Anisin's later study depicts the two categories as a “distinction without a difference” [24, p1]. Böckler et al. compared seven school attackers with seven Islamist attackers and concluded “it might be more from an operational perspective to discuss severe target school violence and terrorist attacks under a common concept of demonstrative violence than to artificially assign them to exclusive classes of violence” [25, p1]. This thinking has begun to seep into practice with the development of joint agency responses to counter attacks by “lone, mentally unstable individuals” [26], application of terrorism threat assessment tools to school shooters [27], and the development of a Targeted Violence and Terrorism Prevention initiative under the U.S. Department of Homeland Security.

Further comparison is warranted to address some of the shortcomings in the earlier mentioned analyses. This article compares 115 public mass murderers and 71 lone-actor terrorists across a range of features beyond demographics and event characteristics [23], incorporating a wider set of behaviors and experiences beyond mental health histories, grievances, and strain [24] whilst also utilizing a large-n dataset [25]. We ask: what (dis)similarities are observable across both offender types and what are the implications for law-enforcement? The next section outlines the unique datasets that are used. We then turn our attention to the individual offender and then offender behavior.

2 DATA AND METHOD

This article's data are dependent on the systematic data collection of openly available information. Previous terrorism research demonstrates richer available granular behavioral data are available for lone-actor terrorists than those who co-offend in larger groupings. For the latter, it is usually only standard sociodemographic information available from open sources [28]. The richness in available information is related to lone-actor terrorisms comparative rareness. For similar reasons, publicly available information on mass murderers is also bountiful [6]. Public mass murder is uncommon compared with other murder types in the United States. Between1976 and 2000, the percentage of murders involving more than one victim was 3%–4% of total events per year [29]. The FBI’s Supplementary Homicide Reports from 2000 to 2012 demonstrates the number of murder incidents involving four or more victims is approximately one-tenth of one percent of all murders.

We limit our sample to post-1990 events. Many of our original data sources are sparse before this period. The lone-actor terrorism data are an update of the data and procedures outlined in Gill et al. [30] and brings the data up to 2014. The mass murder data are outlined in a separate publication [31]. The analysis therefore compares offenders spanning 1990 to 2014 [32].

In terms of definitions, this paper is uses the same definitions as the papers which originally collected both sets of data [30, 31]. Gill et al. [30] defined terrorism as the “use or threat of action where the use or threat is designed to influence the government or to intimidate the public or a section of the public and/or the use or threat is made for the purpose of advancing a political, religious or ideological cause. Terrorism can involve violence against a person, damage to property, endangering a person's life other than that of the person committing the action, creating a serious risk to the health or safety of the public or a section of the public, or facilitating any of the above actions.”

Gill et al. [31] defined mass murder as the killing of four or more people (not including the offender) at one (or multiple but geographically close) location(s) over a relatively short period of time (for further explanation of this inclusion criteria, see Ref. [30]).

To make ecologically valid comparisons, this article made a number of important decisions in its treatment of the data. First, it narrows the analysis to those who perpetrated (or planned) their violence to occur in the United States. This narrows the sample size from the original lone-actor terrorism paper [30]. Second, the mass murderer data are solely focused on those whose incidents occur largely in public places (e.g., similar to how lone-actor terrorism often occurs). We omit homicides exclusively involving family members or intimate partner violence. Other research often treats these types of homicides as conceptually distinct from mass murder [9, 11]. Third, we excluded mass murders resulting from other underlying crimes (e.g., robbery and drug dealing), and gang or organized crime activities. The motivations are distinctively different from other forms of mass murder.[33] Finally, we only examine mass murders committed by a single perpetrator [10].

The codebook consists of over 180 variables based on a review of literature on risk factors for, and potential indicators of, mobilization to violence. Variables include sociodemographic information, pre-attack behaviors, behaviors specific to the predicate offense and postevent behaviors. Three coders independently coded each case. After an observation was coded, the results were reconciled in two stages (coder A with coder B, and then coders AB with C). In the case where three coders could not agree on particular variables, senior members of the project team resolved the differences on examining the original files that the coder's relied on to make their assessment. Such decisions were based on measures of the comparative reliability and quality of the sources (e.g., reports that cover trial proceedings vs. reports in the immediate aftermath of the event) and the source cited in the report. We do not distinguish between missing data and “no” answers. Unless otherwise reported, the analyses were conducted on full subsamples (n = 71 lone-actor terrorists and 115 public mass murderers).

Below, we undertake a series of bivariate and multivariate analyses. The next section outlines the significant bivariate (e.g., chi-square, Fisher's exact tests) results.

3 RESULTS

4 DISCUSSION

Overall, our data inform the emerging idea that lone-actor terrorists and public mass murderers are not distinct offender types. There is more that unites them than divides them. Although the over-arching focus of our results are on the few variables that distinguish them, the vast majority (>80%), of the 180+ variables showed no significant difference. For the most part, they are similar people, with similar mobilization pathways, committing similar forms of violence, with slightly dissimilar motivational structures. Even motivational structures are difficult to demarcate. Although lone-actor terrorists are ideologically influenced, in many cases we see these influences combine with deeply personal grievances to shape aspects of their decision-making. Although mass murderers frequently harbor some form of personal grievance, we often see the perpetrators frame their respective grievances in politicized language, or find others adopting these grievances in the aftermath of the initial attack. For example, in the months after his attack, Elliot Rodger seemed to resemble a typical public mass murderer driven by a highly idiosyncratic, albeit politicized, motivation. Since then, he has come to be seen as more of a terrorist because of his connections to the incel movement, which he helped bring to the awareness of a wider audience. The broader social and political context that follows such an act of violence has a major impact on how we immediately define them. Today's mass murderer, might be another tomorrow's terrorist.

Borum, Fein and Vossekuil's championing of a dimensional approach to understanding lone-actor terrorism is key here and instructive of our findings [34]. They make the case that instead of focusing on typologies of lone-offender terrorists (e.g., based on whether the offender received material support or acted under a formal command and control structure), it may be more useful to view (a) the degree of loneness the offender exhibited, (b) the degree of external direction the offender received, and (c) the depth of their political motivation, as continuums. “Analyzing cases by their features, rather than by their types, might better aid the investigative process, particularly if each dimension is linked to a key facet of the attack and tracked across the spectrum of attack-related activity from idea to action” [34 p. 104]. Loneness measures independence of activity. The loneness continuum plots the degree to which offenders received assistance in initiating planning, preparing for, and executing the attack. Direction measures the level of autonomy the lone actor displayed in decision-making. It plots the degree to which the offender received instruction or guidance on issues concerning whether to attack, what to target and the attack type to deploy. The motivation continuum plots the degree to which the action is primarily ideologically or personally driven. Borum et al. make the case that very few offenders will be placed on the extremes of a continuum but are likely to be found somewhere in between.

Between polar opposites, we see much convergence between lone-actor terrorists and mass murderers. They inhabit similar spaces. This finding may have important implications for law enforcement and threat assessment professionals. For example, when encountering a potential offender who appears to be motivated only by an ideologically based grievance, an investigator should not be surprised by a lack of evidence that others were aware of any planned attack (half of lone-actor terrorists were just as “lone” as the nonideologically motivated public mass murderers). Similarly, in the aftermath of a public mass murder not “claimed” by any terrorist organization, investigators should not discount the possibility that the offender was motivated at least in part by political, religious, or ideological concerns in choosing the target (over a quarter of public mass murderers had some form of ideological aspirations, albeit poorly articulated, fanciful, or lacking a wider milieu of support).

Where we do see consistent differences between the two categories relates to leakage. This, we feel, is directly linked to wider supportive ecologies and the opportunities to leak they provide. Naturally, ideologies have bigger followings than idiosyncratic personal grievances. Where there is a large (physical or virtual) presence of co-ideologues, there is a greater likelihood of some form of leakage occurring. What we found in this regard is true for 1990–2014. However, there is no guarantee, it will stay this way for two reasons. First, some forms of (attempted) shootings traditionally thought of as (attempted) public mass murder are gaining much wider support online. This is true for school shootings [35], “targeted individuals” [36], Incels [37], and various forms of conspiracy theories that cross the Rubicon into inspiring political violence (e.g., QAnon [38]). As these locales grow, it will likely foster more leakage behaviors amongst the few who mobilize to violence. Second, other online affordances such as live streaming violent events may have the opposite effect and lessen the prevalence of leakage behaviors. If one consistent reason for leaking intent is to let others know “why” the violence had to happen, a livestreamed claim of responsibility immediately prior to the violence (as seen in Christchurch) may be far less risky for perpetrators attempting to avoid upstream detection by police and intelligence agencies.

5 CONCLUSION

This article adds to the growth in data-driven analyses comparing samples of violent extremists with other violent populations of interest [39]. The results demonstrate that many major potential risk factors (e.g., existing mental disorders) are just as common among and between lone-actor terrorists and mass murderers. This has major implications for practice and justifies the development of multi-agency approaches that do not discriminate between motivational drivers of violence. Threat/risk assessment and management guidance documents could be equally applicable to both forms of attacker and indeed have already been validated by the Risk Management Authority (e.g., TRAP-18). Different forms of interventions may also be equally suitable. For example, by targeting the social and individual determinants of who crosses the threshold into targeted violence in public spaces, it will aid upstream prevention. Policies that encourage early reporting of warning signs of imminent violence can have an equal impact. Interventions that target the ease of access to weaponry used for such mass violence will equally impact all forms of public violence. Interventions which target harden, and aim to disrupt, deter, and deny hostile behavior will also have similar impacts regardless of the motive driving the violence.