You are on page 1of 10

J Forensic Sci, 2018

doi: 10.1111/1556-4029.13985
PAPER
PSYCHIATRY & BEHAVIORAL SCIENCE

Joel A. Capellan ,1 Ph.D.; Joseph Johnson,1 Ph.D.; Jeremy R. Porter,2 Ph.D.; and
Christine Martin,1 B.A.

Disaggregating Mass Public Shootings: A


Comparative Analysis of Disgruntled
Employee, School, Ideologically motivated,
and Rampage Shooters

ABSTRACT: This study compares the demographic, background, motivation, and pre-event and event-level behaviors across four types of
mass public shooters: disgruntled employee, school, ideologically motivated, and rampage offenders. Using a database containing detailed
information on 318 mass public shootings that occurred in the United States between 1966 and 2017, we find systematic differences in the
characteristics, motivations, target selection, planning, and incident-level behaviors among these offenders. The results show that ideologically
motivated shooters to be the most patient, and methodical, and as a result the most lethal. Conversely, disgruntled employees, who are driven
by revenge, tend to have little time to plan and consequently are the least lethal shooters. These, among other differences, underscore the need
for prevention strategies and policies to be tailored to specific types of offenders. Furthermore, the results also highlight commonalities across
offender type, suggesting that the social and psychological pathways to violence are universal across offenders.

KEYWORDS: forensic science, mass public shootings, ideological extremism, homicide, school shootings, mass murder

The literature on mass shootings has given us valuable the field of mass murder but also across different areas of terror-
insights on the incidence (1), demographic and background char- ism studies.
acteristics of offenders (2), motivations (3), and individual, and In the current study, we expand on this literature by examin-
macrolevel processes that may be responsible for their incidence ing the demographic, background characteristics, motivations,
and distribution (4–6). While their findings have been instrumen- pre-event and event-level behaviors of disgruntled employee,
tal for our understanding, this body of research implicitly ignores school, ideologically motivated, and rampage mass public shoot-
the heterogeneity found in the characteristics, behaviors, and ers in the United States between 1966 and 2017. Identifying the
motivations of mass shooters. Understanding this diversity commonalities and differences among these four types of offend-
requires us to disaggregate offenders into different subtypes that ers is not only crucial for our understanding the etiology of mass
may account for the heterogeneity found within this social phe- shootings, but also for creating policies and crime prevention
nomenon. Different types of mass shooters may be characterized strategies tailored for each type of shooter.
by different patterns, explanations, and situational contexts.
Previous research has compared U.S. mass shooters to differ-
Definitions
ent, but related phenomena such as suicide bombers in the Mid-
dle East (7) and assassins (8). Additionally, studies have also Consistent with the growing literature, we define a mass pub-
made comparisons between different types of mass shooters. lic shooting as an incident of targeted violence where an offen-
These include comparisons between “lone wolf terrorists” and der has killed or unequivocally attempted to kill four or more
“deranged shooters” (9), suicide terrorists to various types of victims on a public stage (e.g., workplace, schools, parks, and
mass casualty shooters (10), family annihilators, felony mass businesses) in one or multiple closely related locations within a
murder, and public mass murderers (11), as well as comparisons 24-h period (2–7,9–12). Both familicides and felony-related mass
of offender by their motivation for the attack (3). These studies shootings are excluded from this operationalization (12). Famili-
highlight the importance of comparative work not only within cide mass shootings are targeted events where the majority of
the victims are members of the offender’s immediate or extended
family. Felony-related mass shootings are exclusive to shootings
1
Law & Justice Studies, Rowan University, 201 Mullica Hill Road, carried out in conjunction with other criminal activity (e.g.,
Glassboro, NJ 08028. gangland, drug-related shootings, home invasion).
2
Brooklyn College, Sociology and Criminal Justice Programs, 2900 As noted, this study includes incidents where offenders unam-
Bedford Avenue, New York, NY 11220.
Corresponding author: Joel A. Capellan, Ph.D. E-mail: capellan@rowan.edu biguously attempted to kill four or more individuals but may
Received 11 Oct. 2018; and in revised form 19 Nov. 2018; accepted 29 Nov. have failed to meet the death-toll criterion. Cases where the
2018.

© 2019 American Academy of Forensic Sciences 814


815 JOURNAL OF FORENSIC SCIENCES

perpetrator unambiguously attempted to kill four or more victims impossible to know for sure whether cases have been missed.
but failed are theoretically relevant cases as the underlying psy- Open-source data collection strategies tend to be biased against
chological and sociological processes that propel an offender to older events (time-period effect) and events that are less publi-
commit a mass public shooting are not associated with his/her cized (publicity effect). Cases that occurred further in the past
ability to do so successfully. Therefore, cases like Robert Lewis and less sensational shootings (i.e., fewer media coverage) are
Dear Jr. (who in 2015, attacked a Planned Parenthood clinic in less likely to be identified and included in the sample. While
Colorado, killing three and injuring nine others) or Jason Bur- bias caused by publicity and time-period effects cannot be elimi-
nam (who shot, but failed to kill, four coworkers in 2007) are nated, this study tries to minimize this in several ways. First, the
just as theoretically relevant as cases like Adam Lanza’s and key words were run through eight different search engines to
should be included in any analysis. maximize exposure and identify as many incidents as possible.
In the current study, we compare four types of mass public Second, once the initial list of potential mass public shootings
shooters: disgruntled employee, school, ideologically motivated, was compiled, it was then cross-referenced with fifty-five addi-
and rampage. Consistent with the mass murder literature, dis- tional lists of mass shooting and databases provided by peer-
gruntled employee shooters are defined as individuals who target review journals, books, news organizations, school-sponsored
their current or former place of work (13–15). We also classify reports, blogs, and online encyclopedias. The cross-reference
individuals who were not necessarily employees but were part- process is critical in identifying additional mass public shootings
ners, co-owners, or investors as disgruntled employee shooters that otherwise would not have been detected.
(13,15). School shooters are defined as individuals who target Another limitation inherent to open-source data collection
their current or former schools or individuals in that school (15– strategy is reliability. Reliability problems come in the shape of
17). Shootings perpetrated by teachers and school staff are conflicting accounts of the same event by different sources.
excluded from this definition as these would be classified as a When inconsistencies in the open-source data where encoun-
disgruntled employee shootings. Additionally, for an attack to be tered, this study followed the guidelines established by Sageman
classified as a school shooting it must occur on school grounds. (22). Trusted sources of information were given more weight. In
Offenders who targeted schoolmates in other venues were not decreasing order of reliability, this study favored (i) court docu-
classified as school shooters (15–18). Mass public shooters who ments, police reports, news reports that provided corroborated
were motivated by ideological extremism (e.g., jihadist, far-right, accounts of witnesses, family members, friends; (ii) uncorrobo-
Black Nationalism) were categorized as ideologically motivated rated statements from witnesses, family members, friends; and
shooters (9). Lastly, the term rampage shooter has been used in (iii) accounts from people that heard the information second-
the media to describe mass public shootings in general (19). hand.
However, in the mass shooting literature, it has been used to cat- Despite these limitations, open-source data offer richer accounts
egorize offenders that do not fall in any of the preceding cate- than official crime statistics. Media reports and government docu-
gories (10). Effectively, a rampage shooter represents the “other” ments often include the names of offenders and victims, their
category. motives, criminal histories, preparation, execution, and conclusion
of homicides, allowing researchers to reconstruct these events in
great detail. Furthermore, while the noted sources of error could
Methods not be eliminated, they can be significantly reduced through a sys-
tematic and rigorous data collection strategy, such as the one
Data Collection Strategy
employed in this study. For example, Parkin and Gruenewald (23)
Consistent with most studies on the subject (1–12,14–19), we have demonstrated that open-data sourcing can produce databases
employed an open-source data collection strategy to identify and of crime that are just as accurate and comprehensive as official
collect information on both failed and successful mass public crime statistics.
shootings that occurred in the United States from 1966 to 2017.
Open-source data are information that is open to the public (20).
Measures
These data often come in the shape of searchable electronic doc-
uments such as newspaper articles and government documents. Although most characteristics presented in this study are rather
In order to identify all relevant cases, specific search terms (e.g., straightforward, some measures are more abstract and require
mass shooting, mass public shooting, random shooting, and further clarification. Ideologically motivated attacks represent
deranged shooting) were employed in eight different search engi- mass public shootings that were motivated by the offender’s ide-
nes (Lexis-Nexis, ProQuest, Yahoo, Google, Copernic, News ological extremist beliefs. Ideological extremism is a set of
Library, Westlaw, and Google Scholar). Double quotations and beliefs that diverge widely from mainstream policy preferences.
Boolean operators were used to increase the number of relevant To determine whether the attack was motivated by ideological
results. Once the final list of mass public shootings was gener- extremism, we relied on the Extremist Crime Dataset’s (ECDB)
ated, eight online search engines were used to obtain information Strength of Association measurement protocol. This coding pro-
on the offenders, victims, and incidents. Open-source materials, tocol is a continuum of certainty based on “pro association” and
such as media accounts, legal documents, blogs, videos, and “con association” evidence (24). This measure ranges from 0 to
government documents, were used to piece together the most 4, with four being the highest level of certainty that the attack
complete picture possible of the attack including information was ideologically motivated. Perpetrators that score greater than
about the offender’s motivation, pre-event behaviors, the location “one” were classified as an ideologically motivated offense.
of each event, victim information, and the manner in which the We also collected information on the offender’s history of
attack was carried out and concluded. mental illness. This is a categorical variable with “0” represent-
It is important to highlight some limitations inherent to open- ing no evidence of mental illness, “1” indicating confirmed men-
source data collection strategies (21). We aimed to capture all tal illness, and “3” representing offenders with a suggested
mass public shootings that met the definitional criteria, but it is history of mental illness. Offenders who received a formal
CAPELLAN ET AL. . DISAGGREGATING MASS PUBLIC SHOOTINGS 816

medical diagnosis were coded as “confirmed mental illness.” shootings across demographic, background characteristics, pre-
Unfortunately, research has demonstrated that significant por- event, and event-level behaviors. To this end, we rely on
tions of criminal offenders have never been diagnosed despite descriptive statistics and bivariate tests (i.e., chi-square, t-tests,
suffering from severe mental illnesses (25). Therefore, relying on and Fisher’s exact test) to identify significant differences across
formal medical diagnoses will ignore the full extent of the men- these different types of mass public shooters.
tal illness in this population. For that reason, we also measure
suggested history of mental illness. To determine suggested his-
Results
tory of mental illness, we reviewed open-source documents for
characterizations by family members and close friends of the Through the data collection strategy, we identified 318 mass
offender in regard to mental health status. To assess the accuracy public shootings in the United States between 1966 and 2017.
of these characterizations, only detailed accounts that included According to the results, rampage shootings are the most com-
examples or statements by the perpetrators that well-predated the monly occurring (36.1%), followed by disgruntled employees
attack were considered. (29.8%), school (19.1%), and ideologically motivated shooters
Lastly, we include a measure of the offender’s level of plan- (14.7%). This trend is not constant through time. Figure 1 pre-
ning. This is an ordinal-level variable with “0” representing no sents the type composition of mass public shootings for each
planning, and levels one, two, and three indicating low, med- decade in the analysis. The results show substantial temporal
ium, and high levels of planning, respectively. Consistent with variation in the incidence between offender type. For example,
Capellan’s (9) study, this measure is based on the observable school mass public shootings used to make up the majority of
indicators before and during the attack. “No planning” involves mass public shootings in the 1960s and 1970s; however, its rela-
an offender who had no time to plan as the catalyst for the tive incidence has been shrinking in the last few decades. Con-
shooting occurred minutes before the shooting. “Low level of versely, rampage and disgruntled employee shooting were not
planning” involves offenders who had an hour or multiple relatively common in the 1960s and 1970s, but it has become
hours pass between the catalyst and attack. Generally, these more common over time. Ideologically motivated mass public
individuals go back to their homes or cars to get firearms in shootings, however, have remained relatively stable, accounting
order to carry out the shooting. These offenders did not use for approximately 15% of all attacks.
additional nonfirearm weapons (i.e., knives and blunt objects)
and additional gear. Offenders who employed multiple firearms
Demographic and Background Characteristics
brought additional nonfirearm weapons and gear (i.e., bullet-
proof jacket, ammunition, and tactical clothing.) were catego- Table 1 compares disgruntled employee, school, ideologically
rized as “medium level of planning.” In addition to the noted motivated, and rampage mass public shooters across a range of
behaviors, offenders who engaged in “high levels of planning” demographic and background characteristics. Consistent with the
also acquired tactical and logistical information though surveil- literature, the results show mass public shootings are almost
lance and online searches. entirely male-dominated with over 96% of shootings executed
by men across the typology. There are, however, significant dif-
ferences in the age and racial background of these offenders.
Analytic Strategy
Compared to disgruntled employee (39.7), ideologically
In the current study, we compare disgruntled employee, motivated (37.7), and rampage (37.7), school shooters are signif-
school, ideologically motivated, and rampage mass public icantly younger with an average age of 20.6. These differences

FIG. 1––Type of mass public shootings through time.


817 JOURNAL OF FORENSIC SCIENCES

TABLE 1––Offender characteristics.

Disgruntled (n = 95) School (n = 61) Ideological (n = 45) Rampage (n = 115) Chi-square/t-test


Male 96.8 98.3 97.8 96.3
Age 39.7(Avg) 20.6(Avg) 37.7(Avg) 37.7(Avg) ***
Race
White 51.5 77 57.4 53.9 ***
Black 32.6 8.2 27.6 22.6
Hispanic 10.5 3.2 0 14.7
Asian 3.16 8.2 0 5.2
Middle-Eastern 0 0 14.8 2.6
Other 2 3.2 0 0.8
Single/divorced 72.6 93.4 76.6 75 ***
Employed 65.9 8.3 34 33.3 ***
Education level
High school or less 72.8 75.8 61.3 71
Some college 3.7 11.6 13.6 10.1
College 16 5 18.1 13
Graduate 7.4 8.3 6.8 6
Criminal record 25.8 6.7 48.9 40.1 ***
Military service 15.2 4.9 27.6 10.6 ***
History of substance abuse 16.6 18 20 30
Mental Health
Confirmed illness 21.2 36 29.7 16.5
Suggested illness 18 13.1 12.7 22.6
*p ≤ 0.05
**p ≤ 0.01.
***p ≤ 0.001.

hide a significant amount of variation. Figure 2 presents Kernel three categories: autogenic, revenge, and ideological. Auto-
density to estimate the age–crime curve across four types of genic motivations are “self-generated” due to the offender’s
offenders. The age–crime curve for disgruntled employee, ideo- internal psychological processes and issues (26). Generally,
logically motivated, and rampage offenders do not follow the autogenic motivations are not anchored in the external world.
traditional crime–age curve; instead, it follows a platykurtic dis- As a result, attacks fueled by autogenic motivations are often
tribution centering around the late 30’s. We also find significant seemed as motiveless or random. Our analysis reveals that
differences in the racial composition of these offenders. School school (60.3%) and rampage (55.6%) shooters are signifi-
shooters are significantly more likely to be White (77%). Dis- cantly more likely to be driven by autogenic motivations.
gruntle employees have the highest share of Black mass public Conversely, disgruntled employees (83.1%) are significantly
shooters (32.6%). Rampage shooters have the highest percentage more likely to be motivated by revenge against specific indi-
of Latino offenders (14.7%), and ideologically motivated shoot- viduals or organizations. This finding is not surprising as
ers have the highest share of offenders of Middle-Eastern des- 73.6% of disgruntled employee shootings could be traced to
cend (14.8%). precipitating events such as getting fired, being bullied, or
Our analysis also reveals significant differences in background reprimanded at work. Not all significant differences are sub-
characteristics across the four offenders. We find that school stantive. Some significant differences are the product of defi-
shooters are significantly more likely to be single (93.4%). Dis- nitional choices. For example, ideologically motivated mass
gruntled employee offenders are significantly more likely to be public shooters (100%) were significantly more likely to be
unemployed at the time of the attack (65.9%). Ideologically motivated by extremist ideologies.
motivated (48.9%) and rampage shooters (40.1%) are Previous research (3) has stressed the act of obtaining a fire-
significantly more likely to have a criminal record. Ideologically arm affords law enforcement a great opportunity to intervene
motivated shooters are also significantly more likely to be an and potentially prevent mass public shootings. Our results sug-
active or former military serviceman at the time of the shooting gest this opportunity to intervene at this stage of the process is
(27.6%). Despite these differences, these offenders share some greatest with school shooters. 79.3% of school shooters acquired
important similarities. For example, there are no significant dif- firearms for the shooting. This finding suggests that most dis-
ferences in educational attainment across these offenders. The gruntled employee, ideologically motivated, and rampage shoot-
majority (approximately 70%) of all offenders have a high ers had access to firearms, eliminating the need to obtain
school or lesser degree. Importantly, there are no significant dif- additional weapons. Although school shooters are significantly
ferences in confirmed or suggested mental illnesses. Approxi- more likely to obtain firearms, the findings suggest that when
mately 40% of all offenders suffer from confirmed or suggested ideologically motivated (x = 2) and rampage mass public shoot-
mental illnesses. ers (x = 2.5) do acquire firearms for their attack, they tend to
acquire a significantly higher number of guns than their counter-
parts.
Motivations and Pre-Event Behaviors
Threats have been long considered by the U.S. Secret Service
Table 2 compares the motivations and pre-event behaviors to be reliable and actionable intelligence that may signal an
of disgruntled employee, school, ideologically motivated, and impending violent act (27). The results show that ideologically
rampage mass public shooters. Research (3) suggests that motivated (51%) and school shooters (49.1%) are significantly
motivations for mass public shootings could be grouped into more likely to make such threats prior to the attack. There are
CAPELLAN ET AL. . DISAGGREGATING MASS PUBLIC SHOOTINGS 818

also significant differences to whom threats are directed. The Consistent with the literature (3,9), the ideologically motivated
analysis reveals that disgruntle employees (60%) and rampage shooters display significantly higher levels of preparatory behav-
shooters (46.1%) are more likely to make threats directly to the iors prior to attacks. Compared to its counterparts, ideologically
potential victims. Conversely, school shooters (63.1%) are more motivated offenders are significantly more likely to train, acquire
likely to make threats around family members and friends. Ideo- tactical gear (25.3%), acquire logistical information (44.6%), and
logically motivated shooters (38.4%) are significantly more conduct surveillance (21.2%) of targets prior the shooting. As a
likely to use public forums (e.g., social media sites) to make result, ideologically motivated offenders (38.3%) are significantly
threats. We also find that not all threats are equally trustworthy more likely to have the highest levels of preparation for the
as actionable intelligence on the potential time, target, and attack. Conversely, rampage offenders are more likely to have
method of impending attack. Among the different offenders, dis- no (13.9%) or low levels of planning (42.6%).
gruntled employees and ideologically motivated mass public
shooters are significantly more likely to follow through on all
Event-level Behaviors
aspects (e.g., time, method, and target) of said threats. Con-
versely, school shooters (63.1%) are more likely to change some Table 3 presents the event-level behaviors of disgruntled
aspects when carrying out their threats. employee, school, ideologically motivated, and rampage mass
public shooters. The results show that mass public shootings are
carried out by lone offenders in that they plan and execute these
shootings on their own and in one location. The findings suggest
that over 90% of all mass public shooters are executed by a lone
actor. Despite this commonality, the results show significant dif-
ferences in whom they target. Ideologically motivated shooters
(91.4%) are significantly more likely to target strangers. Dis-
gruntled employee offenders (88.4) are significantly more likely
to target coworkers and managers. Conversely, school shooters
(81.1%) to target classmates with whom they have or had a per-
sonal or intimate relationship.
Additionally, we find that ideologically motivated mass public
shootings (8.6%) are significantly more likely to involve multi-
ple offenders. We also find disgruntled employees (96.7%) and
school shooters (100%) to be significantly more likely to attack
locations they are familiar with. There are also significant differ-
ences in the type of locations these offenders target. For exam-
ple, the results show that disgruntled employee offenders
(68.4%) are significantly more likely to target businesses. Ideo-
FIG. 2––Kernel Density function of offender’s age.
logically motivated shooters are significantly more likely to

TABLE 2––Motivations and pre-event behaviors.

Disgruntled (n = 95) School (n = 61) Ideological (n = 45) Rampage (n = 115) Chi-square/t-test


Motivations
Autogenic 16.8 60.3 0 55.6 ***
Revenge 83.1 39.3 0 44.3
Ideological 0 0 100 0
Precipitator 73.6 47.5 36.1 55.6 ***
Acquires firearms 36.1 79.3 37.7 26.2 ***
No. firearms acquired 1.4(Avg) 1.6(Avg) 2(Avg) 2.5(Avg) *
Threat 31.5 49.1 51 34.7 *
Threat bystanders
Potential victims 60 21 23 46.1 *
Family/friends 28 63.1 38.4 34.6
Public Forums 12 15.7 38.4 19.2
Threat follow-through
Exactly 48 26.6 46 51.8 *
Somewhat 40 63.1 46.1 29.3
Diverged completely 12 10.5 7.6 18.5
Trained 10.5 9.8 25.3 6.9
Logistics 10.5 21.3 44.6 18.2 ***
Surveillance 7.3 4.9 21.2 3.6 **
Level of planning
No 6.3 3.28 4.26 13.9 ***
Low 32.9 37.7 25.5 42.6
Medium 47.8 31.1 31.9 30.4
High 12.7 27.7 38.3 13.1
*p ≤ 0.05.
**p ≤ 0.01.
***p ≤ 0.001.
819 JOURNAL OF FORENSIC SCIENCES

TABLE 3––Event-level behaviors.

Disgruntled (n = 95) School (n = 61) Ideological (n = 45) Rampage (n = 115) Chi-square/t-test


Number of offenders
Single 100 96.7 91.4 100 **
Multiple 0 3.3 8.6 0
Relationship with victims
Stranger 9.5 22.9 91.4 58.6 ***
Professional 88.4 0 8.6 12.7
Personal/intimate 2.1 81.1 0 28.7
Type of location
Business 68.4 0 14.8 39.1 ***
Government 17.8 0 36.7 12.1
School 8.4 100 4.2 8.7
Religious institution 0 0 17 3.4
Open-space 2.1 0 23.4 21.6
Other 3.1 0 4.2 14.7
Security 21 42.6 55.3 26.9 ***
Number of firearms 1.5(Avg) 1.7(Avg) 1.8(Avg) 1.8(Avg)
Type of firearms
Handgun 63.1 44.2 46.8 56.5 *
Shotgun 11.5 11.4 8.5 6
Rifle 9.4 18 6.3 13.9
Combination 15.7 26.3 38.3 23.4
Additional weapons 9.4 21.3 21.2 14.9
Fatalities 3.4(Avg) 3.2(Avg) 4.5(Avg) 3.8(Avg)
Number injured 2.9(Avg) 4.7(Avg) 6(Avg) 3.8(Avg)
Lethal resistance 19.1 14.7 53.1 40.8 ***
Nonlethal resistance 6.4 49.1 23.4 20.1 ***
Conclusion
Subdued and arrested 5.2 37.7 29.7 13.9 ***
Killed 11.5 8.2 27.6 23.4
Suicide 41 21.3 19.1 33.9
Flees, later arrested 18.9 13.1 17 15.6
Flees, later suicide 10.5 1.6 0 6
Surrenders 12.6 18 6.3 6.9
*p ≤ 0.05.
**p ≤ 0.01.
***p ≤ 0.001.

target government institutions (36.7%) and religious institutions statistically significant differences among different types of
(17%). As noted, some of the significant differences presented offenders. Figure 3 presents the Kernel density to estimate the
here stem from definitional choices. For example, given the way distribution of fatalities among the types of offenders. This graph
school mass public shooters were defined, it is not surprising illustrates several important points. First, there is a great deal of
that 100% of school shooters targeted schools. However, it is variation in the number of victims killed across the four types
important to note that schools are not only targeted by school of shooters. Second, the number of fatalities across all types of
shooters. Schools have been attacked by disgruntled employees offenders follows a negative binomial distribution, with most
of the school, ideologically motivated offenders, and rampage shootings resulting in less than four fatalities. Third, this finding
shooters. Additionally, we find that ideologically motivated exemplifies the need to include and analyze mass public shoot-
offenders are significantly more likely to target locations with ings where the perpetrator intended but failed to kill four or
security officers. more victims. The decision to exclude such cases may bring in
The results show no significant differences in the number of bias, as ‘successful’ mass public shooters may differ systemati-
firearms used during the attack. On average, all offenders use cally from ‘failed’ mass public shooters. Such bias would inhibit
between 1.5 and 1.8 firearms in the shooting. There are also no our ability to identify opportunities to mitigate the lethality of
significant differences in the percent of incidents where addi- these massacres. In order to identify the determinants of ‘failure,’
tional nonfirearm weapons (e.g., knives, IEDs, blunt objects, and we must inevitably compare successful and unsuccessful offend-
others) were used during the attack. However, there are signifi- ers.
cant differences in the types of firearms employed. Compared to The Department of Homeland Security’s active shooter
its counterparts, we find that a disgruntled employee is more response protocol (29) dictates that potential victims evacuate
likely to use handguns (63%) and shotguns (11.5%). Rampage and not engage the shooter. Our findings, however, show poten-
shooters (13.9%) are significantly more likely to use rifles, and tial victims, security, and police officers often resist the shooter
ideologically motivated mass public shooters (38%) are more in lethal (e.g., gunfire) and nonlethal ways. Importantly, we find
likely to use a combination of different firearms. that not all offender types have the same risk of being engaged
Among all the individual- and event-level characteristics that during the shooting. The results show that ideologically moti-
we have discussed, none get more attention than the human vated (53.1%) and rampage shooters (40.8%) are more likely to
costs of these massacres (15,28). Our results suggest that while be met with lethal forms of resistance than their counterparts.
ideologically motivated offenders have the highest number of Conversely, school shooters (49.1%) are significantly more
fatalities (x = 4.5) and injured victims (x = 6), there are no likely to be met with nonlethal forms of resistance compared to
CAPELLAN ET AL. . DISAGGREGATING MASS PUBLIC SHOOTINGS 820

mass public shootings could only occur if public officials


develop crime prevention strategies specific to rampage and dis-
gruntled employee shooters.

Threat Assessment
One promising prevention strategy is threat assessment (31–
33). Threat assessment is the process of identifying, assessing,
and managing the threat that certain persons may pose (27,31–
33). As conceptualized by the Secret Service (27), threat assess-
ment is predicated on communication from potential offenders
that may signal an impending violent act. Our results show that
a considerable number of shooters make threats prior to their
FIG. 3––Kernel Density function of the number of killed victims. attack. Unfortunately, our findings also suggest threat assessment
may not be equally effective across all offender types. School
and ideologically motivated offenders were significantly more
their counterparts. Whether or not a shooter is met with lethal likely to make such threats compared to disgruntled employee
and nonlethal forms of resistance seems to be associated with and rampage shooters.
the conclusion of the attack. For example, school shooters Another potential roadblock to the success of threat assessment
(37.7%) are significantly more likely to be subdued and arrested, is the reporting of threats. Research suggests that an important
whereas ideologically motivated offenders are more likely to be determinant of reporting of crimes and threats is the bystander’s
killed during the attack compared to their counterparts. Relative relationship with the offender (31–34). This literature finds that
to school, ideologically motivated and rampage offenders, dis- threats and crimes made in the presence of friends and family are
gruntled employees are significantly more likely to commit sui- less likely to be reported to the authorities (35). Our results show
cide during the incident (41%) and flee and later commit suicide that while school shooters and ideologically motivated offenders
(10.5%). Additionally, school shooters are more likely to surren- are more likely to express their intention of the impending
der during the attack (12.6%). attacks, they are also more likely to express these threats toward
the people (family and friends) less likely to report them. In addi-
tion to being a window into the offender’s state of mind, the
Discussion Secret Service’s threat assessment model (27) also sees threats as
actionable intelligence on the potential time, target, and method
Reality versus Perception
of an impending attack. Our findings suggest that not all offen-
The results show systematic and substantive differences across der’s threats provide reliable information on the specifics of the
the offender types. Perhaps the most notable difference is related attack. We find that rampage shooters are more likely to follow
to their relative incidence. We find that rampage (36.1%) and through exactly with all said elements in their threats. Con-
disgruntled employee (29.8%) offenders carried out approxi- versely, school shooters are significantly more likely to somewhat
mately 65% of all mass public shootings in the United States follow through with all said elements in their threats.
between 1966 and 2017. This finding is in direct contrast with The differences in threat behavior across offender type bolster
the public attention given to these shooters. Silva and Capellan the underlying justification of this study: Effective prevention
(15) find, for example, that rampage and disgruntled employee strategies must account for the heterogeneity in the characteristics
offenders receive 15-20% of total news coverage on mass public and behaviors of mass public shooters. Our results do not suggest
shootings. Conversely, school and terrorist offenders receive 75- that threat assessment cannot be effective across-the-board, rather
80% of all news coverage. Public officials also pay dispropor- for it to be effective, it must be tailored to specific shooters. For
tionate attention to school and terrorist shootings. Take for example, we believe that threat assessment could be particularly
example, President Barack Obama. Silva and Capellan (15) note useful in preventing school mass public shootings. Schools pos-
that during his tenure the U.S. experienced 123 mass public sess the infrastructure, workforce, and information to carry out
shootings. However, President Obama spoke in the immediate effective threat assessments on a student’s threats. However,
aftermath of fourteen of those incidents. Nine of those remarks schools must account for the potential problem of reporting of
were in response to terrorist mass public shootings, and three threats. One possible solution is to create a public awareness
were in response to school mass public shootings. campaign where students and parents are made aware of warning
This attention has historically led to legislation specific to signs of mass public shootings, and the importance of reporting
types of shootings. Schildkraut & Hernandez (30) identified that all manners of threats. Importantly, students and parents must be
a hundred pieces of legislation were introduced as a direct made aware that all reports will be taken seriously, reports will
response to school shootings. For example, in 1994 the U.S. be confidential to the public, and the process will be conducted
Congress passed the Gun-Free School Act, which allowed in a respectful, holistic, and fair manner for all involved. As is
schools to expel students who bring a firearm onto school the case for school shooters, threat assessment could be tailored
grounds. Similarly, ideologically motivated shootings have to maximize its effects in preventing disgruntled employee, ideo-
sparked policy discussion on banning individuals on terrorist logically motivated, and rampage mass public shootings.
lists from buying firearms among many other policy discussions.
Although the attention and legislative efforts to prevent ideologi- Motivation
cally motivated and school mass public shootings are to be
applauded, such efforts ignore the majority of mass public shoot- We also find significant differences in the motivations among
ings. Our findings demonstrate that a significant reduction in disgruntled employee, school, ideologically motivated, and
821 JOURNAL OF FORENSIC SCIENCES

rampage shooters. Disgruntled employee shooters are signifi- used in their attack, ideologically motivated shooters are signifi-
cantly more likely to be driven by revenge, whereas school cantly more likely to use a combination of different types of
shooters are significantly more likely to be driven by autogenic firearms. Consistent with the noted literature, we find that ideo-
or self-generated motivations. Naturally, all ideologically moti- logically motivated mass public shootings yield, on average, the
vated shooters were motivated by ideological definitions. highest number of fatalities and injured victims. Meloy and
According to Osborne and Capellan (3), the motives behind Richard’s (37) work on mass murder may be vital in understand-
mass public shootings are predictive of behaviors and character- ing how ideology influences the behaviors of ideologically
istics of offenders. Using a script analysis, they find that offend- motivated offenders. According to Meloy and Richards (37),
ers motivated by revenge tend to target individuals with whom ideologically motivated mass murderers tend to be fixated with
they have personal/professional relationships. Their victims are military tactics and weaponry. They see themselves as warriors
connected to what they called “precipitating events”—incidents fighting for a just cause. It is perhaps this warrior mentality that
that caused acute emotional or financial strain. Due to the short influences them to care about planning and execution, which,
time passed between the precipitating event and decision to com- in turn, makes them particularly lethal.
mit the shooting, these offenders have the lowest levels of plan-
ning, which translates into lower death tolls. Osborne and
Resistance and Outcomes
Capellan (3) also find that offenders motivated by revenge
embody different characteristics. These offenders tend to be non- Next to the decision to start the attack, the second most
White, in their late 30’s, and in a relationship/married. Our important decision a mass public shooter can make is to stop
results are consistent with such characterizations. Disgruntled shooting. Often, this is not a decision at all. Our results show
employee shooters who have the highest percentage of offenders that a substantial number of mass public shooters encounter
motivated by revenge (83.1%) are most likely to experience lethal and nonlethal resistance during their attack. Interestingly
acute strain from a precipitating event. Relative to their counter- however, there are systematic differences among offender types.
parts, disgruntled employee shooters are least likely to have high Ideologically motivated shooters are significantly more likely to
levels of planning. Moreover, while not the lowest, disgruntled be met with lethal forms of resistance. This finding may be
employee shooters have the second lowest number of fatalities explained by target selection. Our results show that ideologically
and the lowest number of injured victims. Consistent with motivated offenders are significantly more likely to target loca-
Osborne and Capellan’s (3) findings, disgruntled employee tions with security, such as government institutions. Security
offenders had the highest percent of Black and the second high- officers are trained to meet the lethal form of threat used by
est percentage of Hispanic mass public shooters. offenders and therefore rely on lethal forms of resistance to
Osborne and Capellan (3) note that offenders fueled by bring the shooter to a halt. We also find that school shooters are
autogenic motivations are overwhelmingly White males and are significantly more likely to be met with nonlethal resistance.
relatively younger (mid-twenties). Since the shooting is This finding is surprising given that after the Columbine shoot-
“self-generated” and due to the offender’s internal psychological ings in 1999 first responders were trained to take immediate
processes, these offenders are more likely to have a history of action and to isolate and incapacitate the offender (38). Figure 4
mental illness. These offenders generally have medium levels of presents the percent of mass public shooters that are met with
preparation, making them quite lethal, and second only to ideo- lethal resistance before and after the Columbine school shooting
logically motivated offenders. Our results are also consistent by offender type. This figure shows us that changes in training
with such characterizations. School shooters (60.3%) have the in the wake of the Columbine school shooting have only
highest percentage of autogenic-motivated offenders. As hypoth- impacted school and ideologically motivated shootings. There
esized by Osborne and Capellan (3), they also have the highest has been an 11% and 45% increase in the number of school and
percentage of White offenders with an average age of 20.6. ideologically motivated shootings that have been met with lethal
Although the differences are not significant, school shooters forms of resistance, respectively. The new guidelines seem not
have the highest percentage of confirmed diagnosis of mental ill- to have affected disgruntled employee and rampage shooters.
ness. Consistent with Osborne and Capellan’s (3) characteriza- Undoubtedly, differences in lethal and nonlethal forms of
tions, school shooters have the second highest levels of resistance have an impact on how the attack ends. We find that
planning, and while their attacks result in the lowest number of school shooters who are most likely to encounter nonlethal resis-
deaths (x = 3.2), they yield the second highest number of injured tance are to be the most likely to be subdued and arrested. Con-
victims (x = 4.7). versely, ideologically motivated shooters who are most likely to
Research has consistently shown ideologically motivated mass be encountered with lethal resistance are the most likely to be
public shooters to be particularly methodical and lethal killed during the attack.
(3,9,10,34). According to Osborne and Capellan (3), offenders
motivated by ideological extremism are very patient and unlikely
Target Hardening
to attack because of a precipitating event. They engage in a
range of preparatory behaviors and use a higher number of fire- The link between lethal resistance and the outcome of the
arm and nonfirearm weapons. As a result, ideologically moti- attack underscores the importance of target hardening to prevent
vated offenders have been found to have a higher number of and mitigate the lethality of all mass public shooters. Target
fatalities and injured victims (3,9,10,36). Our results are consis- hardening refers to a wide range of measures to strengthen the
tent with these findings. Our results show that ideologically security of a location. These may range from architectural design
motivated shooters are least likely to be affected by precipitating elements to employing security staff. Although implementing
events. They are however most likely to engage in predatory security measures is likely to reduce the lethality of future
behaviors for the shooting and, consequently, are significantly attacks, our findings suggest that similar to threat assessment,
more likely to have high levels of planning. Although there are target hardening practices must be tailored to specific offenders
not any significant differences in the average number of firearms to maximize its effects. For example, organizations seeking to
CAPELLAN ET AL. . DISAGGREGATING MASS PUBLIC SHOOTINGS 822

FIG. 4––Resistance before and after Columbine shooting by offender type.

prevent disgruntled employee shootings must focus on security public shooters across the board share a very similar dysfunc-
measures that protect from within as the offender, by definition tional) backgrounds. They tend to be single/divorced, unem-
currently works there or is a former employee. This means ployed, and have low levels of education. A significant number
removing access to the workplace to employees that are fired or of these offenders have a history of substance abuse and mental
placed on probation; conducting a risk assessment on current illness. Consistent with previous comparative studies (6–10),
employees involved in conflicts; creating a zero-tolerance policy these similarities suggest that social, economic, and psychologi-
on violence, sexual harassment, and psychological bullying; and cal processes behind mass public shootings are the same for dis-
establishing the mechanisms through which incidents can be gruntled employee, school, ideologically motivated, and rampage
reported and conflicts resolved peacefully. Conversely, organiza- shooters.
tions trying to prevent ideologically motivated attacks, as our Additionally, disgruntled employee, school, ideologically moti-
results suggest, must protect themselves primarily from out- vated, and rampage shooters could be characterized as lone
siders. Accordingly, measures that restrict access to and screens actors, as these shootings are perpetrated by a lone actor whose
outsiders are particularly important. Given the level of sophisti- tactics and methods are absent of direct outside support, com-
cation and planning of ideologically motivated shooters, organi- mand, or direction (40). Given the solitary nature of this phe-
zations must rely on armed and trained professionals that may nomenon, it is almost impossible for law enforcement agencies
repel potential offenders from attempting to enter by force. to detect, infiltrate, and foil these attacks unless perpetrators
Additionally, architectural elements that create natural barriers openly discuss their plans (e.g., threats) or raise other behavioral
divide access to the building, and provide clear exits to potential flags (e.g., leakages) that can be used as intelligence by law
victims could be effective in mitigating the lethality of these enforcement. This finding underscores the importance of threat
attacks. assessment as a prevention tool.

Similarities among Offenders Conclusions and Future Research


The similarities shared among offender types are just as infor- In this study, we aimed to advance our understanding of mass
mative as the differences we have noted above. One of the most public shootings by comparing disgruntled employee, school,
important commonalities these offenders share is their gender. ideologically motivated, and rampage shooters across a wide
We find that mass public shootings are carried out almost exclu- range of individual characteristics, pre, and incident-level behav-
sively by men. This is the most universal and yet the most iors. To that end, we examined and compared 318 mass public
understudied aspect of this phenomenon. Several scholars, how- shootings that occurred in the United States between 1966 and
ever, have argued that to further understand the causes of mass 2017. A number of important differences emerged through this
public shootings we must examine the role that masculinity and comparative analysis. For example, we find that there are signifi-
gender, as well as macrolevel patriarchal hegemony and gender cant differences in the motivations behind the attacks. As dis-
inequality, interact together (39). In addition to gender, mass cussed above, the motivation driving the shooting plays an
823 JOURNAL OF FORENSIC SCIENCES

important role in the preparatory behaviors, target selection, and, federal public health and safety policy. Washington, DC: Congressional
consequently, how deadly these massacres are. Furthermore, Research Service, 2013.
13. Fox JA, Levin J. Firing back: the growing threat of workplace homicide.
some of the differences identified in this research underscore the Ann Am Acad Pol Soc Sci 1994;536(1):16–30.
need for prevention strategies to be tailored to specific types of 14. Holmes RM, Holmes ST. Mass murder in the United States. Upper Sad-
offenders. For example, we noted that for a threat assessment to dle River, NJ: Prentice Hall, 2001.
be successful in preventing school shootings, schools must 15. Silva JR, Capellan JA. A comparative analysis of media coverage of
mass public shootings: examining rampage, disgruntled employee,
account for the potential problem of reporting of the threats.
school, and lone-wolf terrorist shootings in the United States. Crim Jus-
Similarly, the effectiveness of target hardening measures is unli- tice Policy Rev 2018. https://doi.org/10.1177/0887403418786556
kely to be homogeneous. Consequently, organizations must tailor 16. Langman P. Rampage school shooters: a typology. Aggres Violent
their security measures to minimize their specific threats. Behav 2009;14(1):79–86.
Our findings highlight the benefits of disaggregating mass pub- 17. Larkin RW. The Columbine legacy: rampage shootings as political acts.
Am Behav Sci 2009;52(9):1309–26.
lic shootings into typologies that may explain some of the hetero- 18. Vossekuil B. The final report and findings of the Safe School Initiative:
geneity found in the characteristics and behaviors of mass public implications for the prevention of school attacks in the United States.
shooters. It is important to note that no typology will account for Collingdale, PA: DIANE Publishing, 2002.
all of these variations. Accordingly, research should continue to 19. Schildkraut J, Elsass HJ. Mass shootings: media, myths and realities.
Santa Barbara, CA: Praeger, 2016.
disaggregate mass public shootings, and other forms of mass mur-
20. Chermak SM, Freilich JD, Parkin WS, Lynch JP. American terrorism
der, into different typologies and use comparative analyses to iden- and extremist crime data sources and selectivity bias: an investigation
tify the uniqueness and universal characteristics and behaviors in focusing on homicide events committed by far-right extremists. J Quant
this phenomenon. These, in turn, will help in the creation of policy Criminol 2012;28:191–218.
and crime prevention strategies that can be tailored to specific 21. Huff-Corzine L, McCutcheon JC, Corzine J, Jarvis JP, Tetzlaff-Bemiller
MJ, Weller M, et al. Shooting for accuracy: comparing data sources on
shooters or targets. One possibility is to disaggregate mass public mass murder. Homicide Stud 2014;18(1):105–24.
shootings by targets. For example, our research shows that ideo- 22. Sageman M. Understanding terror networks. Philadelphia, PA: University
logically motivated offenders are significantly more likely to target of Pennsylvania Press, 2004.
government institutions. However, those interested in understand- 23. Parkin WS, Gruenewald J. Open-source data and the study of homicide.
ing the full range of motives, preparation, and event-level behav- J Interpers Violence 2017;32(18):2693–723.
24. Freilich JD, Chermak SM, Belli R, Gruenewald J, Parkin WS. Introduc-
iors behind mass public shootings directed at government ing the United States extremist crime database (ECDB). Terror Polit Vio-
institutions and agents must examine all government attacks and lence 2014;26(2):372–84.
compare them to shooters who did not target government institu- 25. Lankford A. Detecting mental health problems and suicidal motives among
tions. The same comparative analysis could be performed for terrorists and mass shooters. Crim Behav Ment Health 2016;26(5):315–21.
26. Mullen PE. The autogenic (self-generated) massacre. Behav Sci Law
schools, private businesses, and religious institutions. 2004;22(3):311–23.
27. O’Toole ME. The school shooter: a threat assessment perspective.
Collingdale, PA: DIANE Publishing, 2000.
References 28. Silva JR, Capellan JA. The media’s coverage of mass public shootings
in America: fifty years of newsworthiness. Int J Comp Appl Crim Justice
1. Kelly R. Active shooter report: recommendations and analysis for risk 2018. https://doi.org/10.1080/01924036.2018.1437458
mitigation. New York, NY: New York City Police Department, 2012. 29. Department of Homeland Security. Active shooter: how to respond.
2. Blair JP, Martindale MH. United States active shooter events from 2000 Washington, DC: U.S. Department of Homeland Security, 2007.
to 2010: training and equipment implications. San Marcos, TX: Texas 30. Schildkraut J, Hernandez TC. Laws that bit the bullet: a review of leg-
State University, 2013. islative responses to school shootings. Am J Crim Justice 2014;39
3. Osborne JR, Capellan JA. Examining active shooter events through the (2):358–74.
rational choice perspective and crime script analysis. Secur J 2015;30 31. Capellan JA, Lewandowski C. Can threat assessment help police prevent
(3):880–902. mass public shootings? Testing an intelligence-led policing tool Policing
4. Lankford A, Tomek S. Mass killings in the United States from 2006 to 2018. https://doi.org/10.1108/PIJPSM-07-2018-0089.
2013: social contagion or random clusters? Suicide Life Threat Behav 32. Amman M, Bowlin M, Brunell KF, Buckles L, Burton KC, Gibson KA,
2018;48(4):459–67. et al. Making prevention a reality: identifying, assessing, and managing
5. Capellan J. Looking upstream: a sociological investigation of mass public the threat of targeted attacks. Washington, DC: DOJ, FBI, CIRG, 2017.
shootings. In: Daly SE, editor. Assessing and averting the prevalence of 33. Meloy JR, Hoffmann J, editors. International handbook of threat assess-
mass violence. Hershey, PA: IGI Global, 2018;99–128. ment. New York, NY: Oxford University Press, 2014.
6. Capellan JA, Anisin A. A distinction without a difference? Examining 34. Bennett S, Banyard VL, Edwards KM. The impact of the bystander’s rela-
the causal pathways behind ideologically motivated mass public shoot- tionship with the victim and the perpetrator on intent to help in situations
ings. Homicide Stud 2018;22:235–55. involving sexual violence. J Interpers Violence 2017;32(5):682–702.
7. Lankford A, Hakim N. From Columbine to Palestine: a comparative 35. Nicksa SC. Bystander’s willingness to report theft, physical assault, and
analysis of rampage shooters in the United States and volunteer suicide sexual assault: the impact of gender, anonymity, and relationship with
bombers in the Middle East. Aggress Violent Behav 2011;16(2):98–107. the offender. J Interpers Violence 2014;29(2):217–36.
8. McCauley C, Moskalenko S, Van Son B. Characteristics of lone-wolf 36. Gruenewald J, Pridemore WA. A comparison of ideologically-motivated
violent offenders: a comparison of assassins and school attackers. Per- homicides from the new extremist crime database and homicides from the
spect Terror 2013;7(1):4–21. supplementary homicide reports using multiple imputation by chained equa-
9. Capellan JA. Lone wolf terrorist or deranged shooter? A study of ideo- tions to handle missing values. J Quant Criminol 2012;28(1):141–62.
logical active shooter events in the United States, 1970–2014. Stud Con- 37. Hempel A, Meloy JR, Richards T. Offender and offense characteristics
flict Terror 2015;38(6):395–413. of a nonrandom sample of mass murderers. J Am Acad Psychiatry Law
10. Lankford A. A comparative analysis of suicide terrorists and rampage, 1999;27(2):213–25.
workplace, and school shooters in the United States from 1990 to 2010. 38. Blair JP, Schweit KW. A study of active shooter incidents in the United
Homicide Stud 2013;17(3):255–74. States between 2000 and 2013. Washington, DC: U.S. Department of
11. Fridel EE. A multivariate comparison of family, felony, and public mass Justice, 2014.
public murders in the United States. J Interpers Violence 2017. https://doi. 39. Kimmel MS, Mahler M. Adolescent masculinity, homophobia, and vio-
org/10.1177/0886260517739286 lence. Am Behav Sci 2003;46(10):1439–58.
12. Bjelopera JP, Bagalman E, Caldwell SW, Finklea KM, McCallion G. 40. Spaaij R. The enigma of lone wolf terrorism: an assessment. Stud Con-
Public mass shootings in the United States: selected implications for flict Terror 2010;33(9):854–70.

You might also like