ArticlePDF Available

Abstract

This exploratory study examines approximately 1,000 shooting threats made at K–12 schools gathered from publicly available news reports over a 4-year academic period, including prepandemic, pandemic, and postpandemic data. The content analysis finds violent threats increased dramatically in 2021–2022. A majority of individuals who make school shooting threats are male students at large public high schools, which is consistent with those who perpetrate mass shootings at K–12 schools. However, those who threaten shootings are a more diverse population than perpetrators themselves in that they are a wider variety of ages and nearly one fifth are female. Text analysis identified words indicating the specificity of a threat and showed threats tended to be both negative and angry. In 40% of cases, it was unclear if the threat was real or a joke/hoax, yet the most common outcome was to arrest the individual making the threat and charge them with a felony. This study concludes with a call for research to better understand who makes school shooting threats and the challenges to this line of research.
An Exploration of K12 School Shooting Threats in the United States
Jillian Peterson
1
, James Densley
2
, David Riedman
3
,
Jamie Spaulding
1
, and Hannah Malicky
4
1
Department of Criminal Justice and Forensic Science, Hamline University
2
Department of Criminal Justice, Metropolitan State University
3
Department of Criminal Justice, University of Central Florida
4
Department of Peace and Conict Studies, Grinnell College
This exploratory study examines approximately 1,000 shooting threats made at K12
schools gathered from publicly available news reports over a 4-year academic period,
including prepandemic, pandemic, and postpandemic data. The content analysis nds
violent threats increased dramatically in 20212022. A majority of individuals who make
school shooting threats are male students at large public high schools, which is consistent
with those who perpetrate mass shootings at K12 schools. However, those who threaten
shootings are a more diverse population than perpetrators themselves in that they are
a wider variety of ages and nearly one fth are female. Text analysis identied words
indicating the specicity of a threat and showed threats tended to be both negative and
angry. In 40% of cases, it was unclear if the threat was real or a joke/hoax, yet the most
common outcome was to arrest the individual making the threat and charge them with a
felony. This study concludes with a call for research to better understand who makes
school shooting threats and the challenges to this line of research.
Public Signicance Statement
The results of this exploratory examination of 1,000 school shooting threats over a
4-year period show that threats are increasing. Those who threatened school shootings
were a more diverse group than perpetrators of school shootings; and in 40% of cases,
it is unclear if the threat is real or a joke. Future research is needed to understand and
design appropriate intervention strategies for various forms of threats.
Keywords: school violence, school shooting, mass shooting, threat assessment, gun
violence
Supplemental materials: https://doi.org/10.1037/tam0000215.supp
Threats of school shootings are a serious
concern for schools and communities, yet it
is currently unknown the extent to which they
are happening, where they are happening most
frequently, who is making them, the nature of
the threats, and how schools are responding to
them. This exploratory study is the rst step to
understanding the nature and extent of school
shooting threats in the United States in an effort
to identify trends and patterns, inform future
research, and help to promote school safety and
prevent violence from occurring. The present
study examines approximately 1,000 K12
school shooting threats gathered from publicly
available news reports over a 4-year period. It
explores the content of the threats using text-
based sentiment analysis and public reporting
on threats to answer the following research
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Content may be shared at no cost, but any requests to reuse this content in part or whole must go through the American Psychological Association.
This article was published Online First September 14, 2023.
Jillian Peterson https://orcid.org/0000-0002-3331-7691
The authors have no known conicts of interest to disclose.
Correspondence concerning this article should be
addressed to Jillian Peterson, Department of Criminal
Justice and Forensic Science, Hamline University, 1536
Hewitt Avenue, Saint Paul, MN 55105, United States.
Email: jpeterson68@hamline.edu
106
Journal of Threat Assessment and Management
© 2023 American Psychological Association 2024, Vol. 11, No. 2, 106120
ISSN: 2169-4842 https://doi.org/10.1037/tam0000215
questions: Who threatens to commit a school
shooting? Do individuals who threaten a school
shooting look similar to past perpetrators of
K12 school mass shootings and if so, Can this
comparison help differentiate individuals who
truly intend violence from others who merely
want to threaten or draw attention to themselves?
How are schools responding to shooting threats?
Are responses consistent across cases and do
they follow best practices established by threat
assessment and management research?
Prevalence of School Shootings and School
Shooting Threats
The typical U.S. school system consists of
elementary, middle, and high schools that cover
grades kindergarten through 12. Students usually
start in kindergarten (K) at Age 5 or 6 and
complete high school by Age 17 or 18. About
50 million studentswere enrolled in public schools
(funded by local, state, or federal government) in
2021, while private schools (funded by tuition)
had about 5 million students (National Center
for Education Statistics, 2022). That same year,
the K12 School Shooting Database (Riedman,
2023), which, based on media and police reports,
documents whenever a gun is brandished, is red,
or a bullet hits school property for any reason,
regardless of the number of victims, time of day,
or day of weekgoing back to 1970, reported a
record 250 shots red on U.S. school campuses
(Riedman, 2023). That number was more than
double the numbers for 2018, 2019, and 2020,
respectively, and 410 times more than any other
year between 1970 and 2017.
With 130,000 schools nationwide and about
180 school days every year, the statistical odds
of being shot at school are slim, but that is little
solace to thousands of grieving families and
survivors of school gun violence (Cox, 2021),
and new survey results point to rising safety
concerns among parents of students enrolled in
grade school (Brenan, 2022). Most in school
student fatalities are associated with mass kill-
ings: 20 students at the Sandy Hook Elementary
School in Connecticut, 19 students at Robb
Elementary School in Uvalde, Texas, 14 students
at the Marjory Stoneman Douglas High School in
Florida, 12 students at Columbine High School in
Colorado, six students at Red Lake Senior High
School in Minnesota, eight students at Santa Fe
High School in Texas, ve students at Cleveland
Elementary School in California, ve students
at the Amish West Nickel Mines School in
Pennsylvania, four students at Marysville High
School in Washington, and four students at
Westside Middle School in Arkansas (Peterson &
Densley, 2022). The deadliest school shooting of
2021 was on November 30, when 11 people were
shot, four fatally, at Oxford High School, about
45 miles north of Detroit. This mass killing
brought new attention to the responsibility of
schools to handle threats and warning signs
because the weekend before the shooting, the
alleged shooter, a 15-year-old boy, posted photos
of the murder weapon on social media with the
caption: Just got my new beauty today. SIG
SAUER 9 mm(Elamroussi, 2021, para 11).
The day before the Oxford shooting, a school
administrator also called the shootersparents
because their son was allegedly looking at rearms
ammunitiononline during class. His mother texted
her son, LOL Im not mad at you. You have to
learn not to get caught(Peterson & Densley,
2023, para 9). Then, on the day of the shooting, the
parents were called into school because their son
drew a gun on his math worksheet with the words:
The thoughts wontstopHelpmeand Blood
everywhere(Pete rson & Densley, 2023, para 10).
The parents did not immediately take their son out
of school, and later that afternoon, he murdered
four of his classmates.
While shots red at a school are statistically
rare, thousands of school shooting threats are
made every year that come in many formsdirect
communication, social media posts, written plans
and drawings, messages on the bathroom wall,
hearsay, and anonymous 911 calls. Some are real,
while others are hoaxes. Few states and school
districts publicly report the school shooting
threats they receive, making it impossible to
know the true nature and extent of the problem.
Oxford Community School District in the
suburbs of Detroit, Michigan, United States,
typically receives about 500 threats per year, but
following December 2021 shooting at Oxford
High School, 35,000 threats were reported in a
month (Brooks & Brown, 2022). In Texas,
67,440 threats were reported to school ofcials in
20212022 across 16,845 schools total, for an
average of four threats per school (New, 2022).
Schools are reportedly dealing with a rise in false,
automated, swattingthreats (Yousef, 2022),
and in the 20222023 school year, at least nine
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Content may be shared at no cost, but any requests to reuse this content in part or whole must go through the American Psychological Association.
AN EXPLORATION OF K12 SCHOOL SHOOTING THREATS 107
schools in Wisconsin and a dozen schools in
Maine received school shooting threats on the
same day (Jacobo, 2022;WABI News Desk,
2022). On December 17, 2021, schools nation-
wide also closed in response to an anonymous
threat on the social media platform TikTok
warning against National Shoot Up Your School
Day(Riedman et al., 2021, para 4).
Decisions for how to handle school threats
typically begin and end locally with school and
police personnel. Formal threat assessmentis
required by law in 18 U.S. states, encouraged in
ve states, and is in noncodied policy in 16
states (National Association of State Boards of
Education, 2022). Less than half of U.S. schools
have a formal threat assessment team, dened
as a formalized group of persons who meet
regularly with the common purpose of identify-
ing, assessing, and managing students who may
pose a threat of targeted violence in schools,and
only 9% have a threat assessment team that meets
more than once a month (National Center for
Education Statistics, 2018). Many threats made
against schools lack context and other key details
required to conduct a formal behavioral threat
assessment.
School Mass Shootings
Several studies have examined the perpetrators
who actually carry out school mass shootings.
One of the largest studies, The Violence Project
study of mass shootings, was a 3-year study
funded by the National Institute of Justice to
build a database of mass public shootings and
conduct interviews with mass shooting perpe-
trators, families, survivors, and rst responders
(Peterson & Densley, 2021). The Violence Project
Database (VPD; Peterson & Densley, 2022)
tracks mass shootings in which four or more
people were killed in a public space (excluding
the perpetrator), where most of the victims were
nonfamily members and the shooting was not
related to other underlying criminal activity.
VPD includes 186 mass shootings perpetrated
by 190 mass shooters (in four cases, there
were two perpetrators acting together) between
1966 and December 2022, coded on nearly
200 different life history variables (Peterson &
Densley, 2022).
There are a total of 15 perpetrators in the
database who committed a mass shooting at a
K12 school. All of the school mass shootings
in the database were carried out by men or boys,
81% of whom were White (Peterson & Densley,
2022). The median age of those involved in
carrying out the attacks was 17. The majority of
school mass shootings were carried out by a lone
gunman, with just twoJonesboro in 1998 and
Columbine in 1999carried out by two gunmen.
School mass shooters tend to have a connection
to the school they target. Thirteen of the 15 school
shooters in the database were either current
or former students at the school (Peterson &
Densley, 2022).
Three quarters (73%) of school mass shooters
in VPD had a known history of childhood
trauma (i.e., physical, sexual, and emotional
abuse; Peterson & Densley, 2022). Most perpe-
trators (85%) showed signs of a crisis prior to
the shooting and 93% were suicidal before the
shooting or died by suicide in the shooting
(Peterson & Densley, 2022). Most perpetrators
(85%) showed a high degree of planning before
the shooting, and 64% of school mass shooters
showed an interest in previous mass shootings
and studied them through online searches,
documentaries, or books (Peterson et al., 2023).
The most common way for a school mass shooter
to obtain a rearm was to take it from a family or
friend21 of 37 guns (58%) were accessed this
way (Peterson & Densley, 2022). Other studies
of school shooters have found similar results. For
example, the National Threat Assessment Center
(2019) examined 35 perpetrators of targeted
school violence between 2008 and 2017, nding
94% of perpetrators had a history of adverse
childhood experiences and 100% had experienced
recent stressors.
Research shows a majority of K12 school
shooters communicate violent intent ahead of
time, either overtly stating or veiling threats,
and this is a crucial intervention point on the
pathway to violence (Langman, 2021;Meloy
et al., 2014;National Threat Assessment Center,
2019;Peterson & Densley, 2021;Peterson et al.,
2021). In VPD (Peterson & Densley, 2022),
14 out of the 15 school mass shooters leaked
their plans ahead of time (93%). Of those 14
perpetrators who leaked their plans, six told
another person, six posted online, and two wrote it
down somewhere else. The most common person
they leaked their plans to was a classmate. Other
studies have identied leakage 81% of the time
among perpetrators of targeted school violence
(Vossekuil et al., 2004) and 87% of the time
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Content may be shared at no cost, but any requests to reuse this content in part or whole must go through the American Psychological Association.
108 PETERSON, DENSLEY, RIEDMAN, SPAULDING, AND MALICKY
among perpetrators of the 15 deadliest mass
shootings (Lankford et al., 2019).
Prior Research on Threats of
School Violence
Prior research has explored the dynamics of
averted school shootings (e.g., Cowan et al.,
2022;Mads, 2020;National Threat Assessment
Center, 2021;Rocque et al., 2022) and with
funding support from the Community Oriented
Policing Services ofce, the National Police
Foundation (2021) maintains an Averted School
Violence (ASV) database of completed and ASV
narratives from across the country. However,
incidents of violence on school grounds not
related to the school, cases where the motive or
intent to carry out targeted school violence are
unclear, and social media threats not deemed
credible by law enforcement are excluded from
the database. There are 230 cases going back to
1999 in the ASV databasemore than 170
averted incidents and over 60 carried-out attacks
(National Police Foundation, 2021). Yet, a 2022
analysis of just 43 months of SafeOregon tip line
data found 228 tips, which illustrates the size of
the existing data gap (Hendrix et al., 2022). In the
Oregon study, most were direct threats to shoot up
the school made at the school by a male student
(Hendrix et al., 2022). Further, threats made while
at school were more likely to mention a specic
date for the violence, and threats that mentioned a
specic date were more likely to elicit a police
response.
Behavioral threat assessment is a deductive
process to help stop school shootings before they
occur (Follman, 2022;NASP School Safety &
Crisis Response Committee, 2020;National
Threat Assessment Center, 2018;Reddy et al.,
2001). It is focused primarily on a persons
behavior and communications rather than on
specic characteristics, and what they tell us
about that persons potential to do harm (Meloy &
OToole, 2011). Some individuals truly intend
violence (known by some in the threat assessment
eld as hunters), while others merely want to
threaten or draw attention to themselves (how-
lers), for instance (Calhoun & Weston, 2016).
Still, some argue threat assessment may impact
the school environment in ways that make it
feel unsafe or unwelcoming for students
(Kelly, 2018) or it may unduly label students
as potential school shooters, thus contributing to
the school-to-prison pipeline (Whitaker et al.,
2020). While the concept of hunters versus
howlers (Calhoun & Weston, 2016) may have
some utility in certain contexts; moreover, it may
not be an appropriate or effective framework for
talking about children in a school setting.
Studies show threat assessment can improve
campus climate (Fein et al., 2004;Nekvasil &
Cornell, 2015), reduce the use of suspensions and
other punitive disciplinary measures for students
who receive a threat assessment (Maeng et al.,
2020), and increase the use of counseling for
said students (Cornell et al., 2012). An analysis of
1,836 threat assessments also found no signicant
differences in threat assessment outcomes by
race, although Black students were referred for
threat assessment at a higher rate than White
students (Cornell et al., 2018).
One of the reported benets of threat
assessment is that it can help create alternatives
to arrest in the case of school threats (Maeng
et al., 2020), such as diversion, which involves
redirecting the individual who made the threat
away from the criminal justice system and
toward other interventions, such as counseling
or mental health treatment. This approach
recognizes that many individuals who make
threats may be struggling with underlying
mental health or emotional issues, and that
addressing these issues may be more effective in
preventing future harm than simply punishing
the individual (Peterson et al., 2021). Other
alternatives to arrest may include restorative
justice practices or empathy training, which aim
to repair harm and restore relationships between
individuals (van Berkhout & Malouff, 2015),
or community-based interventions, such as
mentoring or job training programs. An open
question explored in this research is just how
often these alternatives to arrest are used in
response to school shooting threats.
Method
A series of daily Google News alerts were set
up to identify news reports that contained
the phrases school shooting threat,”“school
threat,”“school shooting,and school shooter.
Everyday, for 4 years, all media reports published
in the United States referring to a school threat
were reviewed by a trained research assistant.
While using open-source media reports for
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Content may be shared at no cost, but any requests to reuse this content in part or whole must go through the American Psychological Association.
AN EXPLORATION OF K12 SCHOOL SHOOTING THREATS 109
primary data collection has limitations, there is no
government agency that collects, tracks, or
publishes reports on school shooting threats.
Threats of a school shooting were entered into
the database if they met the following criteria:
occurred at a K12 school (i.e., threats to colleges
and universities were excluded), occurred in the
United States (i.e., international threats were
excluded), and threatened a school shooting (i.e.,
bomb threats or general threats of violence were
excluded). Threats were tracked for a 4-year
period that covered academic years 20182019,
20192020, 20202021, and 20212022. The
study was deemed exempt by the institutional
review board at Hamline University because it
only used publicly available information and
there was no risk to human subjects.
Data Coding
Each threat was coded on variables related to
where it occurred (school name, city, state,
region, urbanicity, type of school, enrollment,
public, or private), when the threat occurred (day,
month, year, weekday, time of day), the person
who made the threat (gender, race, age, relation-
ship to school, access to a gun, outcome from
making the threat, criminal charges), and the
threat itself (how it was made, whom it was made
to, the specic language used, type of threat
joke or real). For example, a fth-grade student
in Florida was arrested for texting his friends
a picture of ArmaLite rie (AR-15) ries that he
claimed to purchase after scamming a classmate
out of a billion dollars (Robinson, 2022). The
article includes the name and geographic infor-
mation for the school, when and how the threat
was made, and a description of the threat: boy
texted someone pictures of cash and four AR-15
ries.The article also includes demographic
information about the student who made the
threat, the outcome (arrest), and the fact that
the statement was intended as a joke because the
10-year old did not have 1 billion dollars in cash
or four AR-15 ries.
Ten advanced undergraduate students were
trained on the variables and codebook. Five
reliability cases were initially coded by all research
assistants to check the utility of the codebook and
discuss any coding discrepancies. Each case was
coded initially by one research assistant and then
double checked by a different coder. Basic
frequency data were analyzed using SPSS.
Text and Sentiment Analysis
Text mining, examination, and sentiment
analysis were also conducted using R (R Core
Team, 2022) in this study. First, the threats were
cleaned (i.e., punctuation removed, whitespace
removed) and a corpus was constructed. Next, the
corpus was used to create a term-document matrix
that describes the frequency of term usage that
occurs throughout the threat corpus. The term-
document matrix was then used to assess word
frequency to reect the signicance of a word
throughout the threat collection and develop term
association linkage networks using the count-
based evaluation methods in the text mining
architecture of the tmpackage (Feinerer
et al., 2008).
Next, text-based sentiment analysis was
performed on the corpus using two different
functions integrated from the syuzhetpackage
(Jockers, 2020). The syuzhetpackage provides
four sentiment dictionaries with crowdsourced
lexicons developed by the National Resource
Council Canada for sentiment extraction of
text strings (Jockers, 2020). The rst approach
utilized in this work measured the sentiment
valence (from positive to negative) of the threat
text strings using a dictionary approach (Jockers,
2020). Each threat was assigned a continuous
score within a range from a minimum negative
value (6) to a maximum positive value (+6) for
the sentiment valence of the string. The second
approach consisted of categorical classications
of the threat text strings into the following
sentiment classes: anger, anticipation, disgust,
fear, joy, sadness, surprise, trust, negative, and
positive.
Both approaches for sentiment analysis rely
on lists of words and phrases with positive and
negative connotations. Examination of valence
shifter, that is, negators, ampliers (intensiers),
deampliers (downtoners), and adversative
conjunctions, presence was conducted using
functions provided by the sentimentpackage
(Rinker, 2021). Valence shifters are signi-
cant since their occurrence affects the polariza-
tion of words in the threats, ultimately changing
the meaning of the statement. Additionally,
the equation used by the syuzhetpackage
algorithm to assign value to the polarity of
each sentence rst utilizes a sentiment dictio-
nary to tag and account for polarized words
(Jockers, 2020).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Content may be shared at no cost, but any requests to reuse this content in part or whole must go through the American Psychological Association.
110 PETERSON, DENSLEY, RIEDMAN, SPAULDING, AND MALICKY
Data Analysis
Finally, the similarity of the threats was
examined using functions within the stringdist
package (van der Loo, 2014). Pairwise string
distances between all threats were calculated
using JaroWinkler distance, a string metric used
for the measurement of edit distance between two
sequences. Edit distance is commonly used in
natural language processing as a metric or way of
quantifying how dis/similar two strings are by
counting the minimum number of operations
needed to transform one string into the other
(Winkler, 1990). Using the computed Jaro
Winkler distance, a dendrogram was developed
to demonstrate the hierarchical clustering of the
threats, outlining which threats are most similar in
their construction.
Results
Schools Receiving Threats
The most common states for threats to occur in
this sample were Florida (13.4% of threats) and
California (9.8% of threats), but it is worth noting
that Florida and California are two of the three
largest states by population. Threats were mostly
likely in the South (40.4%), followed by the
Midwest (22.8%), West (20.5%), and Northeast
(16.3%). Threats were more likely at suburban
schools (55.9%) than urban (21.0%) or rural
schools (23.1%). Enrollment at schools receiving
the threats ranged from 41 to 4,788 with a mean of
1,247 students. In total, 98% of threats were made
to public schools. Of the schools receiving the
threats, 4.6% were elementary schools, 22.8%
were middle schools, 62.5% were high schools,
and 10.1% were a combination of grades.
Timing of Threats
A total of 1,038 threats had enough informa-
tion available to be included in this analysis. Only
threats that took place during the school year
between September and June when students were
on campus were analyzed. Table 1 shows the
number of threats by academic year. The fewest
threats took place in the 20202021 academic
year while most students were attending
school virtually owing to the COVID-19 pan-
demic (5.7%). Upon return to the classroom, the
20212022 academic year had the largest number
of threats (61.7%).
The most common months for threats to occur
were transitional months in the school year:
December (21.4%), September (13.6%), October
(12.8%), and January (11.7%). Threats were
mostly likely to occur on Mondays (19.1%) and
Thursdays (19.9%). Of cases where the time of
day was known, 60% of threats took place during
the school day.
Nature of the Threats
Threats were most commonly made online (see
Table 2). Of cases with known data (78.7%),
threats were specic (specied a school shooting
threat) 42.1% of the time and nonspecic 57.9%
of the time. In total, 28.9% of threats were
determined to be genuine, while 31.5% were
jokes or hoaxes. However, it was most often
unclear if the threat was a real/credible threat or a
joke/hoax because in 40% of cases, the nature of
the threat was unknown.
Individuals Making Threats
Threats were most commonly made by a
specic person (67.4%), as opposed to anony-
mous threats (28.8%) or threats coming from fake
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Content may be shared at no cost, but any requests to reuse this content in part or whole must go through the American Psychological Association.
Table 1
Prevalence of School Shooting Threats by Year
School year Frequency % of threats
20182019 186 17.9
20192020 153 14.7
20202021 59 5.7
20212022 640 61.7
Total 1,039 100.0
Table 2
Types of School Shooting Threats
Mode Frequency Percentage
In person 136 13.1
Letter 11 1.1
Other writing 103 9.9
Phone/text 123 11.8
Online/social media 538 51.8
Anonymous tip 19 1.8
Other 23 2.2
Total 953 91.8
AN EXPLORATION OF K12 SCHOOL SHOOTING THREATS 111
online accounts (3.8%). Of cases with informa-
tion available (80.6%), threats were most often
made by a current student (63.6%), followed by
anonymous threats (23.7%), someone unafli-
ated with the school (9.6%), a former student
(2.4%), or teacher (0.8%seven cases).
Of cases with information available (only
41.7% of cases), the person making the threat was
usually male (82.7%), but threats in 17.3% of
known cases came from females. The racial
background was only able to be coded in 7.2% of
cases using publicly available records. The age of
the person making the threat ranged from 5 to 70
with a mean of 16.8 (Figure 1). The vast majority
(75.8%) of individuals who made threats were
minors under the age of 18.
The individual making the threat had
known access to a rearm in 8.6% of cases,
no access to a gun in 16.5% of cases, and in the
majority of cases it was unclear whether or
notthepersonmakingthethreathadaccesstoa
gun (74.9%). For 78.5% of cases with known
outcome data, the most common outcome was
an arrest of the person making the threat (63.7%
of cases). The person making the threat was
not found in 16.4% of cases. The case was
investigatedanddismissedin12.1%ofcases
and there was some other type of intervention in
7.7% of cases.
In cases where an arrest was made, individuals
were charged with a felony, a crime of high
seriousness, 87.0% of the time. Of the 44 threats
that occurred at elementary schools, arrest was
still the most common outcome (students were
arrested in 24 cases).
Threat Text Analysis
For this analysis, to be as inclusive as possible,
language was also used from the 99 cases where
the threat took place outside of the regular school
year, over the summer. There were 396 cases
where the exact language of the threat was
publicly available and examined in this study.
Evaluation of word usage throughout the threat
archive was completed using the term-document
matrix. A bar plot of the most frequently used
words is given in Figure 2. The terms school
and shootwere most frequently used in the
threats. The terms tomorrowand Monday
were also among the most utilized words,
speaking to the timing of the threat relative to
the commission of the incident. The word
frequencies are further exhibited through word
clouds in Figure 3.
Next, the most frequent words from the term-
document matrix were examined to determine if
their occurrence was correlated with one another.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Content may be shared at no cost, but any requests to reuse this content in part or whole must go through the American Psychological Association.
Figure 1
Age of the Individual Making the Threat
112 PETERSON, DENSLEY, RIEDMAN, SPAULDING, AND MALICKY
A network was created using terms that were
correlated as occurring together within the
threats. A plot of the term-document matrix
which visualizes the correlations over 0.5
between frequent terms is given in Figure 4.
Note that terms that occur in combination more
often have thicker edges connecting the terms in
Figure 4. Overall, the term schoolis the most
utilized and connected term across the content, as
evidenced by the thicker edges originating at
that node. Additionally, the word killis used
less frequently (<40 occurrences); however, it is
also highly connected with the other terms when
utilized.
Another important aspect of the content is the
sentiment of the threats. The sentiment extraction
and analysis were performed using the algorithm
and lexicons developed by the National Research
Council. The categorical sentiment classica-
tions of the threats are provided in Figure 5.
The highest sentiment detected was negative;
however, fearand angerwere the highest
emotions detected in the threats. Both negative
and angerwould be expected; however, the
high frequency of fear extracted from the data
is somewhat surprising. The reason that fear
returned such a high score was due to the usage of
the words killand killing,both of which have
very high emotion intensity scores within the
National Research Council lexicon of 0.962 and
0.906, respectively. In the context of a school
shooting threat, these terms would be expected
to be associated with negativesentiment and
angeremotion, which is shown, but is also
inating the fearemotion frequency as detected
by the algorithm. Another key nding is the high
emotion associated with trustin the data. Very
few of the terms associated with trustin the
lexicon were present in the threat data, which
would indicate that the individual was writing
in a polarizing manner. For example, one threat
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Content may be shared at no cost, but any requests to reuse this content in part or whole must go through the American Psychological Association.
Figure 2
Word Usage Frequencies in the Constructed Term-Document Matrix; Most Frequent Terms Shown
Figure 3
Word Cloud Constructed From the Term-Document
Matrix Where Size and Color Indicate Relative
Occurrence
Note. See the online article for the color version of this
gure.
AN EXPLORATION OF K12 SCHOOL SHOOTING THREATS 113
included all of the following: inltrate the school
from all entrances,”“start little by little killing
people,”“should probably steal the teachers
vehicles,and have good armor so we would not
die.illustrates an array of sentiment and
emotion. In this case, the following sentiment
and emotion scores were computed: anger3,
anticipation4, disgust0, fear5, joy1,
sadness4, surprise2, trust3, negative4,
and positive2. Most notable in this threat are
the scores for anticipationand surprisewhile
demonstrating anger,”“disgust,and negative
emotion/sentiment. This trend of many different
emotions was observable throughout the threats.
Additionally, the sentiment was evaluated by
valence by using a dictionary approach where
each threat was assigned a continuous score (from
6to+6). A sentiment score was calculated for
each threat and the scores were aggregated into a
histogram given in Figure 6. The analyses suggest
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Content may be shared at no cost, but any requests to reuse this content in part or whole must go through the American Psychological Association.
Figure 4
Word Association Networks From the Term-Document Matrix by Words Used Greater Than 20 Times (Top) and
40 Times (Bottom)
Note. The linkages depict terms used concurrently in threats.
114 PETERSON, DENSLEY, RIEDMAN, SPAULDING, AND MALICKY
that on average, threats were characterized with
neutral sentiment with some skew to the negative
valence. The combination of sentiments and
emotions lead to approximately 12% of the data
demonstrating text polarity within the sentiment
valence. Of these polarized strings, 14% demon-
strated a negator and an additional 11%
demonstrated an amplier. A negator ips the
sign of a polarized word (e.g., I do not like it),
whereas an amplier or intensier increases the
impact of a polarized word (e.g., I really like it;
Rinker, 2021).
Finally, the threats were examined to deter-
mine pairwise similarity through the calculation
of JaroWinkler distance, or the number of
transformations required to convert one string to
another. Using these distances, threats could be
associated with their closest neighbors hierar-
chically. The diagrammatic representation of the
hierarchical clustering of the threats is given as a
dendrogram in Figure 7. Furthermore, rectangles
were added to the dendrogram around branches to
highlight corresponding clusters of threats. The
largest cluster of threats is given as an excerpt in
Figure 7 (bottom) to demonstrate the lowest level
of association among the threats.
Discussion
This exploratory study looked at approxi-
mately 1,000 threats over a 4-year academic
period, prepandemic, during the pandemic when
many schools were closed, and postpandemic
when schools reopened. Text analysis identied
words indicating the specicity of a threat (i.e.,
tomorrow and Monday) and showed threats
tended to be both negative and angry. The
majority of individuals who make school
shooting threats are male students at large public
high schools, which is consistent with those
who perpetrate mass shootings at K12 schools
(Peterson & Densley, 2021). However, those who
threatened shootings in this study were a more
diverse population than actual shooting perpe-
trators in that they are a wider variety of ages (as
young as Age 5 and as old as 70, Figure 1) and
nearly one fth were female. Perpetrators of
school mass shootings tend to be older teenage
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Content may be shared at no cost, but any requests to reuse this content in part or whole must go through the American Psychological Association.
Figure 5
Text-Based Sentiment Analysis by National Resource
Council Sentiment Type for All Content
Note. See the online article for the color version of this
gure.
Figure 6
Average National Resource Council Sentiment by Threat
AN EXPLORATION OF K12 SCHOOL SHOOTING THREATS 115
current or former students of the school, and they
are 100% male (Peterson & Densley, 2021).
There was a large jump in school shooting
threats after the pandemica 985% increase in
2021/2022 from 2020/2021, and a 244% increase
in 2021/2022 compared to prepandemic rates.
Although in 40% of cases, it was unclear if the
threat was real or a joke/hoax, the most common
outcome was to arrest the individual making the
threat and charge them with a felony. Felony
charges result in lifelong consequences and
juvenile justice involvement is a strong correlate
to future adult offending (Gatti et al., 2009). For
these reasons, the long-term impact of bringing
criminal charges against students for obvious
jokes/hoaxes needs further evaluation.
Limitations
This study is one of the rst to examine over
1,000 threats of school violence quantitatively,
but it has several limitations. One signicant
potential limitation of this study is that the sample
was restricted to media reports, which may not
necessarily reect the full scope and nature of
school shooting threats. It is likely only the most
serious or egregious threats that result in arrests
make the news, thus biasing results. Schools
likely quietly handle thousands of threats of
shootings that do not make the news, making the
totality of threats difcult to study. The study also
relied exclusively on publicly available informa-
tion reported by the media. Even when a case is
reported by the media, often little information is
available (such as the race and age of the person
making the threat), especially because most
people making threats are students at the school
who are underage.
The rise of shooting threats in recent years may
also be an artifact of social media lowering the
threshold for posting threats of violence in some
cases (Peterson et al., 2023). Social media has
become an increasingly important platform for
communication and information sharing, and it is
possible that it has impacted the prevalence and
nature of school shooting threats. Social media
platforms provide a quick and easy way for
people to communicate and share content with a
wide audience, and this can create a sense of
anonymity and detachment that may make it more
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Content may be shared at no cost, but any requests to reuse this content in part or whole must go through the American Psychological Association.
Figure 7
Dendogram of Threats by Threat Index
Note. Complete dendrogram (Top) of threats by threat index illustrating the arrangement of the hierarchical clusters
produced through JaroWinkler distance analysis. A subset (Bottom) is also provided to illustrate the clustering of threats at
lower levels of the dendrogram. See the online article for the color version of this gure.
116 PETERSON, DENSLEY, RIEDMAN, SPAULDING, AND MALICKY
likely for some people to post threatening or
aggressive content (for a review, see Peterson &
Densley, 2017). In addition, social media can also
create a sense of immediacy and urgency, which
may make it more tempting for some people to
post threats or other types of aggressive content in
the heat of the moment, without fully considering
the consequences.
Implications for Research
The process of identifying, evaluating, and
responding to potential threats of violence in
schools is complicated. An exploratory study such
as this can helpgenerate hypotheses forfurther and
necessaryresearch on school shootingthreats. One
important area of research is to identify patterns
and motivations that may differentiate between
those who carry out school mass shootings and
those who merely threaten them. This information
can help to inform law enforcement and school
ofcials on how to respond to threats and develop
effective prevention strategies. For instance, if
the data show that those who threaten school
shootings are often motivated by a desire for
attention online, then interventions that focus
exclusively on providing mental health support
and treatment in person may not be effective in
preventing future incidents.
This study hints at the use of natural language
processing techniques to analyze the text of a
threat and identify keywords or phrases that might
indicate a high level of credibility or risk. Future
research should further explore the potential for
articial intelligence (AI) to analyze the specic-
ity and severity of violent threats to determine
credibility, focusing on the context in which
the threat was made, such as the location, the
relationship between the person making the threat
and the intended target, and any other relevant
factors. Future research is needed using larger
data sets, but the ndings here suggest that AI is
likely to be most effective when used as part of a
larger system that includes human expertise and
judgment. For example, AI might be used to help
prioritize and triage threats, but it would be up
to a team of trained professionals to assess the
risk based on the individual circumstances of the
student in question.
At the same time, generative AI like chat
Generative Pre-Training Transformer could
empower people to craft well-written, automated,
and coordinated threats that are harder for school
ofcials to decode and dismiss. To this end, more
research is needed into the reasons why students
make threats of school shootings with no intention
to follow through. Perhaps threats are a way to gain
attention or notoriety or to copy or emulate others
who have carried out similar attacks in the past. If
so, research is needed to identify the appropriate
threat response for students or school staff,
especially in cases where the threat has been
determined to be a joke or a hoax. Research is also
needed into what types of programming could
prevent this behavior among students. Or perhaps
a threat is a cry for help and an early indication of
risk, providing a key moment for holistic and
individualized intervention.
Implications for Practice
The results of this study are consistent with
other reports of a marked rise in school shooting
threats after the pandemic (Yousef, 2022).
Prior research has found that threats of school
shootings that are leaked online or to other people
are often a cry for helpcorrelated with prior
counseling and suicide attempts (Peterson et al.,
2021). The pandemic had devastating effects
on the mental health and well-being of students
(Sparks, 2022). A nationally representative
survey of high school students conducted by
the Center for Disease Control and Prevention in
2021 found that 37% of high school students
reported they experienced poor mental health,
44% reported they persistently felt sad or
hopeless, and 55% experienced emotional abuse
by a parent in the home (Center for Disease
Control and Prevention, 2022). At the same time,
access to rearms increased as U.S. state gun laws
became more permissive and the United States
saw a historic spike in gun sales during the
pandemic (Helmore, 2021).
In this context, the most common response to
threats in the data was to arrest and criminally
charge the student. It is important to carefully
weigh the risks and benets of this approach.
A punitive response may make students less
likely to report concerns about their classmates.
This can make it more difcult to identify
and intervene in cases of potential violence.
Furthermore, a punitive response can intensify
feelings of crisis and suicidality for the student
making the threat, while intensifying feelings of
anger and grievance toward the school. Students
may be more likely to deny or cover up their
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Content may be shared at no cost, but any requests to reuse this content in part or whole must go through the American Psychological Association.
AN EXPLORATION OF K12 SCHOOL SHOOTING THREATS 117
actions, which can make it more difcult to
identify and address future threats of violence.
Approaches such as Fixated Threat Assess-
ment Centers in Britain may be a promising
approach, where threats are assessed using a
team of both police and mental health profes-
sionals, and a diversion approach is taken to
avoid punitive responses and improve access
to necessary care and resources (Barry-Walsh
et al., 2020).
With so many threats coming from so many
different angles, a national crisis hotline that
functions much like the national poison control
number (staffed 24 hr a day, 7 days a week, free
and condential) could help triage a variety of
threats and provide immediate support and
intervention to those in need. Individuals who
may be contemplating violent actions could be
identied and connected with the appropriate
resources. Several states, such as Colorado, Ohio,
and Pennsylvania, have established versions of
a crisis hotline. A national crisis hotline would
provide a centralized location for such reports
to be made, ensuring that they are promptly
and appropriately addressed as well as tracked
over time. Additionally, laws requiring schools
to report all threats to a centralized agency
for assessment and tracking will help improve
our understanding and create more research
opportunities.
Conclusion
Threats of violence can have serious con-
sequences for the safety and well-being of
students, teachers, and other school staff. They
can have a signicant impact on the overall
learning environment, as well as on the physical
and mental health of a community. By under-
standing the characteristics of threats and the
factors that may contribute to their occurrence,
researchers can develop strategies and interven-
tions to prevent violence from occurring and
keep schools safe. Research that compares
people who perpetrate school mass shootings
with those who threaten them but do not carry
them out is crucial to inform prevention and
intervention efforts. The ndings of this study
provide some initial evidence that those who
threaten school shootings are a more diverse
group than the perpetrators who carry out school
mass shootingsthe question for future research
is why? More information is needed about the
backgrounds of students who threaten shootings
on a large-scale level to be statistically analyzed
to provide data-driven recommendations for
threat assessment and management.
References
Barry-Walsh, J., James, D. V., & Mullen, P. E. (2020).
Fixated threat assessment centers: Preventing harm
and facilitating care in public gure threat cases and
those thought to be at risk of lone-actor grievance-
fueled violence. CNS Spectrums,25(5), 630637.
https://doi.org/10.1017/S1092852920000152
Brenan, M. (2022, September 1). Parent, student
school safety concerns elevated. Gallop.https://news
.gallup.com/poll/399680/parent-student-school-
safety-concerns-elevated.aspx
Brooks, L., & Brown, E. (2022, November 30).
Experts talk school safety measures in the wake of
Oxford shooting. WILX 10.https://www.wilx.com/
2022/11/30/experts-talk-school-safety-measures-
wake-oxford-shooting/
Calhoun, F. S., & Weston, S. W. (2016). Threat
assessment and management strategies: Identifying
the howlers and hunters. Routledge. https://doi.org/
10.1201/b19689-2
Center for Disease Control and Prevention. (2022,
March 31). New CDC data illuminate youth mental
health threats during the COVID-19 pandemic.
https://www.cdc.gov/media/releases/2022/p0331-
youth-mental-health-covid-19.html
Cornell, D., Allen, K., & Fan, X. (2012). A
randomized controlled study of the Virginia student
threat assessment guidelines in grades K-12. School
Psychology Review,41(1), 100115. https://
doi.org/10.1080/02796015.2012.12087378
Cornell, D., Maeng, J. L., Burnette, A. G., Jia, Y.,
Huang, F., Konold, T., Datta, P., Malone, M., &
Meyer, P. (2018). Student threat assessment as a
standard school safety practice: Results from a
statewide implementation study. School Psychology
Quarterly,33(2), 213222. https://doi.org/10.1037/
spq0000220
Cowan, R., Tedeschi, P., Corbin, M., & Cole, R.
(2022). A mixed-methods analysis of averted mass
violence in schools: Implications for professional
school counselors. Psychology in the Schools,
59(4), 817831. https://doi.org/10.1002/pits.22647
Cox, J. W. (2021). Children under re: An American
crisis. Ecco.
Elamroussi, A. (2021, December 4). The shooter
methodically and deliberatelyred at students.
A timeline of a school shooting tragedy. CNN.
https://www.cnn.com/2021/12/04/us/michigan-oxford-
high-school-shooting-timeline/index.html
Fein, R., Vossekuil, B., Pollack, W., Borum, R.,
Modzeleski, W., & Reddy, M. (2004). Threat
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Content may be shared at no cost, but any requests to reuse this content in part or whole must go through the American Psychological Association.
118 PETERSON, DENSLEY, RIEDMAN, SPAULDING, AND MALICKY
assessment in schools: A guide to managing
threatening situations and to creating safe school
climates. U.S. Secret Service, U.S. Department of
Education.
Feinerer, I., Hornik, K., & Meyer, D. (2008). Text
mining infrastructure in R. Journal of Statistical
Software,25(5), 154. https://doi.org/10.18637/jss
.v025.i05
Follman, M. (2022). Trigger points. Dey Street.
Gatti, U., Tremblay, R. E., & Vitaro, F. (2009).
Iatrogenic effect of juvenile justice. Journal of Child
Psychology and Psychiatry, and Allied Disciplines,
50(8), 991998. https://doi.org/10.1111/j.1469-7610
.2008.02057.x
Helmore,E.(2021,May31).U.S. gun sales spiked
during pandemic and continue to rise. The Guardian.
https://www.theguardian.com/us-news/2021/may/31/
us-gun-sales-rise-pandemic
Hendrix, J. A., Planty, M. G., & Cutbush, S. (2022).
Leakage warning behaviors for mass school
violence: An analysis of tips reported to a state
school safety tip line. Journal of Threat Assessment
and Management,9(1), 3351. https://doi.org/10
.1037/tam0000171
Jacobo, V. (2022, October 20). Multiple alleged
swatting callsof false active shooters made across
southeastern Wisconsin school districts. KTVZ.
https://ktvz.com/news/2022/10/20/multiple-alleged-
swatting-calls-of-false-active-shooters-made-across-
southeastern-wisconsin-school-districts/
Jockers, M. (2020, October 29). Extracts sentiment
and sentiment-derived plot arcs from text. Cran.
https://cran.r-project.org/web/packages/syuzhet/
syuzhet.pdf
Kelly, S. R. (2018). The school psychologistsrolein
leading multidisciplinary school-based threat assess-
ment teams. Contemporary School Psychology,22(2),
163173. https://doi.org/10.1007/s40688-017-0153-y
Langman, P. (2021). Warning signs: Identifying
school shooters before they strike. Langman
Psychological Associates.
Lankford, A., Adkins, K. G., & Mads, E. (2019).
Are the deadliest mass shootings preventable? An
assessment of leakage, information reported to law
enforcement, and rearm acquisition prior to attacks
in the United States. Journal of Contemporary
Criminal Justice,35(3), 315341. https://doi.org/10
.1177/1043986219840231
Mads, E. (2020). How to stop school rampage killing:
Lessons from averted mass shootings and bombings.
Palgrave Macmillan Cham. https://doi.org/10.1007/
978-3-030-37181-4
Maeng, J., Cornell, D., & Huang, F. (2020). Student
threat assessment as an alternative to exclusionary
discipline. Journal of School Violence,19(3),
377388. https://doi.org/10.1080/15388220.2019
.1707682
Meloy, J. R., Hoffmann, J., Roshdi, K., &
Guldimann, A. (2014). Some warning behaviors
discriminate between school shooters and other
students of concern. Journal of Threat Assessment
and Management,1(3), 203211. https://doi.org/
10.1037/tam0000020
Meloy, J. R., & OToole, M. E. (2011). The concept of
leakage in threat assessment. Behavioral Sciences &
the Law,29(4), 513527. https://doi.org/10.1002/
bsl.986
NASP School Safety and Crisis Response Committee.
(2020). Behavior threat assessment and manage-
ment: Best practice considerations for K-12
schools, brief overview. National Association of
School Psychologists.
National Association of State Boards of Education.
(2022). State police database: Threat assessment.
https://statepolicies.nasbe.org/health/categories/
physical-environment/threat-assessment
National Center for Education Statistics. (2018). What
are threat assessment teams and how prevalent are
they in public schools? https://nces.ed.gov/blogs/
nces/post/what-are-threat-assessment-teams-and-
how-prevalent-are-they-in-public-schools
National Center for Education Statistics. (2022).
Enrollment in public elementary and secondary
schools, by level, grade, and state or jurisdiction,
Fall 2021.https://nces.ed.gov/programs/digest/d22/
tables/dt22_203.40.asp
National Police Foundation. (2021). Averted school
violence (ASV) database: 2021 analysis update.
Ofce of Community Oriented Policing Services.
National Threat Assessment Center. (2018). Enhancing
school safety using a threat assessment model:
An operational guide for preventing targeted
school violence. U.S. Secret Service, Department
of Homeland Security.
National Threat Assessment Center. (2019).
Protecting Americas schools: A U.S. Secret
Service analysis of targeted school violence. U.S.
Secret Service, Department of Homeland Security.
National Threat Assessment Center. (2021). Averting
targeted school violence: A U.S. Secret Service
analysis of plots against schools. U.S. Secret
Service, Department of Homeland Security.
Nekvasil, E., & Cornell, D. (2015). Student threat
assessment associated with positive school climate
in middle schools. Journal of Threat Assessment
and Management,2(2), 98113. https://doi.org/10
.1037/tam0000038
New, B. (2022, November 18). Texas schools received
more than 67,000 threat reportslast year. CBS
News.https://www.cbsnews.com/dfw/news/texas-
schools-received-more-than-67000-threat-reports-
last-year/
Peterson, J., & Densley, J. (2017). Cyber violence:
What do we know and where do we go from here?
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Content may be shared at no cost, but any requests to reuse this content in part or whole must go through the American Psychological Association.
AN EXPLORATION OF K12 SCHOOL SHOOTING THREATS 119
Aggression and Violent Behavior,34, 193200.
https://doi.org/10.1016/j.avb.2017.01.012
Peterson, J., & Densley, J. (2021). The violence
project: How to stop a mass shooting epidemic.
Abrams Press.
Peterson, J., & Densley, J. (2022). The violence
project database of mass shootings in the United
States.https://www.theviolenceproject.org/mass-
shooter-database/
Peterson, J., & Densley, J. (2023, March 28). Opinion:
Crumbley parents lose their appeal. Why that could
be a game changer on school shootings. CNN.
https://www.cnn.com/2023/03/28/opinions/school-
shooting-crumbley-parents-trial-peterson-densley/
index.html
Peterson, J., Densley, J., Spaulding, J., & Higgins, S.
(2023). How mass public shooters use social media:
Exploring themes and future directions. Social
Media +Society,9(1). https://doi.org/10.1177/
20563051231155101
Peterson, J., Erickson, G., Knapp, K., & Densley, J.
(2021). Communication of intent to do harm
preceding mass public shootings in the United
States, 19662019. JAMA Network Open,4(11),
Article e2133073. https://doi.org/10.1001/jamane
tworkopen.2021.33073
R Core Team. (2022). R: A language and environment
for statistical computing. R Foundation for
Statistical Computing. https://www.R-project.org/
Reddy, M., Borum, R., Berglund, J., Vossekuil, B.,
Fein, R., & Modzeleski, W. (2001). Evaluating risk
for targeted violence in schools: Comparing risk
assessment, threat assessment, and other ap-
proaches. Psychology in the Schools,38(2), 157
172. https://doi.org/10.1002/pits.1007
Riedman, D. (2023). Incidents per year 19702022
[Database record]. K-12 school shooting database.
https://k12ssdb.org
Riedman, D., Densley, J., & Peterson, J. (2021,
December 20). School closures are costly caution
But are still not enough to prevent tragedies. Star
Tribune. https://www.startribune.com/school-closu
res-are-costly-caution-but-are-still-not-enough-to-
prevent-tragedies/600128982/
Rinker, T. (2021, October 13). sentimentr: Calculate
text polarity sentiment (Version 2.9.0). GitHub.
https://github.com/trinker/sentimentr
Robinson, C. (2022). Arrest report: Florida 10-year-
old texted picture of AR-15 ries, said get
ready. The Associated Press. https://www.wesh
.com/article/orida-child-school-shooting-threat/
40150970#
Rocque, M., Gerdes, M., Fox, J. A., Duwe, G., & Clark,
M. (2022). Averting tragedy: An exploration of
thwarted mass public shootings relative to completed
attacks. Criminal Justice Review. Advance online
publication. https://doi.org/10.1177/073401682211
17107
Sparks, S. (2022, January 18). New research shows
how bad the pandemic has been for student
mental health. Education Week.https://www
.edweek.org/leadership/new-research-shows-how-
bad-the-pandemic-has-been-for-student-mental-
health/2022/01
van Berkhout, E., & Malouff, J. M. (2015). The
efcacy of empathy training: A meta-analysis of
randomized controlled trials. Journal of Counseling
Psychology,63(1), 3241. https://doi.org/10.1037/
cou0000093
van der Loo, M. P. J. (2014). The stringdist package for
approximate string matching. The R Journal,6(1),
111122. https://doi.org/10.32614/RJ-2014-011
Vossekuil, B., Rein, R. A., Reddy, M., Borum, R., &
Modzeleski, W. (2004). The nal report and
ndings of the Safe School Initiative: Implications
for the prevention of school attacks in the United
States. U.S. Secret Service, United States
Department of Education.
WABI News Desk. (2022, November 15). Multiple
Maine schools report threats Tuesday. WABI 5.
https://www.wabi.tv/2022/11/15/multiple-maine-
schools-report-threats-tuesday/
Whitaker, A., Torres-Guillén, S., Morton, M., Jordan,
H., Coyle, S., Mann, A., & Sun, W. (2020). Cops
and no counselors: How the lack of school mental
health staff is harming students. American Civil
Liberties Union. https://www.aclu.org/sites/default/
les/eld_document/030419acluschooldiscipline
report.pdf
Winkler, W. E. (1990). String comparator metrics and
enhanced decision rules in the Fellegi-Sunter model of
record linkage. Education Resources Information
Center. https://les.eric.ed.gov/fulltext/ED325505.pdf
Yousef, O. (2022, October 7). False calls about
active school shooters are rising. Behind them is a
strange pattern. NPR.https://www.npr.org/2022/
10/07/1127242702/false-calls-about-active-shooters-
at-schools-are-up-why
Received December 18, 2022
Revision received July 27, 2023
Accepted August 15, 2023
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Content may be shared at no cost, but any requests to reuse this content in part or whole must go through the American Psychological Association.
120 PETERSON, DENSLEY, RIEDMAN, SPAULDING, AND MALICKY
... Prevention research usually draws from deterministic criminological theory to understand the predominant characteristics, triggers, and strains associated with mass and public shooters. For example, many studies examining mass public shooters offer a distinctly offender-oriented framework directed toward prevention (Aitken et al., 2008;Banay, 1956;Bruch, 1967;Hempel et al., 1999;Lankford, 2015a;Lankford & Hakim, 2011;Mullen, 2004;Newman & Fox, 2009;Newman et al., 2004;Palermo, 1997;Peterson et al., 2023;Vossekuil et al., 2004). Most scholars and professional stakeholders are likely to agree that outright prevention of active shootings is favorable to harm mitigation (Knoll & Pies, 2019), since even an attempted active shooting may traumatize witnesses. ...
Article
Research summary Situational crime prevention (SCP) is an environmental crime control perspective with enormous practical and policy relevance due to its practitioner‐friendly theoretical approach. This study examines whether SCP interventions reduce incident casualty outcomes in active shooter incidents. We used an inductive, open‐source data set of 555 active shooter and mass public shooting sites to study the applicability of SCP to active shooter and mass public violence. Our findings suggest a harm mitigation role for SCP: active shooter sites with stronger holistic SCP had fewer casualties. We assessed perpetrator motivation to test displacement, a core critique of SCP, and found that the harm mitigation potential of SCP persists even in the presence of a highly motivated offender. Policy summary SCP could be a practical and effective method to decrease casualties in the event of an active shooting, which is a highly motivated crime type that is difficult to predict and prevent. Public locations may select a range of appropriate SCP techniques based on individual resources and needs. The totality and interactions of these techniques may contribute to public safety in general, with diffuse benefits. This policy solution is highly oriented toward practice and real‐life application, and may be used to supplement existing preventative measures like threat assessment and gun legislation.
... Three studies concentrated on examining school shooting incidents using various methods. Peterson et al. (2023) in a content analysis, investigated how often alternatives to arrest are implemented in response to school shooting threats. Unsurprisingly, the authors found that arrest was the most common method for dealing with individuals making school shooting threats, and 87% of the time, the perpetrator was charged with a felony. ...
Chapter
Full-text available
High-profile school shootings (i.e., rampage school shootings) have fueled empirical and general inquiry into school shooting characteristics and have been a point of emphasis for stakeholders (e.g., parents, teachers, and students) nationwide. The current chapter's goals are to provide a systematic overview of the available evidence examining school shooting incidents. Specifically, this review is intended to gather, summarize, and review relevant peer-reviewed studies on the topic, integrate and identify key findings and themes, identify gaps and limitations in the literature, and discuss implications for directions for future research. Further, it is expected this systematic review can inform practitioners, academics, researchers, policymakers, and stakeholders on the current state of school shooting research and promote innovation in methodological approaches for investigating school shooting incidents.
Article
Full-text available
This mixed-methods study examines social media use among public mass shooters in the United States as an extension of a comprehensive database of 170 mass shooters from 1966 to 2021. Here, we report findings from a systematic content analysis of public data and detailed timelines that were constructed for 44 mass shooters’ social media habits and changes to those habits during the period of time leading up to their shooting. The paper also presents as a case study, a sentiment analysis, and term-linkage network for one perpetrator’s total 3,000 tweets. Several themes were found in the data—there were shooters who changed their posting habits and in some cases, stopped using social media entirely in the lead up to their crime; shooters who used hate speech and were “radicalized” to violence online; shooters with a demonstrable interest in violence, who referenced past mass shooters in their own communications; shooters who exhibited signs of mental illness and suicidality; shooters who were already known to authorities; and shooters who like those described above, actively posted while shooting, presumably to boost their own celebrity status. The findings from this study provide insight into commonalities among mass shooters in terms of their social media usage, which could lead to new pathways for prevention and intervention.
Article
Full-text available
Tip lines are a promising strategy for promoting school safety by creating a structured system where students, staff, parents, and community members can anonymously report safety concerns. To date, little is known about the characteristics of tips, how tips are reported, and what implications for practice can be developed from studying tip line data. We analyzed 228 tips collected over the course of 43 months by the SafeOregon tip line in which one or more persons were reported for signaling intentions or desires to commit an act of mass violence at a school (i.e., “leakage”). Tips were coded and analyzed to generate critical knowledge about the leakage, the person(s) being reported, how the tip was reported, and how the school responded. Given the urgency of timely reporting, we also investigated the odds of same day reporting by several tip and school characteristics. Results show that most tips described direct threats to shoot up the school by a known male student, usually made at school. Notable patterns can be observed by disaggregating tips according to time of day, day of the week, and month, although more research is needed to understand temporal fluctuations. When a specific date for violence was mentioned in the leakage, the reporter was significantly more likely to engage in same day reporting and the police were more likely to be called. Results highlight the importance of training students on how to use tip lines effectively, including describing leakage events clearly and with as much detail as possible.
Article
Full-text available
Importance Understanding the motivation of a mass shooter’s intent to do harm can help practitioners and policy makers develop more effective intervention strategies. Objective To examine the prevalence of communication of intent to do harm, known as leakage, in a sample of 170 mass public shooters from 1966 to 2019; the characteristics of perpetrators who do and do not leak their plans; and whether leakage is a form of fame-seeking behavior or a cry for help among individuals who are in crisis or suicidal. Design, Setting, and Participants This cross-sectional study included perpetrators who killed 4 or more people in a public space from 1996 to 2019 and were included in a comprehensive database of US mass shootings. That database was built from August 2017 to December 2019, and analysis took place from January to May 2021. Main Outcomes and Measures Leakage was identified and coded using publicly available records. Any communication to a third party of an intent to do harm prior to the shooting was coded as leakage. Logistic regression models were used to examine the risk factors associated with leakage. Models estimating leakage were examined to assess the 2 hypothesized pathways to leakage (the cry-for-help model and the fame-seeking model). Results The 170 participants in this sample included 166 (97.7%) male perpetrators and 3 (2.3%) female perpetrators, with a mean (SD) age of 34 (12) years. Overall, 161 participants had known race and ethnicity: 11 (6.8%) Asian individuals, 35 (21.7%) Black individuals, 14 (8.7%) Latinx individuals, 7 (4.4%) Middle Eastern individuals, 3 (1.9%) Native American individuals, 89 (55.3%) White individuals, and 2 (1.2%) individuals with other race and ethnicity. Overall, 79 mass shooters (46.5%) leaked their plans. Of perpetrators who leaked their plans, 35 (44.3%) leaked specific plans about a mass shooting, and 44 (55.1%) leaked nonspecific plans about generalized violence. The study findings indicate that leakage was associated with receiving counseling (odds ratio, 7.0; 95% CI, 2.0-24.8) and suicidality (odds ratio, 3.7; 95% CI, 1.0-13.6), suggesting that leakage may best be characterized as a cry for help from perpetrators prior to their act. Conclusions and Relevance In this study, nearly half of the perpetrators of mass shootings leaked their plans. Leakage was associated with receiving counseling and suicidality. Leakage is a critical moment for mental health intervention to prevent gun violence. Opportunities to report threats of violence need to be increased. Traditional threat assessment models focused on specific threats of violence may miss critical opportunities for intervention.
Book
Full-text available
An examination of the phenomenon of mass shootings in America and an urgent call to implement evidence-based strategies to stop these tragedies Using data from the writers’ groundbreaking research on mass shooters, including first-person accounts from the perpetrators themselves, The Violence Project charts new pathways to prevention and innovative ways to stop the social contagion of violence. Frustrated by reactionary policy conversations that never seemed to convert into meaningful action, special investigator and psychologist Jill Peterson and sociologist James Densley built The Violence Project, the first comprehensive database of mass shooters. Their goal was to establish the root causes of mass shootings and figure out how to stop them by examining hundreds of data points in the life histories of more than 170 mass shooters—from their childhood and adolescence to their mental health and motives. They’ve also interviewed the living perpetrators of mass shootings and people who knew them, shooting survivors, victims’ families, first responders, and leading experts to gain a comprehensive firsthand understanding of the real stories behind them, rather than the sensationalized media narratives that too often prevail. For the first time, instead of offering thoughts and prayers for the victims of these crimes, Peterson and Densley share their data-driven solutions for exactly what we must do, at the individual level, in our communities, and as a country, to put an end to these tragedies that have defined our modern era.
Article
Full-text available
The rediscovery of the importance of mental illness in the risk assessment and management of those who threaten, approach, and harm public figures has led to a new way of dealing with those that threaten public figures. This approach emphasises the role of “fixation” which may be defined as an intense preoccupation pursued to an abnormally intense degree. It integrates a threat assessment paradigm with the literature on stalking. The need for such an approach was highlighted in research on the prevalence of harassment of public figures. Psychiatry has a key role in this approach which sees mental health clinicians and Police work together in Fixated Threat Assessment Centres (FTACs). An FTAC functions by assessing the level of concern and sharing information to facilitate interventions that are often mental health based. The purpose is not the hopeless task of identifying those who will go on to perpetrate serious violence, rather to intervene in the group they emerge from, to prevent harm. As well as decreasing risk to the persons fixated upon, this approach improves care to the mentally disordered people who harass and threaten them and, in doing so, decreases the likelihood of their criminalization while enhancing their quality of life. As expertise in the area has grown, policing and security agencies in several countries have expanded the FTAC model to cover individuals thought at risk of lone-actor grievance-fueled violence, a term that captures both different forms of mass killing and lone actor terrorism.
Article
Scholarship on mass public shootings has increased in recent years as comprehensive datasets have become more available. As a result, much is known about the contextual and offender related characteristics of such attacks. However, less research has been conducted on attacks that were planned but ultimately did not occur. Understanding how mass public shootings may be thwarted or averted is important for both policy and theoretical reasons. In this paper, we describe a new dataset of averted mass public shooting threats (N = 194) from 2000–2019 and compare them to mass public shootings that were completed during this time (N = 97). Several noteworthy findings emerged, including that nearly half of the averted cases were reported by a friend or acquaintance, most targeted a specific location or group, and averted cases were more likely to involve school targets and co-offenders. Implications are discussed.
Article
The purpose of this study is to examine how recent incidents of planned mass violence in high schools within the United States were successfully averted. Using the Averted School Violence database, this study utilizes a mixed-methods content analysis approach to analyze 82 incidents of averted mass violence in high schools that occurred post-Columbine era. The following three themes were identified: attack motives and logistics, plot discovery, and attack aversion. Implications for professional school counselors in preventing similar mass violence attacks in the future are discussed, including fostering trusting relationships between students and adults, providing education to parents about social media platforms and proper firearms storage, establishing threat assessment teams, and educating parents, students, and school personnel about warning signs for violence. Key points • • School counselors play key roles in preventing mass violence as a part of their comprehensive school counseling programs. • • Based on a mixed-methods content analysis, implications for school counselors in preventing mass violence attacks are discussed.
Book
Dr. Langman’s timely book presents practical, research-based guidance on anticipating and preventing mass attacks. Warning Signs is for all members of a school community, including school personnel, mental health professionals, law enforcement officers, parents, and students. Drawing on lessons learned from dozens of thwarted plots and deadly attacks, Warning Signs provides the insight necessary to keep schools safe.
Book
This book tackles the important question of how we can understand and learn from the school rampage killings that have been prevented. In the flood of recent accounts and analyses of deadly school rampage killings that plague society and inspire widespread public fear, very little attention has been given to the incidents that almost were. Building on Madfis’ previous book, The Risk of School Rampage: Assessing and Preventing Threats of School Violence (2014), this vital work addresses key gaps in school violence scholarship through the examination of averted school rampage incidents in the United States and advances existing knowledge through ground-breaking insights from the latest research on mass murder, violence prevention, bystander intervention, disciplinary policy, and threat assessment in school contexts. This empirical study utilizes in-depth interviews conducted with school and police officials (administrators, counselors, security guards, police officers, and teachers) directly involved in averting potential school rampages to explore the processes by which threats are assessed and school rampage plots are thwarted. Madfis finds that many common contemporary school violence prevention policies and practices are ineffective at preventing rampage attacks and may actually increase the likelihood of their occurrence. Rather than uncritically adopting such problematic approaches, Madfis argues that schools must model prevention practices upon what has proven successful in averting potentially deadly incidents.
Article
Threat assessment has been proposed as a method for schools to respond to student threats of violence that does not rely on exclusionary discipline practices (e.g., suspension, transfer, expulsion, arrest). The present study compared disciplinary consequences for 657 students in 260 schools using the Comprehensive Student Threat Assessment Guidelines (CSTAG) with a comparison group of 661 students in 267 schools using a more general threat assessment approach. The odds that students receiving a threat assessment in CSTAG schools would receive a suspension (OR = 0.59) or law enforcement action (OR = 0.47) were less than those in schools using a general approach. Students in CSTAG schools were expelled at lower rates (0% versus 1.7%) than students in comparison schools. These results indicate that schools using the CSTAG model are less likely to respond to student threats with exclusionary discipline.