AMERICAN MASS SHOOTERS AND SUICIDE 

By 

Kylei Elizabeth Brown 

 

 

 

 

 

 

 

 

 

 

 

 

A THESIS 

Submitted to 

Michigan State University 

in partial fulfillment of the requirements 

for the degree of 

Criminal Justice—Master of Science 

2018 

 

ABSTRACT 

AMERICAN MASS SHOOTERS AND SUICIDE 

By 

Kylei Elizabeth Brown 

Mass shootings are traumatic events associated with highly publicized and often 

gruesome homicides that raise public alarm. Yet, researchers know relatively little about shooters 

and mass shooting events. Therefore, it is important to further extend the research of mass 

shootings to characterize different types of shooters and how they ultimately perpetrate their 

attacks. As such, the current study investigated the differences between American mass shooters 

who commit suicide after their attacks and mass shooters who do not. This study examined 

shooter vitality in 185 mass shootings in America perpetrated by 194 mass shooters between 

2000 and 2016. Each shooter was categorized as having survived the attack or the manner in 

which they died; whether by self-inflicted suicide or being shot by another person. This revised 

data about how the mass shooters died was compared with thirty-three different independent 

variables that looked at the characteristics of the mass shooters as well as the characteristics of 

the mass shootings using both analysis of variance and binary logistic regression analytic 

techniques. At the same time, an original version of this dataset was tested alongside the revised 

dataset to look for similarities and comparisons in the results. The results of the revised dataset 

showed that mass shooters with a prior criminal record are more likely to survive their attacks 

and those who kill more victims are less likely to survive their attacks. These results are useful to 

inform both law enforcement personnel intervening in a mass shooting as well as lawmakers 

developing policies in an attempt to decrease the number of mass shootings in America. 

 

 

 

 

 

 

Copyright by 

KYLEI ELIZABETH BROWN 

2018 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

This thesis is dedicated to Mom and Dad. 

Thank you for your continued love and support, and for always believing in me. 

 

 

 

 

 

 

 

iv

 

 

 

 

 

 

ACKNOWLEDGEMENTS 

 

Thank you to Dr. Christina DeJong for acting as my mentor throughout this arduous process. I 

honestly do not know if I would have ever completed this thesis if it were not for your ongoing 

guidance and assistance. I would also like to thank Christine Kwiatkowski for meticulously 

editing every word that I wrote.  

 

 

 

 

 

 

 

 

 

v

 

 

 

 

 

 

 

TABLE OF CONTENTS 

 

LIST OF TABLES…………….……….………………………………………………………...vii 

American Mass Shooters and Suicide……………………………….………….…………………1 

Literature Review……….……………………………….……………….………………...……...4 
Defining Mass Shootings……….……….……….………….…………………………….4 
 
 
Information about Mass Shooters……………...………………….……………….……...5 
Suicide………………………………………………………………………….…….……7 
 

 

Variables for the Current Study…………………………………………….…….……….9 

Data and Methods………………………………………………….………….…………...…….14 
Data………………………………………………………………………………………14 
 
Variables….………………….…………………………….…………………………….16 
 
 
Subjects…………………………….………………….…………………………………18 

 

Analysis….………………….…………………………………………………………....23 

Results….……………………………………………….…………...…………………………...28 

Discussion….…………………………………….……….………….…………….….…………30 
 
Major Findings…….……………………………………….………….…………………31 
Hypotheses…...…………………………………….………………………………….…31 
 
 
Limitations…...…………………………………………….………………………….…33 

 

Future Directions……………………………….………….…………………………….36 

BIBLIOGRAPHY………………………………………………………………………….…….38 

 

 

 

 

 

 

 

vi

LIST OF TABLES 

 

 

Table 1: Mass shooter descriptive statistics from the revised dataset and the original dataset….19 
 
Table 2: Mass shooting descriptive statistics from the revised dataset and the original dataset...21 
 
Table 3: ANOVA of the mass shooter characteristics for the revised dataset and the original 
dataset…………………................................................................................................................23 
 
Table 4: ANOVA of the mass shooting characteristics for the revised dataset and the original 
dataset…...……………………………………………………………………………………….25 
 
Table 5: Binary logistic regression using the revised dataset and the original dataset….……….28 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

vii

American Mass Shooters and Suicide 

 

On a pleasant fall night, commemorating the first day of October, thousands of people 

were enjoying a Jason Aldean concert with friends and family when their lives were instantly 

changed forever. A 64-year-old man opened fire from his 32nd-floor window at the Mandalay 

Bay Resort and Casino. He killed 58 people and injured more than 500 before he turned the gun 

on himself and pulled the trigger (Liston, Wan, Somashekhar, & Davis, 2017). To this day, the 

2017 Las Vegas shooting is the deadliest mass shooting in American history and, like other mass 

shootings, fueled a national conversation about mass shooters, their motivations, and how they 

can be stopped.  

A mass shooting is a large-scale homicide event that involves multiple people dying in 

one event. After a mass shooting, numerous questions are asked by politicians, media personnel, 

and the public, wondering how this could have happened, why it happened, and how a similar 

event can be prevented in the future. While all of these questions are important and worthwhile 

to answer, people generally do not ask other important questions following a mass shooting, such 

as, why the mass shooter committed suicide or not, or how mass shooters that commit suicide 

may be different than those who do not. Are incidents with mass shooters who commit suicide 

deadlier? Do those who commit suicide and those who do not choose similar targets? Do they 

have similar backgrounds? 

 

The number of mass shootings in America also appear to be on the rise. Research by the 

Congressional Research Service found that the prevalence of mass public shootings “…increased 

in the 1970s and 1980s, and continued to increase, but not as steeply, during the 1990s and 200s, 

and first four years of the 2010s” (Krouse & Richardson, 2015, para. 5). Mass shootings clearly 

impact the survivors of such attacks and their families, along with the families of the deceased. 

 

1

In addition, the effects of mass shootings are potentially amplified by vigorous media coverage 

of these events. Through different types of media, such as social media, television, and news 

articles, the public is captivated by the stories of mass shootings, which increases their coverage 

and sociocultural power. For instance, the constant replaying of a shooting incident could 

reinforce trauma in the bystanders in the forms of survivor’s guilt or post-traumatic stress 

disorder (PTSD). The severity of these effects may vary with different types of shooters, such as 

those who commit suicide versus those who do not, but such details remain unknown.  

Although a significant social problem, there is little quantitative and empirical research 

on mass shooters. One important limitation of mass shooting research is the lack of availability 

of systematically collected data. Indeed, policymakers often base policy recommendations on 

anecdotal or hastily collected data. There are few datasets available to study this topic, and even 

fewer that have passed close academic scrutiny. Mass shooting events are difficult to study 

because they are always examined in retrospect with information drawn from highly limited data 

sources. Due to the numerous limitations of previous research, there are few studies comparing 

different types of mass shooters, and only one known study (Lankford, 2015) that compares 

shooters who commit suicide to those who do not. However, other research in related areas, such 

as terrorism, provide insight into potential differences between attackers that may apply to mass 

shooters.  

Terrorism is similar to mass shootings in that both of these types of murderers, 

“…planned their attacks for months or years before finally striking, and some of them also left 

behind notes or explanations in which they claimed to be fighting for a cause,” but differs 

because these events are ideologically motivated (Lankford, 2014, p. 356). Despite differences, 

however, the terrorism literature can inform the study of mass shootings because there is more 

 

2

research examining variation in plots and perpetrators, including difference between attackers 

who commit suicide versus those who do not. For example, research by Freilich and colleagues 

(2017) looked at the differences between American jihadist and far right terrorists who did and 

did not commit suicide in the following their attacks. Researchers showed that, “…suicide 

mission offenders were nearly 3 times more likely to have served in the military, 13 times more 

likely to have received movement training, and almost 2.5 times more likely to have a high 

commitment to their ideology,” (Freilich et al., 2017, p. 11) when compared to non-suicide 

missions. Additionally, researchers found that members of Al Qaeda who committed terrorist 

attacks were more likely to kill themselves when compared to far-right terrorists. Together, these 

findings suggest that those who perpetrate terrorist attacks vary in the presence and severity of 

certain risk factors by suicide status and in the likelihood of committing suicide. More generally, 

this provides preliminary evidence that perpetrators of mass homicides differ depending on 

suicidality, a finding that can be applied to mass shooter research.  

In sum, it is important to compare mass shooters who commit suicide to mass shooters 

who do not commit suicide because there are many unanswered questions about shooter 

suicidality that could reveal differences in risk and/or crime perpetration that could guide 

prevention efforts for this growing social problem. As such, the current study will explore how 

mass shooters who commit suicide differ from those who do not. The comparative focus is on 

whether mass shooter suicide (or not) is affected by the location of the mass shooting, age of the 

shooter, region of the country, marital status of the shooter, prior criminal record of the shooter, 

the number of victims, or the number of mass shooters involved in the event. Additionally, some 

of the issues that will be explored in the current research include the nature of the mass shooter’s 

suicide (such as killing themselves vs. suicide by police, the timing of the suicide) and the 

 

3

motive for committing suicide (such as wanting to be a martyr for a larger cause). Policy 

implications and future directions will be discussed. 

Literature Review 

Defining Mass Shootings 

Mass shootings are not a new form of violence, but rather have existed in the United 

States for over a century. Schildkraut and Elsass (2016) reported that the first mass shooting in 

the United States occurred at a school in Baton Rouge, Louisiana in May of 1880. Since then, 

mass shootings have occurred in various locations, such as workplaces, nightclubs, private 

homes, universities, public streets, and movie theaters, and most people believe that the 

prevalence of mass shootings is increasing. Given the diversity of events that have been 

considered a mass shooting, these events have become difficult to define and there is no 

consensus on a scholarly definition (Schildkraut & Elsass, 2016). Some scholars argue that mass 

shootings have to involve at least four people being killed—not including the perpetrator—while 

others argue that only three people need to be killed, or that only three people need to be 

wounded. Other definitional issues that are debated in the scholarly literature is the length of 

time between killings, the motivation behind the killing, and whether or not the shootings occur 

at one location (Schildkraut & Elsass, 2016). 

In their review, Schildkraut and Elsass (2016) systematically evaluated how other 

scholars have defined mass shooters, weighed the pros and cons of different definitions, and 

provided what is a more inclusive and more complete definition of mass shooting:    

A mass shooting is an incident or targeted violence carried out by one or more shooters at 

one or more public or populated locations. Multiple victims (both injuries and fatalities) 

are associated with the attack, and both the victims and the location(s) are chosen either 

 

4

at random or for their symbolic value. The event occurs within a single 24-hour period, 

though most attacks typically last only minutes. The motivation of the shooting must not 

correlate with gang violence or targeted militant or terroristic activity (p. 28).  

This definition identifies several difficult issues in creating a cohesive definition of a mass 

shooting. By focusing on the issues of the time period around a mass shooting, the motivations 

behind a shooting that would not qualify as a mass shooting, and the number of mass shooters, 

this mass shooting definition by these authors could considerably help people understand what a 

mass shooting is and what it is not. This definition of a mass shooting because is arguably the 

most amalgamated definition of a mass shooting that can easily be found in the literature. It has 

been incorporated into published literature, showing that it is reliable in the studying of mass 

shootings. 

Information about Mass Shooters 

Though little is known about mass shooters, there are a few noteworthy studies that focus 

on mass shootings independently of other crime. In this work, the characteristics of mass 

shooters and how they commit their crimes have been described. For example, Schildkraut and 

Elsass (2016) identified 304 mass shootings committed by 314 perpetrators in the United States 

from 1880 through 2016. They found that just over twice as many mass shooters were white than 

black and that the mass shooters tended to be male (Schildkraut & Elsass, 2016). In another 

study by Fox and DeLateur (2014), researchers found that the majority of mass shootings are 

committed with semiautomatic handguns, accounting for approximately 1% of the country’s gun 

homicides.  

Mass shootings are considered a type of mass murder, and the majority of available data 

is in the general mass murder literature. Mass murder is a broad category of homicide that is 

 

5

defined as, “a number of murders (four or more) occurring during the same incident, with no 

distinctive time period between the murders,” and may include events involving all manners of 

death in addition to firearm violence, or mass shootings (Federal Bureau of Investigation, 2008, 

p. 8). Therefore, research that discusses mass murder, in general, provides a reasonable amount 

of background information on this topic, one must look to the broader topic of mass murders.  

In a recent study of 308 perpetrators of mass murder in America between 2006 and 2014, 

Lankford (2015) examined the differences between mass murderers who died after their attacks 

and those that survived. Of these 308 mass murderers, the differences between the perpetrators 

who lived after their attack and those who died after their attack were examined, with 94 

perpetrators dying after their mass murder and 214 living after their mass murder. Findings 

suggest that perpetrators of mass homicide differed on several measures depending on whether 

they lived or died after their attack. On average, the age of the perpetrators who died after their 

attack was 36.5-years-old while the average age of the perpetrators who lived after their attack 

was 30.2-years-old. Mass murderers who died after their attack killed an average of 5.5 victims 

while mass murderers who lived after their attack killed an average of 4.6 victims. Perpetrators 

who died after their attack were less likely to have a co-offender than perpetrators who lived 

after their attack. Taken together, these results suggest that perpetrators of mass homicide differ 

in their demographic characteristics, the severity of their attacks, and some situational 

characteristics (e.g., number of offenders) depending upon whether they live or die after an 

attack. While this research by Lankford (2015) includes mass shooting events, a more detailed 

analysis is necessary to replicate findings for mass shooters, specifically, while controlling for 

other factors, such as the type of location where the mass shooting occurred, the region of the 

 

6

country where the mass shooting happened, if the mass shooter had a prior criminal record, and 

the marital status. 

Research in the mass murder literature also describes psychosocial factors that contribute 

to violence risk. For example, Knoll (2012) describe common risk factors across incarcerated 

mass murderers, including extreme feelings of anger and revenge, social alienation, rumination 

on violent revenge fantasies, variable psychiatric illness, precipitating social stressors, and 

significant planning before the offense. In addition, researchers found that the perpetrators were 

all bullied as children, saw themselves as loners, and often seemed open to death, feeling it as 

giving them power and infamy (Knoll, 2012). Though this information represents only those who 

lived after their attack, it is likely that mass murderers who died after their attack also possessed 

some of these risk factors. 

Suicide 

 

At some point in their lives, roughly 9% of people report having significant thoughts of 

suicide and 3% attempt suicide (Borges et al., 2010). In order to understand why a mass shooter 

would commit suicide, it is important to understand suicide itself and the factors that could lead 

anyone to commit suicide. As part of the World Health Organization’s World Mental Health 

Survey Initiative, Borges and colleagues (2010) administered 108,705 surveys to adult 

participants in twenty-one countries between 2001 and 2007 to determine risk factors for 

attempting suicide. Researchers found that every mental health disorder in the study was strongly 

correlated with suicidal ideation (Borges et al., 2010). Some of these disorders are panic 

disorder, agoraphobia without a history of panic disorder, generalized anxiety disorder, specific 

phobia, social phobia, and post-traumatic stress disorder, to name a few. It was also discovered 

that those more likely to display suicidal behaviors were participants of the female sex, were of a 

 

7

younger age, had a lower education and income, were unmarried, were unemployed, their 

parents had a history of psychopathology, they dealt with childhood adversities, had the presence 

of one of the mental disorders aforementioned, and had psychiatric comorbidity (Borges et al., 

2010). 

 

Other research has examined the pattern of suicidal behavior across geographical space to 

determine if different regions of the United States are riskier for suicide. One such study 

conducted by Kessler and colleagues (2005) investigated trends in suicide ideation, suicide plans, 

suicide gestures, and suicide attempts during two different time periods; 1990 to 1992 and 2001 

to 2003. Researchers found that the region of the country was not statistically significant in their 

relation to suicidal ideation, plan, gesture, or attempt (Kessler, Berglund, Borges, Nock, & 

Wang, 2005). This negative result could be because suicidal behavior truly does not differ by 

region, or it is possible that regional analyses may have prevented finding finer geographical 

patterns, such as a specific state or county. Thus, the possible relationship between suicide and 

region of the country should be further reconnoitered. 

In addition to the known characteristics of suicidal behavior, research also shows that 

suicidal individuals interact with the health care system. In their retrospective study of 5,894 

individuals during the year prior to their suicide, Ahmedani and colleagues (2014) found that 

64% went to a primary care appointment and 62% went to a medical specialty appointment, but 

none received a mental health diagnosis. Moreover, within the week before their deaths, 25% 

went to a medical specialist and 21% went to a primary care doctor, but, again, did not receive a 

mental health diagnosis (Ahmedani, Simon, Stewart, Beck, Waitzfelder, Rossom, Lynch, Owen-

Smith, Hunkeler, Whiteside, Operskalski, Coffey, & Solberg, 2014). This is important because 

medical professionals have the knowledge and resources to intervene in suicide but are still 

 

8

missing critical prevention opportunities. Overall, these findings reveal a potential novel point of 

intervention in mass shooting events if suicidality is both important to perpetration and potential 

shooters to the doctor prior to the commission of their crime. 

It is very difficult to know the exact reason why a mass shooter would choose to take his 

life after he commits a violent act because those who committed suicide are deceased. 

Sometimes, the reason for a suicide will be left in a letter, social media post, or video, but it is 

often unknown. In these cases, researches investigate factors associated with shooter suicide. 

Findings from a study of school mass shooters showed that most of the male mass shooters who 

committed suicide after their attacks were continually bullied and beaten up because their 

aggressors believed the shooters were homosexual (Kalish & Kimmel, 2010). It is certain that 

not all of the school mass shooters in United States have been homosexuals, but the majority of 

these shooters were said to be different or odd compared to other boys by their peers, which may 

have lead them to being haphazardly labeled by others according to malicious cultural 

stereotypes. 

Variables for the Current Study 

The current study seeks to determine the role of shooter suicidality in mass shooting 

events and both event and shooter characteristics associated with shooter suicide. For this reason, 

the current study will examine variables about the mass shootings and variables about the mass 

shooters in an effort to understand the role of shooter suicidality in mass shooting events. The 

variables studied will be pulled from the existing literature about mass shootings and suicide. 

The primary outcome of interest—and thus, the dependent variable—in the current study 

is whether or not the shooter in each mass shooting event committed suicide. This variable was 

created from data regarding shooter vitality and the manner of the shooter’s death, if applicable. 

 

9

This variable was selected to extend previous work in terrorism and mass murder research that 

suggest differential offender profiles and attack severity depending upon shooter survival status. 

In line with previous research on suicide, terrorism, and mass murder, the current study 

investigates the relationship between shooter suicide and other risk factors for mass shooting 

perpetration as well as event severity. As such, the mass shooting event independent variables of 

this study—also referred to as the mass shooting characteristics—include the region of the 

country, the location type, if multiple shooters were involved, how many people were killed, how 

many people were wounded, the total number of people killed and wounded, if multiple weapons 

were used, and the types of weapons used (handgun, shotgun, rifle, or other). The independent 

variables that focus on the individual mass shooters—denoted as the mass shooter 

characteristics—include the age of the mass shooter, race, gender, marital status, and whether or 

not they have a prior criminal record. The decision to focus on these variables was driven by 

what was readily available in the database and what could be collected reliably. This will be 

further discussed in the data and methods section of the paper. 

 

Age has a known influence on both the likelihood of committing suicide and dying after a 

mass murder event (Lankford, 2015). Evidence suggests that certain age groups are more likely 

to commit suicide compared to others. Specifically, Shah (2012) noted that data from the World 

Health Organization shows that in almost every country, suicide rates tend to increase as people 

age. As mentioned in the review of the work of Lankford (2015), the average age of the mass 

murderers who died after their attack was 36.5 years old, while the mass murderers who survived 

their attack had an average age of 30.2 years old. Another article noted that in a study comparing 

adult and adolescent mass murderers, 53% of the adult mass murderers but only 9% of the 

adolescent mass murderers committed suicide or were killed by the police (Meloy, Hempel, 

 

10

Gray, Mohandie, Shiva, & Richards, 2004). Therefore, age may be one variable that makes a 

mass shooter, specifically, more or less likely to commit suicide after their attack.  

 

Though inconsistent, some evidence suggests that geographical location affects the 

likelihood of suicide, which may also have important implications for mass shooting 

perpetration. Though Kessler and colleagues (2005) concluded that region did not correlate 

significantly with suicide, research by Stack (2000) found that community-level economic 

stressors are strongly related to people committing suicide. The inconsistency between these 

findings may be related to the size of the geography analyzed with variation masked in the study 

considering larger regions. In addition, other factors that might vary by region include weather 

patterns, population density, and unemployment rates. For example, if mass shooters live in 

regions that are economically deprived, they may be more likely to commit suicide, which may 

affect the nature of their attack. Therefore, the current study will explore if there is a relationship 

between geographical region and mass shootings.  

 

Marital status has a known role in suicide risk (Kposowa, 2000), and a potential role in 

mass murder perpetration as an element of social isolation or psychosocial adversity (Knoll, 

2012). Kposowa (2000) found that people who were divorced or separated were more than twice 

as likely to commit suicide as those who were married. Being widowed or single, however, had 

no significant interaction with suicide rates (Kposowa, 2000). Given its role in suicide risk, 

marital status may be correlated with mass shooter suicide.  

 

Criminal history is a known risk factor in suicide risk. Thus, this study posits that a prior 

criminal record may contribute to the likelihood of a mass shooter suicide. In a large-scale 

investigation of suicide in South Carolina utilizing multiple data sources, Weis and colleagues 

(2006) found that 196 of 491 suicides in 2004 involved individuals with a prior criminal record. 

 

11

Given that nearly 40% of all 491 suicides in the 2004 dataset represented people with a prior 

criminal history, criminal history needs to be considered as a potential risk factor for suicide, 

more broadly, and mass shooter suicide.  

 

The size of a mass shooting is depicted by the media based on how many victims were 

harmed or killed. Therefore, it is noteworthy to examine how many victims there were for a mass 

shooting event, including victims who were both murdered and injured. In his comparison of 

mass murderers who survived and those that died after their attack, Lankford (2015) found that 

mass murderers who did not survive their attack killed an average of 5.5 victims while mass 

murderers who survived their attack murdered an average of nearly 4.5 victims. Since that study 

focused on the suicide of mass murderers, and mass shootings are considered a type of mass 

murder, research should be conducted to determine if the correlations between age and suicide of 

mass murderers are similar to the age and suicide of mass shooters. 

 

The research from Lankford (2015) was also useful in regard to discussing the number of 

shooters and suicide of mass murderers. Through a comparison between mass murderers that 

survived their attack and those that died after their attack, it was divulged that, “Holding all other 

independent variables constant, for each additional co-offender, a mass murderer was 4.8 times 

more likely to survive his or her crime” (Lankford, 2015, p. 592). If this result was found in mass 

murderers, it would be expected that a similar result would be found in mass shooters, which was 

aforementioned as a type of mass murder. 

 

The final independent variable included in the analysis is the race of the mass shooter. 

Schildkraut and Elsass (2016) found that, out of all the mass shootings that have happened in the 

United States between 1880 and 2016, approximately 55% were white, 25% were black, and the 

other twenty percent of mass shooters were members of racial minority groups. In another study 

 

12

of the racial make-up of mass murderers and their likelihood to die after a mass murder, 

researchers found that mass murderers who were white were the most likely to die after a mass 

murder (Lankford, 2016). While this study broadly categorized mass murderers as alive or dead 

following their attack and did not consider how a mass murderer died, it would be useful to delve 

deeper to compare how the mass murderers died with the race of the mass murderers in order for 

more comparisons to be drawn. The current study will do exactly this, rather with mass shooters 

instead of mass murderers. 

 

Though there has been some research on the suicidality of mass shooters, this topic is 

severely lacking an adequate amount of quantitative research. Therefore, the purpose of this 

study is to analyze factors that differentiate mass shooters who commit suicide after their attack 

from mass shooters who do not. The factors included in this study are location type, region, how 

many shooters were involved, the age of the shooter, the race of the mass shooter, the marital 

status of the shooter, the number of victims, and if the mass shooter has a prior criminal record. 

Thus, the primary research question of this study asks what differences exist between the mass 

shooters who commit suicide after their attacks and mass shooters who do not. Below are the 

five hypotheses, based on the research presented previously in this section, that will be examined 

through the statistical analysis of the data. 

H1: Mass shooters who commit suicide following their attack are more likely to be older 

than mass shooters who survive their attacks. 

H2: Mass shooters who are divorced or separated will be more likely to die than survive 

their mass shooting event. 

H3: Mass shooters who die after their attack are more likely to have a prior criminal 

record. 

 

13

H4: Mass shooters who murder more victims during their attack will be more likely to die 

afterwards. 

H5: Mass shooters who are white will be more likely to die after their attack than mass 

shooters of another race.  

Data and Methods 

Data 

 

To further study what variables are associated with a shooter committing suicide (or not) 

after a mass shooting, a dataset created by Schildkraut (2014) was used. The original dataset 

from Schildkraut (2014) included 91 mass shootings that occurred between the year 2000 and 

2012, but the updated and expanded version includes cases through 2016. The updated version 

was used for this study with 226 mass shootings in the database that occurred between the years 

2000 and 2016. 

 

The data was collected by obtaining media information about the different mass 

shootings in the United States through systematic searches on LexisNexis. While there are other 

datasets that are publicly available, they are often flawed. For instance, the Mother Jones mass 

shooting spreadsheet only selects a few spree mass shootings to be included in their dataset and 

fails to explain why. This dataset does not randomly omit mass shooting events and includes 

thoroughly explained inclusion and exclusion criteria. 

Although the dataset presented by Schildkraut (2014) primarily looks at the media’s 

coverage on mass shootings, these data can be used to examine other aspects of mass shootings.  

The dataset is organized chronologically, and includes the date, city, state, location type, region 

of the country, name of the shooter, age of the shooter, race of the shooter, gender of the shooter, 

how many shooters were involved in the event, how many people were killed in the event, how 

 

14

many people were wounded in the event, if the shooter survived the attack, if multiple weapons 

were used, if a handgun was used, if a shotgun was used, if a rifle was used, and if another type 

of gun was used (Schildkraut, 2014, p. 55). This version of the dataset used includes 226 cases 

between 2000 and 2016.  

In Schildkraut and Elsass’ (2016) definition of a mass shooting, they do not give a 

specific number of people who need to be wounded or killed in one specific event for the event 

to be considered a mass shooting.  In the discussion below, these data will be reviewed to as the 

‘original dataset.’ The original dataset consists of 226 mass shootings committed by 236 mass 

shooters.  As there is considerable debate about how many victims should (or should not) be 

included in the definition, this research will compare the results from analysis with the ‘original 

dataset’ with results generated by restricting the number of victims.  For comparison purposes, 

only mass shootings involving, “three or more shooting victims” that are not necessarily fatalities 

(Stanford Geospatial Center, 2016).  This dataset will be referred to as the ‘revised dataset.’  The 

revised dataset consists of 185 mass shootings by 194 mass shooters.  

In addition, the following variables were added to the original and revised datasets to test 

hypotheses related to suicidality among mass shooters, including how the shooter died (if 

applicable), prior criminal record of the mass shooter, and the marital status of the shooter. It is 

important to get more background information on the shooter(s) so that more precise conclusions 

can be drawn about suicidality in mass shooters. To obtain the necessary data to include the new 

variables introduced in this study, searches for each mass shooting incident were conducted 

using open sources. The searches for the mass shootings began by using Google or LexisNexis. 

For the mass shootings that were heavily reported, it was not uncommon to see twenty or more 

articles pertaining to one mass shooting incident. Therefore, when available, articles were 

 

15

primarily read from The New York Times. When articles were not written by The New York 

Times about a particular mass shooting incident, other popular news websites were browsed 

including CBS News, Fox News, ABC News, The Los Angeles Times, and The Chicago Tribune. 

In the rare instance that two different sources varied on a mass shooter’s criminal history or 

marital status, additional sources were read and the answer for that mass shooter that would be 

chosen would be the answer that was provided the most frequently. 

Variables 

 

The dependent variable for this research is ternary coded, consisting of mass shooters 

committing self-inflicted suicide, mass shooters who died from being shot by another person, 

and mass shooters who survived their mass shooting. All mass shooters that successfully took 

their own lives were coded as committing suicide. It could be argued that mass shooters who 

were killed by the police, or another person, should be included as committing suicide because 

often they want to die and use suicide by cop as a way of accomplishing death (de Similien & 

Okorafor, 2017). However, it cannot be assumed that those mass shooters who were killed by 

someone else wanted to die after their attack because some of these mass shooters may have 

attempted to flee the scene or others may have been under the influence of alcohol or drugs. 

Also, not all of the mass shooters who committed self-inflicted suicide or were shot by another 

person after their attack left behind a suicide note or video to detail their intention to die. 

Although arguments for suicide by cop are reasonably compelling, to avoid inserting 

assumptions, those who were killed by another person were given their own category in the 

dependent variable.  

 

Regarding the creation of the binary logistic regression for the results section of this 

study, the dependent variable was altered from having three categories (self-inflicted suicide, 

 

16

shot by another person, and survived) to only consisting of two categories (died or survived). 

This change was made because the results of the category ‘shot by another person’ had a small 

number of cases and thus, it would be difficult to make confident conclusions about that data. It 

was also determined that the ‘shot by another person’ category did not yield significant results in 

the ANOVAs, thus making it unnecessary to include this category in the binary logistic 

regression. With all of this in mind, the best possible option was to collapse the three categories 

of the dependent variable into two categories for the binary logistic regression. 

 

In this research, the independent variables are the location type of mass shooting, age, 

region of the country, marital status, known criminal record, the number of offenders, the gender 

of the mass shooter, the race of the mass shooter, the number of killed victims, the number of 

wounded victims, the total number of victims killed and wounded, weapons used (shotgun, rifle, 

handgun, and/or other type of gun), if multiple weapons were used, and if the mass shooter was 

deceased after the mass shooting. The dependent variable of this study is how the mass shooter 

died after their attack, such as through self-inflicted suicide, being shot by another person, or 

surviving the event. 

Several demographic variables about the mass shooters will be explored. The age of the 

shooter will be a continuous variable, so the mean will be used. For the variable of multiple 

perpetrators being involved, this will be binary coded as either yes or no. The gender of the 

shooter will either be coded a male or female with female serving as the reference category. For 

marital status, four categories will be coded: married, divorced/separated, widowed, or single. 

Single perpetrators will be the reference category. Relating to this variable is the variable of how 

the shooter died, with the possible choices of self-inflicted suicide, shot by another person, or 

survived. The race of the mass shooter will also be included. Race will be coded into six 

 

17

categories, consisting of white, black, Latino, Asian, Native American, or other. The race 

category of “other” will be used when a perpetrator is multiracial, the race of the perpetrator 

could not be found, or when the perpetrator is a race that is not listed in the other categories. 

White perpetrators will serve as the reference category. Lastly, the prior criminal record of the 

mass shooter will be binary coded as either no or yes. 

The other variables relate to the nature of the incident. The type of mass shooting 

category consists of specific locations, including the following; school, workplace, place of 

worship, restaurant/nightclub, residential, shopping/entertainment, government/military, 

multiple/spree, or other. This variable will be dummy coded with schools serving as the 

reference category. The region of the country variable consists of the Northeast region, the 

Southern region, the Western region, and the Midwest region. (U.S. Census Bureau, 2013). The 

Northeast region will serve as the reference category. The number of perpetrators per mass 

shooting event will also be collected, along with the number of victims killed, the number of 

victims wounded, and the total number of victims that were either killed or wounded. The 

different types of weapons that were used in a mass shooting—including handguns, shotguns, 

rifles, and other weapons—will all be binary coded as no or yes. 

Subjects 

 

This revised dataset study sample, focused on the revised dataset, includes 185 mass 

shootings that happened in the United States of America from 2000 through 2016. Incidents were 

included if they conformed to the aforementioned definition of a mass shooting. These 185 mass 

shooting events involved 194 mass shooters. Additionally, the original dataset information is 

included, which consisted of 226 mass shooting events occurring from 2000 to 2016 with 236 

mass shooters. The descriptive statistics for the variables about the mass shooters and the mass 

 

18

shootings can be found in Table 1 and Table 2, respectively. This information was broken up into 

two different tables due to the large number of variables involved in this study. Table 1 observes 

the following characteristics about the mass shooters: the age, race/ethnicity, the gender, if the 

shooter had a criminal record prior to the mass shooting, if the shooter died after the mass 

shooting, and how the mass shooter died, with an option of “survived” for those who did not die 

after a mass shooting. 

Table 1: 
Mass shooter descriptive statistics from the revised dataset and the original dataset 

 

Revised 

Original 

Variables 

N 

Frequency  Mean/Percentage 

SD 

N 

Frequency  Mean/Percentage 

SD 

Age 

194 

Race/Ethnicity  194 

White 

Black 

Latino 

Asian 

Native 
American 

Another Race 
or Ethnicity 

 

 

 

 

 

 

Gender 

194 

Female 

Male 

 

 

Marital Status 

194 

Single 

Married 

 

 

Separated or 
Divorced 

194 

Widowed 

Shooter Dead 
Afterwards? 

No 

Yes 

 

 

 

 

How the 
Shooter Died 

194 

Self-Inflicted 
Suicide 

 

 

 

 

 

92 

56 

22 

7 

4 

13 

 

7 

187 

 

126 

68 

 

135 

28 

30 

1 

 

79 

 

35.16 

2.04 

47.4% 

28.9% 

11.3% 

3.6% 

2.1% 

6.7% 

14.536  236 

1.414 

234 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.187 

236 

3.6% 

96.4% 

 

 

 

 

 

0.478 

236 

64.9% 

35.1% 

0.47 

69.6% 

14.4% 

15.5% 

0.5% 

 

 

 

 

0.770 

236 

 

 

 

 

 

 

 

 

 

0.493 

236 

 

 

113 

72 

25 

8 

5 

11 

 

9 

227 

 

159 

77 

 

167 

36 

32 

1 

 

35.16 

2.00 

47.9% 

30.5% 

10.6% 

3.4% 

2.1% 

4.7% 

0.96 

3.8% 

96.2% 

0.33 

67.4% 

32.6% 

0.44 

70.8% 

15.3% 

13.6% 

0.4% 

0.56 

40.7% 

 

 

 

 

 

104 

44.1% 

 

 

14.934 

1.443 

 

 

 

 

 

 

0.192 

 

 

0.470 

 

 

0.738 

 

 

 

 

0.498 

 

 

 

19

Table 1 (cont’d) 

 

Shot by 
Another 
Person 

 

115 

 

59.3% 

 

 

 

 

 

132 

 

55.9% 

 

 

Survived  194 

 

0.99 

0.910 

236 

 

1.06 

0.907 

 

 

The ages of the mass shooters in the revised dataset study ranged from 12-years-old to 

74-years-old with a mean age of 35.16-years-old. Approximately 50% of the shooters were white 

and slightly more than a quarter of the shooters were black. The majority of the mass shooters 

had no prior criminal record. Almost five times as many mass shooters were found to be single, 

rather than divorced or separated. After their mass shooting events, roughly 60% of the mass 

shooters died, whether it was by their own hand or someone else’s. Of those mass shooters that 

died after the mass shooting, 41.8% killed themselves while 17.5% were killed by someone else. 

 

Table 1 also shows the original dataset descriptive statistics of mass shooters. It is 

presented that, in the original dataset, the mean age of mass shooters was 35.16-years-old. The 

age of the mass shooter ranged from 12 to 88-years-old. The majority of the mass shooters were 

found to be white and almost all of them were males, similar to the mass shooters of the revised 

dataset. Most of the mass shooters in this dataset did not have a prior criminal record and were 

single; a statistic also seen in the revised dataset. Unlike the revised dataset, more of the mass 

shooters in the original dataset survived their mass shootings, rather than committing self-

inflicted suicide. 

The next table, Table 2, shows the descriptive statistics for the 185 mass shooting events 

that occurred between 2000 and 2016 in the revised dataset, along with the 226 mass shooting 

events that occurred between 2000 and 2016 in the original dataset. This table includes the 

following variables about the mass shootings: the region of the country, the location type, the 

number of people killed, the number of people wounded, the total number of people killed and 

 

20

wounded, if there were multiple shooters involved, if multiple weapons were used, if a handgun 

was used, if a shotgun was used, if a rifle was used, and if another gun was used. 

Table 2: 

Mass shooting descriptive statistics from the revised dataset and the original dataset 

 

Revised 

Original 

Variables 

N 

Frequency 

Mean/ 

SD 

N 

Frequency  Mean/Percentage 

SD 

Percentage 

 

1.057 

226 

13.5% 

31.4% 

22.7% 

32.4% 

 

 

 

 

 

 

 

 

 

2.894 

226 

16.2% 

27.0% 

3.8% 

9.2% 

4.9% 

8.1% 

5.4% 

14.1% 

11.4% 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

29 

75 

49 

73 

 

41 

62 

8 

18 

9 

19 

15 

26 

28 

 

25 

58 

42 

60 

 

30 

50 

7 

17 

9 

15 

10 

26 

21 

2.73 

12.8% 

33.2% 

21.7% 

32.3% 

4.36 

18.1% 

27.4% 

3.5% 

8.0% 

4.0% 

8.4% 

6.6% 

11.5% 

12.4% 

1.050 

 

 

 

 

2.930 

 

 

 

 

 

 

 

 

 

Region 

185 

Northeast 

South 

Midwest 

West 

 

 

 

 

Location 
Type 

185 

 

 

 

 

 

 

 

 

 

School 

Workplace 

Place of 
Worship 

Restaurant or 
Nightlife 

Residential 

Entertainment 
or Shopping 

Government 
or Military 

Spree 

Another 
Location 

Number of 
Victims 
Killed 

Number of 
Victims 
Wounded 

Total Number 
of Killed and 
Wounded 
Victims 

Multiple 
Shooters? 

185 

696 

3.76 

5.135 

226 

733 

3.24 

4.787 

185 

812 

4.39 

6.917 

226 

857 

3.79 

6.393 

185 

1,508 

8.15 

10.64

226 

1,590 

7.04 

9.915 

5 

185 

 

 

0.178 

226 

 

0.03 

0.161 

No 

Yes 

 

 

Multiple 
Weapons? 

185 

179 

6 

 

96.8% 

3.2% 

 

 

 

 

 

0.485 

226 

220 

6 

 

97.3% 

2.7% 

0.32 

 

 

0.469 

 

21

Table 2 (cont’d) 

Handgun 
used 

No 

Yes 

 

 

185 

No 

Yes 

 

 

Shotgun used 

185 

No 

Yes 

 

 

Rifle used 

185 

No 

Yes 

 

 

Another 
Weapon used 

185 

No 

Yes 

 

 

 

 

 

116 

69 

 

38 

147 

 

150 

35 

 

135 

50 

 

184 

1 

 

62.7% 

37.3% 

 

 

 

 

 

 

 

0.405 

226 

20.5% 

79.5% 

 

 

 

 

 

0.393 

226 

81.1% 

18.9% 

 

 

 

 

 

0.445 

226 

73.0% 

27.0% 

 

 

 

 

 

0.074 

226 

99.5% 

0.5% 

 

 

 

 

 

153 

73 

 

51 

175 

 

187 

39 

 

166 

60 

 

225 

1 

 

67.7% 

32.3% 

0.77 

22.6% 

77.4% 

0.17 

82.7% 

17.3% 

0.27 

73.5% 

26.5% 

0 

99.6% 

0.4% 

 

 

 

0.419 

 

 

0.379 

 

 

0.443 

 

 

0.067 

 

 

The revised dataset descriptive statistics in Table 2 showed that the majority of the mass 

shooting events occurred in the Western region of the United States, but they were also very 

likely to occur in the Southern region. 27% of the mass shootings were workplaces shootings and 

16.2% of mass shootings took place at a school. The least likely location type for a mass 

shooting was shown to occur at places of worship, with only 3.8% of all mass shootings 

occurring there. Per mass shooting event, the average number of victims killed was 3.76 and the 

average number of victims wounded was 4.39, with an average total of 8.15 people wounded or 

killed per mass shooting. Approximately 3% of the mass shootings involved more than one mass 

shooter and 37.3% of the mass shooters only used one weapon to commit their mass shootings. 

The most common weapon used in a mass shooting was a handgun, which is consistent with 

previous findings in the literature (Fox & DeLateur, 2014). 

Table 2 also shows the descriptive statistics for the original dataset. Unlike the mass 

shooting descriptive statistics for the revised dataset, the majority of the mass shootings from the 

 

22

original dataset occurred in the South, rather than in the West. Most of the mass shootings in the 

original dataset occurred in workplaces and schools; similar to the result found for the revised 

dataset in Table 2. The mean number of people killed per mass shooting event for the original 

dataset was 3.24, with an average of 3.79 people being wounded each event. The most common 

weapon of choice used in the original dataset was a handgun; a finding that also supports the 

claim by Fox and DeLateur (2014) that the most commonly used type of gun in mass shootings 

are handguns.  

Analysis 

The data were analyzed by completing bivariate comparisons and multivariate logistic 

regression. The bivariate comparisons were completed by running analysis of variance tests, also 

known as “ANOVAs.” The ANOVAs were generated using SPSS. Two different ANOVAs were 

used to assess the differences in the means of the variables for both datasets; with one ANOVA 

focusing on the mass shooter characteristics for the revised and original datasets and another 

ANOVA looking at the mass shooting characteristics for the revised and original datasets. Two 

different ANOVAs were used due to the large number of independent variables in this study. The 

first of these ANOVAs looked at the characteristics of the mass shooters in the population and is 

shown in Table Three. The other type of ANOVA observed the characteristics of the mass 

shooting events and is located in Table Four. 

Table 3: 

ANOVA of the mass shooter characteristics for the revised dataset and the original dataset 

 

Variables 

Self-Inflicted 

Suicide 
(n=81) 

36.88 

 

Age * 

 

 

Revised 

Shot by 
Another 
Person 
(n=34) 

37.47% 

 

Survived 
(n=79) 

Self-Inflicted 

Suicide 
(n=90) 

32.41 

37.21 

 

 

Original 

Shot by 
Another 
Person 
(n=42) 

37.69 

 

Survived 
(n=104) 

32.37 

 

23

Table 3 (cont’d) 

Gender (% 
Male) 

 

98% 

Race/Ethnicity 

 

White 

55.56% 

Black 

24.69% 

Latino 

Another Race 

Marital Status 

9.88% 

9.88% 

 

Single 

69.14% 

Married 

13.58% 

Separated or 
Divorced 

16.05% 

Widowed 

1.23% 

Prior Criminal 
Record 

27% 

 

97% 

 

35.29% 

32.35% 

11.76% 

20.59% 

 

58.82% 

20.59% 

20.59% 

0% 

38% 

 

95% 

 

44.30% 

31.65% 

12.66% 

11.39% 

 

74.68% 

12.66% 

12.66% 

0% 

42% 

 

96% 

 

52.81% 

29.21% 

10.11% 

7.87% 

 

68.89% 

15.56% 

14.44% 

1.11% 

24% 

 

98% 

 

42.86% 

28.57% 

11.90% 

16.67% 

 

59.52% 

21.43% 

19.05% 

0% 

40% 

 

96% 

 

46.60% 

33.01% 

10.68% 

9.71% 

 

76.92% 

12.50% 

10.58% 

0% 

37% 

º= p<0.05 for the “revised dataset” and *= p<0.05 for the “original dataset” 

 

The ANOVA, shown in Table 3, did not yield any significant results for the revised 

dataset. Although none of the results were significant, some of the results highlighted important 

trends in the outcome variables that may have occurred due to chance. The variable of age shows 

that younger mass shooters could be more likely to survive a mass shooting compared to the 

older mass shooters, who could be more likely to be shot by another person or commit suicide. 

For the variable of race, black mass shooters could be less likely to commit suicide than be shot 

by another person, if they died. Conversely, white mass shooters could be more likely to commit 

suicide than be shot by another person. Mass shooters that are single could be more likely to 

survive a mass shooting than be shot by another person. Being married, separated, or divorced 

makes a mass shooter potentially more likely to be shot by another person than to survive a mass 

shooting. Finally, mass shooters with a prior criminal record could be more likely to survive a 

mass shooting, or be shot by another person, than commit suicide.  

 

The Table 3 ANOVA found one significant result in the original dataset. The results of 

this ANOVA indicated that at least one mean age is different from the others across the manners 

 

24

of death. With the mean age of a mass shooting surviving being 32.37-years-old—compared to 

the ages of committing self-inflicted suicide and being shot by another person of 37.21-years-old 

and 37.69-years-old, respectively—this result was found to be statistically significant. While age 

was the only significant result, the other independent variables yielded some interesting trends 

that were not significant. A white mass shooter could be more likely to commit self-inflicted 

suicide, compared to being shot by another person or surviving. A mass shooter, whose race is 

not white, black, or Latino, could be more likely to be shot by another person, rather than 

commit self-inflicted suicide or surviving their mass shooting. Being married could also make a 

mass shooter more likely to be shot by another person, rather than committing self-inflicted 

suicide or surviving. Lastly, a mass shooter with a criminal record could be more likely to 

commit self-inflicted suicide instead of being shot by another person or surviving.  

Table 4: 

ANOVA of the mass shooting characteristics for the revised dataset and the original dataset 

 

Variables 

Self-Inflicted 

Number of 
Victims Killed 
* 

Number of 
Victims 
Wounded 

Total Number 
of Killed and 
Wounded 
Victims 

Multiple 
Shooters? 

Multiple 
Weapons? * 

Handgun used 

Shotgun used 

Rifle used 

 

Revised 

Shot by 
Another 
Person 
(n=33) 

5.06 

Survived 
(n=72) 

Self-Inflicted 

Suicide 
(n=89) 

2.79 

3.84 

Original 

Shot by 
Another 
Person 
(n=41) 

4.22 

Survived 
(n=96) 

2.27 

Suicide 
(n=80) 

4.10 

3.66 

5.76 

4.57 

3.34 

4.88 

3.75 

7.76 

10.82 

7.36 

7.18 

9.10 

6.02 

0% 

41% 

79% 

20% 

26% 

6% 

48% 

82% 

24% 

39% 

6% 

28% 

79% 

15% 

22% 

25

0% 

40% 

81% 

18% 

25% 

5% 

41% 

78% 

22% 

37% 

4% 

21% 

74% 

15% 

24% 

Table 4 (cont’d) 

Another 
Weapon used 

Region 

0% 

 

Northeast 

12.35% 

South 

25.93% 

Midwest 

28.40% 

West 

33.33% 

Location Type 

 

School º * 

14.81% 

Workplace º * 

37.04% 

Place of 
Worship 

3.70% 

 

3% 

 

8.82% 

32.35% 

23.53% 

35.29% 

 

5.88% 

14.71% 

0% 

 

0% 

 

15.19% 

39.24% 

16.46% 

29.11% 

 

26.58% 

20.25% 

5.06% 

 

0% 

 

12.36% 

28.09% 

26.97% 

32.58% 

 

14.61% 

39.33% 

4.49% 

 

2% 

 

7.32% 

34.15% 

24.39% 

34.15% 

 

7.32% 

9.76% 

0% 

 

0% 

 

15.63% 

37.50% 

15.63% 

31.25% 

 

26.04% 

23.96% 

4.17% 

Restaurant or 
Nightlife 

Residential * 

Shopping or 
Entertainment 

Government or 
Military * 

6.17% 

11.76% 

11.39% 

5.62% 

12.20% 

8.33% 

4.94% 

7.41% 

11.76% 

8.82% 

1.27% 

7.59% 

4.49% 

7.87% 

9.76% 

7.32% 

1.04% 

9.38% 

3.70% 

11.76% 

3.80% 

3.37% 

17.07% 

5.21% 

Spree 

11.11% 

Another 
Location 

11.11% 

20.59% 

14.71% 

15.19% 

8.86% 

8.99% 

11.24% 

17.07% 

19.51% 

11.46% 

10.42% 

º= p<0.05 for the “revised dataset” and *= p<0.05 for the “original dataset” 

 

 

The ANOVA, shown above in Table 4, did yield two significant results for the revised 

dataset. First, there was a significant effect of the school location type on the dependent variable 

of how the mass shooter died at the p<0.05 level. There was also a statistically significant 

difference between the workplace location type and how the mass shooter died at the p<0.05 

level. While not significant, the ANOVA also detailed some interesting trends across the 

categories of the dependent variable. The ANOVA showed that the more victims a mass shooter 

killed, the more likely they could be to be shot by another person, rather than committing 

suicide. The same was found for the variable of the total number of victims killed or wounded; 

the more victims killed or wounded, the more likely the mass shooter could be to be shot by 

another person, compared to suicide. No mass shooters involved in a mass shooting with more 

 

26

than one shooter committed suicide, and for the mass shootings that involved more than one 

shooter, they could be equally as likely to survive the mass shooting or be killed by another 

person. If a mass shooter used multiple weapons, that mass shooter could be less likely to survive 

than commit suicide or be killed by another person. Few cases involved the use of a rifle, but for 

the mass shooters who did, they could be more likely to be shot by another person than to 

survive a mass shooting. The region variables yielded interesting results from the ANOVA. In 

the South, mass shooters could be almost four times as likely to commit suicide than be shot by 

another person. School mass shooters could be more likely to survive mass shootings, compared 

to being shot by another person, mass shooters committing workplace shootings could be more 

likely to commit suicide than be shot by another, and spree shooters were could be likely to be 

shot by another person than commit suicide. 

 

The ANOVA in Table 4 generated three significant results for the original dataset. The 

first of these was that a significant effect exists between the independent variable of the school 

location type and the dependent variable of how the mass shooter died at the p<0.05 level. It was 

also determined that there was a statistically significant difference between the workplace 

location type and how the mass shooter died. The final significant result was that there was a 

significant effect of the government/military location type on how the mass shooter died. Along 

with the statistically significant results, other trends were exhibited in the ANOVA that may 

have arisen due to chance and thus, are not significant. Mass shootings that involved multiple 

shooters could make those shooters less likely to commit self-inflicted suicide, rather than be 

shot by another person or surviving. The usage of multiple weapons in a mass shooting could 

make a mass shooter less likely to survive a mass shooting, instead of committing self-inflicted 

suicide or being shot by another person. A mass shooter that uses a rifle could be more likely to 

 

27

be shot by another person, compared to surviving or committing self-inflicted suicide. Finally, 

mass shooters that attack in the Midwest could be less likely to survive than committing self-

inflicted suicide or being shot by another person.  

Results 

Binary logistic regression was used to determine the relationship between the 

independent variables with the dependent variable. Two factors of the dependent variable—self-

inflicted suicide and shot by another person—were collapsed into one category referred to as not 

surviving. Understandably, those who survived their mass shooting were in the survival 

category. A binary logistic regression was used instead of a multinomial logistic regression 

because the case size of the factor ‘shot by another person’ was diminutive and therefore, could 

have adversely affected the results. The independent variable of the gender of the mass shooter 

was excluded because the variable is highly skewed due to the nature of its distribution: which is 

caused by approximately 96% of the mass shooters being males. The binary logistic regression 

table is shown below in Table 5. Table 5 displays the binary logistic regression for the revised 

dataset and original dataset. In these statistics, the reference group of the dependent variable was 

not surviving and thus, the results are discussed in the terms of surviving a mass shooting. 

Dummy variables were created to make the variables of white, single, region of the country, and 

location type from the variables of race, marital status, region of the country, and location type, 

respectively.  

Table 5: 
Binary logistic regression using the revised dataset and the original dataset 

 

 

Revised 

Original 

Variables 

B 

Age 

White 

Latino 

-0.013 (0.013) 

0.358 (0.420) 

0.370 (0.576) 

B 

-0.013 (0.011) 

0.426 (0.366) 

0.259 (0.537) 

Exp(B) 

0.987 

1.532 

1.295 

Exp(B) 

0.987 

1.431 

1.447 

28

Table 5 (cont’d) 

 

Another Race 

0.496 (0.617) 

Separated or 

Divorced 

0.038 (0.496) 

Prior Criminal 

0.829* (0.347) 

Record 

Workplace 

-0.825 (0.564) 

Spree 

Restaurant or 

Nightlife 

-0.779 (0.625) 

-0.447 (0.701) 

Another Location 

-0.889 (0.529) 

South 

Midwest 

West 

-0.141 (0.520) 

-1.150 (0.590) 

-0.585 (0.539) 

Handgun used 

0.057 (0.411) 

Number of Victims 

-0.120* (0.054) 

Killed 

 

1.643 

1.039 

2.290 

0.438 

0.459 

0.640 

0.411 

0.869 

0.317 

0.557 

1.059 

0.887 

 

0.838 (0.611) 

0.040 (0.468) 

0.779* (0.326) 

-0.877 *0.493) 

-1.038 (0.593) 

-0.791 (0.659) 

-1.092 (0.471) 

-0.327 (0.481) 

-1.381* (0.546) 

-0.898 (0.509) 

-0.111 (0.360) 

-0.157* (0.057) 

 

2.312 

1.041 

2.180 

0.416 

0.354 

0.453 

0.335 

0.721 

0.251 

0.407 

0.895 

0.854 

*= p<0.05.  The standard error is written in parentheses. 

 

 

 The revised dataset’s binary logistic regression, seen on Table 5, yielded two significant 

results. First, the binary logistic regression indicated that mass shooters with a prior criminal 

record were more likely to survive their attacks, rather than die. This result is pertinent to the 

third hypothesis of the present study. The statistic also displayed that for every additional victim 

a mass shooter kills, the more likely the mass shooter is to die than survive their mass shooting. 

This finding is relevant to the fourth hypothesis of the current study. 

 

The binary logistic regression that was based on the original dataset and presented in 

Table 5 found three significant results. These statistically significant results are similar to those 

of the revised dataset. Similar to the revised dataset’s binary logistic regression shown in Table 

5, this binary logistic regression, based on the original dataset, found that mass shooters with a 

prior criminal record are more likely to survive their attacks. It was also established that 

perpetrators of mass shootings in the Midwest are less likely to survive their attacks compared to 

those in the East.  The final significant statistic was that a mass shooter grows less likely to 

 

29

survive their attack as they kill more victims; another outcome that was also found for the 

revised dataset.  

Discussion 

 

As more mass shootings occur in America, the citizens look to the government for 

answers about how these atrocities can be stopped. However, how is the government expected to 

know how to combat these mass shootings with the limited research that is available about them? 

This is the problem with the previous research on mass shootings; there is just not enough of it. 

There is also no readily found research available about mass shooters committing suicide, which 

is part of the reason why it was essential for this study to be completed. By observing the 

differences between mass shooters who commit suicide and mass shooters who survive their 

mass shootings, this study highlights the characteristics that set these two types of mass shooters 

apart. Factors such as the region of the country, the location type, the marital status of the mass 

shooter, and more, displayed their roles in interacting with what happens to the mass shooter 

after the shooting. 

 

The inclusion of the data from Schildkraut (2014) with the usage of two definitions of a 

mass shooting—from Schildkraut and Elsass (2016) and the Stanford Geospatial Center 

(2016)—to comprise the ‘revised dataset’ was important because the combination of these two 

definitions helped form a new definition of a mass shooting that is comparable to a majority of 

the other common definitions of mass shooting found in the relevant literature (Bjelopera et al., 

2013; Krouse & Richardson, 2015). It was also important to simultaneously run Dr. Jaclyn 

Schildkraut’s dataset, only using the definition of a mass shooting from Schildkraut and Elsass 

(2016) to determine what changes happened to the data with the addition of more cases and of 

cases involving fewer wounded and killed victims.  

 

30

Major Findings 

 

As previously mentioned, the binary logistic regression yielded two significant results 

that are important for the study of mass shooters committing, or not committing, suicide after 

their attacks. The first major finding from the results of this study is that mass shooters with a 

prior criminal record were found to be more likely to survive their attacks. It was also revealed 

that for each additional victim a mass shooter kills, the less likely the mass shooter becomes to 

survive their attack.  

These significant results can be useful for a number of reasons. Particularly, researchers, 

law enforcement, and mental healthcare workers could use these findings to postulate what will 

happen to mass shooters after their attacks have been completed, based on their race, the location 

type, the prior criminal record, and the number of victims killed. While research seems to over 

exaggerate the number of people who will become mass murderers, thus discrediting the 

importance of prediction (Knoll, 2012), it is still wise for different places across the country to be 

prepared for what to do in case of a mass shooting. For example, after the mass shooting at 

Virginia Polytechnic Institute and State University in 2007, a great deal of colleges across the 

nation tried to implement strategies to help reduce the amount of harm a mass shooting could 

cause (Fox & Savage, 2009). While the statistics show that mass shootings are extremely rare 

events, the schools with plans in place for mass shootings can be basing their plans on the results 

of studies similar to this one.  

Hypotheses 

 

This research study had five hypotheses. The first hypothesis was that mass shooters who 

commit suicide following their attack are more likely to be older than mass shooters who survive 

their attacks. The results failed to reject this hypothesis and is shown in the binary logistic 

 

31

regression. The binary logistic regression showed that for every year a mass shooter ages, the 

less likely they are to survive, than commit suicide. This result, while not found to be statistically 

significant, is consistent with the research from Lankford (2015), which showed that mass 

murderers who committed suicide were an average of 36.5-years-old while mass murderers who 

survived their events were an average of 30.2-years-old. Therefore, the present study provides 

support for this result from Lankford (2015). 

The second hypothesis was that mass shooters who are divorced or separated will be 

more likely to commit suicide than survive their mass shooting event. This hypothesis was 

rejected by the binary logistic regression. Interestingly, the binary logistic regression showed that 

divorced or separated mass shooters were more likely to survive than commit suicide. Previous 

research by Kposowa (2000) showed that people who are divorced or separated are twice as 

likely to commit suicide than those who are married; a finding that was not supported by the 

binary logistic regression in this study. 

The third hypothesis pertained to criminal records, postulating that mass shooters who 

commit suicide after their attack are more likely to have a prior criminal record. This hypothesis 

was rejected as demonstrated with the binary logistic regression. The binary logistic regression 

provided the significant result that mass shooters with a prior criminal record were actually more 

likely to survive, rather than die, after their attacks. This finding was not consistent with the 

research from Weis et al., (2006) that showed that 196 out of 491 suicides in South Carolina, 

during a given time period, were committed by people with a prior criminal record.  

Fourth, mass shooters who murder more victims during their attack will be more likely to 

commit suicide. The results failed to reject this hypothesis by the binary logistic regression. The 

binary logistic regression showed that for every additional victim a mass shooter killed, the less 

 

32

likely the mass shooter was to survive than commit suicide. This is important because it supports 

the research by Lankford (2015) which showed that mass murderers who didn’t survive their 

attacks killed an average of 5.5 victims and mass murderers who did survive their attacks killed 

an average of 4.5 victims.  

Lastly, the fifth hypothesis purports that mass shooters who are white will be more likely 

to commit suicide than shooters of another race. This hypothesis was rejected by the binary 

logistic regression. The binary logistic regression showed that white mass shooters were more 

likely to survive their mass shooting attacks. This is important because a study completed by 

Lankford (2016) showed that white mass murderers were more likely to commit suicide after 

their murders, compared to mass murderers of other races. Thus, the present finding about white 

mass shooters does not support the aforementioned claim by Lankford (2016). 

Limitations 

 

With only two of the hypotheses being supported by the results of this study, this research 

did have its limitations. One of the primary limitations of this research was the lack of 

information available. To more thoroughly examine the action of mass shooters committing 

suicide, it would have been more helpful to have prior medical health diagnoses, possible suicide 

notes of the mass shooters, details about the crimes committed by those with a criminal history, 

and so forth. Having information that allowed insight into the mental health of these mass 

shooters would have allowed for more comparisons and conclusions to be made about mass 

shooters committing suicide, possibly even increasing the validity of the suicide by cop 

contention. 

 

The credibility of open sources news stories is a limitation of this study. Maier (2005) 

conducted a study of 4,800 news sources in fourteen different newspapers to look at the accuracy 

 

33

of reporting and found that 61% of the local news and feature stories contained errors. The 

finding that over half of the stories reported in the news were inaccurate could easily make one 

ponder the issue of truth in journalism. Therefore, the use of news stories in this study could be 

viewed as problematic. However, it was the best source available at the time for this study. 

 

Another important limitation of this study, related to the lack of available data, pertained 

to the variables of the prior criminal record and the marital status of the mass shooters. 

Specifically, if no information could be found on these variables in any open sources news 

article, senior researchers advised to mark a perpetrator as not having a prior criminal record 

and/or mark a perpetrator as being single. This instruction could have skewed the data and the 

results obtained by this study. However, if this was not done, then a great deal of data for these 

two variables would have been left out, which would also cause a problem when it came time to 

analyze the data. Therefore, more accurate results about marital status and prior criminal records 

could have been found had this information been available in open sources articles. 

 

The assumptions drawn from the results of this study may be considered more reliable if 

a larger sample size of mass shootings and mass shooters had been used. With only 185 mass 

shootings and 194 mass shooters, it is difficult to draw definitive conclusions from the results 

obtained. The statistical power of the results would have been enhanced had they been based on 

a larger sample size. This is not to say that the results from this study should simply be 

discredited, but rather observed as the starting point of the discussion into suicide of mass 

shooters.  

 

Lastly, the use of the definition of a mass shooting from Schildkraut and Elsass (2016), 

combined with the definition from the Stanford Geospatial Center (2016) could be viewed as a 

weakness of this study, as compared to using the definition of a mass shooting according to the 

 

34

Federal Bureau of Investigation (FBI). The definition of a mass shooting created by Schildkraut 

and Elsass (2016) was published two years ago and therefore, has not been tested over an 

extensive period of time, whereas the mass shooting definition from the FBI has existed 

relatively unchanged for nearly a decade. Another weakness is reflected in the statement about 

the location of the mass shooting. Simply stating that, “…location(s) are chosen either at random 

or for their symbolic value,” is a very broad way of describing the location choice. The definition 

also does not state the number of victims that are required to be killed in order for a multiple 

homicide event to be classified as a mass shooting. The authors argued that they did not include 

this in their definition because there are a lot of situational factors that can determine whether or 

not a person dies, such as if the shooting occurs far away from a hospital or if the shooter keeps a 

victim hostage after they are shot (Schildkraut & Elsass, 2016). This is problematic because 

without a victim count, shooting events in which two people were shot could be included as mass 

shootings, and could one easily argue that a shooting in which two people were shot is easily 

comparable to a shooting in which fifty people were shot? An additional limitation of this 

definition that could be made about using the definition from Schildkraut and Elsass (2016) is 

that the definition is too narrow and therefore, restrictive. Lastly, the definition excludes the 

category of mass shootings related to domestic violence, such as familicides. Although these 

events meet almost every criterion for a mass shooting, one could easily argue that they 

inherently different from mass shootings. However, by excluding mass shooting familicides, one 

who follows this definition would omit a great volume of data, which is unfortunate because 

having a larger quantity of data would make the conclusions and trends derived about mass 

shootings more valid. By not including mass shootings that involve domestic and/or family 

violence, not including a count of victims killed, and being vague about the location type, this 

 

35

definition could be limiting itself to a minute number of mass shooting cases available to 

research. 

Future Directions 

 

Future work should attempt to find more information about the mass shootings. Gather 

more information about the crimes committed—which could possibly be done by getting in 

contact with police departments or performing background checks—that would put certain 

perpetrators into the category of having a prior criminal record. Obtain copies of marriage 

licenses or divorce papers for all of the mass shooters to be certain about their marital statuses. 

Attain police reports of the actual mass shootings, rather than simply relying on news media to 

make sure that the facts are truthful. More information, gathered through more reputable 

channels, could make this study more reliable. It could also be more useful to look at a larger 

sample of mass shootings. 

 

Other researchers who choose to explore this topic should also delve deeper into the 

dependent variable category of the mass shooter being shot by another person. While none of the 

hypotheses included this category, results for being shot by another person were shown to be 

consistent and therefore, should be further researched. It may also be beneficial to break down 

this category even more. For this study, the category of being shot by another person consisted of 

people being shot by a police officer or a citizen. Many arguments could be made that those shot 

by police could be categorized as suicide by cop, however, there was not enough evidence 

present in the data found for this study to defend that argument. If more information was 

collected—such as suicide notes or medical files from a mental healthcare worker—then an 

argument for suicide by cop could be made here, therefore allowing some of those in the shot by 

another category to be combined with the self-inflicted suicide category.  

 

36

 

It would also be important to include a temporal issue into research on suicide and mass 

shootings. Specifically, mass shootings can happen so quickly that they can end before law 

enforcement is able to arrive on the scene. Therefore, it may be valuable to look at the rates of 

suicide by mass shooters before and after the police arrive. The present study was only able to 

report if a mass shooter survived, committed suicide, or was shot by another person during their 

mass shooting, but it would have been useful to have temporal information about when the 

suicides occurred. With this information, it is possible that many implications could be drawn 

from the research, such as highlighting the importance of quick response times by the police. 

 

The case could also be made to do more in-depth analyses of the mass shooters at specific 

location types. For example, there may be characteristics of school mass shooters that 

differentiate them from workplace mass shooters, rather than just their ages and how they 

typically die or live after mass shootings. A more thorough look the mass shooters in specific 

location types may potentially help law enforcement and mental healthcare workers create a 

typecast to help them try to identify mass shooters before they attack. Along with a typecast 

being useful to categorize specific traits and behaviors to different mass shooters in various 

location types, this could also be useful in hypothesizing if they would attempt to commit suicide 

after their mass shootings, which could be an important addition to the results drawn by the 

present study. 

 

 

 

 

 

 

37

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

BIBLIOGRAPHY 

38

BIBLIOGRAPHY 

 

 

Ahmedani, B. K., Simon, G. E., Stewart, C., Beck, A., Waitzfelder, B. E., Rossom, R., Lynch, F., 

Owen-Smith, A., Hunkeler, E. M., Whiteside, U., Operskalski, B. H., Coffey, M. J., & 
Solberg, L. I. (2014). Health care contacts in the year before suicide death. Journal of 
General Internal Medicine, 29(6), 870-877.  

 

Bjelopera, J. P., Bagalman, E., Caldwell, S. W., Finklea, K. M., & McCallion, G. (2013). Public 

mass shootings in the United States: Selected implications for federal public health and 
safety policy. Washington, D.C.: Congressional Research Service. 

 

Borges, G., Nock, M. K., Haro Abad, J. M., Hwang, I., Sampson, N. A., Alonso, J., Andrade, L. 

H., Angermeyer, M. C., Beautrais, A., Bromet, E., Bruffaerts, R., de Girolamo, G., 
Florescu, S., Gureje, O., Hu, C., Karam, E. G., Kovess-Masfety, V., Lee, S., Levinson, 
D., Medina-Mora, M. E., Ormel, J., Posada-Villa, J., Sagar, R., Tomov, T., Uda, H., 
Williams, D. R., & Kessler, R. C. (2010). Twelve-month prevalence of and risk factors of 
suicide attempts in the WHO world mental health surveys. Journal of Clinical 
Psychiatry, 71(12), 1-25.  

 

de Similien, R. H., & Okorafor, A. (2017). Suicide by cop: A psychiatric phenomenon. The 

American Journal of Psychiatry Resident’s Journal, 12(1), 20-22.  

 

Federal Bureau of Investigation (2008). Serial murder: Multi-disciplinary perspectives for 

investigators. Washington, D.C.: National Center for the Analysis of Violent Crime. 

 

Fox, J. A., & DeLateur, M. J. (2014). Mass shootings in America: Moving beyond Newtown. 

Homicide Studies, 18(1), 125-145.  

 

Fox, J.A., & Savage, J. (2009). Mass murder goes to college: An examination of changes on 

college campuses following Virginia Tech. American Behavioral Scientist, 52(10), 1465-
1485. 

 

Freilich, J. D., Parkin, W. S., Gruenewald, J., & Chermak, S. M. (2017). Comparing extremist 

perpetrators of suicide and non-suicide attacks in the United States. Terrorism and 
Political Violence, 1-23.  

 

Kalish, R., & Kimmel, M. (2010). Suicide by mass murder: Masculinity, aggrieved entitlement, 

and rampage school shootings. Health Sociology Review, 19(4), 451-464.  

 

 

39

Kessler, R. C., Berglund, P., Borges, G., Nock, M., & Wang, P. S. (2005). Trends in suicide 

ideation, plans, gestures, and attempts in the United States, 1990-1992 and 2001-2003. 
Journal of the American Medical Association, 293(20), 2487-2495. 

 

Knoll, J. L., IV. (2012). Mass murder: Causes, classification, and prevention. Psychiatric Clinics 

of North America, 35, 757-780.  

 

Kposowa, A. J. (2000). Marital status and suicide in the National Longitudinal Mortality Study. 

Journal of Epidemiology and Community Health, 54, 254-261.  

 

Krouse, W. J., & Richardson, D. J. (2015). Mass murder with firearms: Incidents and victims, 

1999-2013. Washington, D.C.: Congressional Research Service. 

 

Lankford, A. (2014). Précis of the myth of martyrdom: What really drives suicide bombers, 

rampage shooters, and other self-destructive killers. Behavioral and Brain Sciences, 37, 
p. 351-393. 

 

Lankford, A. (2015). Mass murderers in the United States: Predictors of offender deaths. The 

Journal of Forensic Psychiatry & Psychology, 26(5), 586-600.  

 

Lankford, A. (2016). Race and mass murder in the United States: A social and behavioral 

analysis. Current Sociology, 64(3), 470-490. 

 

Liston, B., Wan, W., Somashekhar, S., & Davis, A. C. (2017). Las Vegas gunman Stephen 

Paddock was a high-stakes gambler who ‘kept to himself’ before massacre. The 
Washington Post. Retrieved from https://www.washingtonpost.com/news/post-
nation/wp/2017/10/02/las-vegas-gunman-liked-to-gamble-listened-to-country-music-
lived-quiet-retired-life-before-massacre/?utm_term=.6a1342ba497d  

 

Maier, S. R. (2005). Accuracy matters: A cross-market assessment of newspaper error and 

credibility. Journalism and Mass Communication Quarterly, 82(3), 533-551. 

 

Meloy, J. R., Hempel, A. G., Gray, T., Mohandie, K., Shiva, A., & Richards, T. C. (2004). A 

comparative analysis of North American adolescent and adult mass murderers. 
Behavioral Sciences and the Law, 22, 291-309.  

 

Schildkraut, J. V. (2014). Mass murder and the mass media: An examination of the media 

discourse on U.S. rampage shootings, 2000-2012. (Unpublished doctoral dissertation). 
Texas State University, San Marcos, TX. 

 

 

40

Schildkraut, J., & Elsass, H. J. (2016). Mass shootings: Media, myths, and realities. Santa 

Barbara, CA: ABC-CLIO, LLC. 

 

Shah, A. (2012). Suicide rates: Age-associated trends and their correlates. Journal of Injury and 

Violence Research, 4(2), 79-86. 

 

Stack, S. (2000). Suicide: a 15-year review of the sociological literature part I: Cultural and 

economic factors. Suicide and Life, 30(2), 145-162. 

 

Stanford Geospatial Center (2016). Mass shootings in America. Retrieved from 

https://library.stanford.edu/projects/mass-shootings-america.  

 

United States Census Bureau (2013). Health status, health insurance, and medical services 

utilization: 2010. (United States Department of Commerce Publication No. P70-133RV). 
Washington, D.C. 

 

Weis, M. A., Bradberry, C., Carter, L. P., Ferguson, J., & Kozareva, D. (2006). An exploration 
of human services system contacts prior to suicide in South Carolina: An expansion of 
the South Carolina Violent Death Reporting System. Injury Prevention, 12, 17-21. 

 

41