Introduction

In 2023, there were over 600 shootings in the United States (US) where at least one person was injured or killed (Gun Violence Archive, 2024).Footnote 1 When focusing on just public mass shootings where at least three people died, there were twelve in 2023 (Follman et al., 2023).Footnote 2 While only a small percentage of the gun violence in US, public mass shootings cause great harm to the American psyche and social well-being (Lowe & Galea, 2017; Soni & Tekin, 2023). There has been a significant rise in mass shootings over the last decade, with the vast majority of the perpetrators being white and men (The Nelson A. Rockefeller Institute of Government, 2022).Footnote 3 With the increase in incidents over recent years, there has also come a lot of scrutiny towards law enforcement’s handling of such cases. One specific incident that brought significant public censure towards law enforcement was the handling of Dylann Roof. In 2015, Dylann Roof, a white man, killed nine Black people at a historically Black church in Charleston, South Carolina while shouting hate-filled speech. Despite this, Roof was peacefully detained and provided Burger King following his arrest (Elliot, 2020). Such deference was not granted to the numerous people of Color who have lost their lives at the hands of police in traffic stops or other minor infraction encounters with law enforcement in recent years (e.g., Michael Brown, George Floyd, Breonna Taylor, Atatiana Jefferson, Daunte Wright, and so many others (see Chughtai, 2020; The Washington Post, 2023)). While everyone who comes in contact with law enforcement should be treated with dignity and respect, the issue arises when race determines whether someone receives that respect, and ultimately the right to live.

According to Mapping Police Violence (2023), police have killed more Black people to date in late 2023 (at time of authorship) than by the same time in 2022. Furthermore, data collected over the last ten years find that Black people are nearly 3 times more likely to die at the hands of police than their white counterparts. When Black people commit crimes, sometimes minor crimes, they are often not treated with regard, but instead seen immediately as a threat, and many times met with lethal force. For example, George Floyd, a Black man, died in the custody of three police officers due to one of them kneeling on Floyd’s neck for over eight minutes. Law enforcement was dispatched originally because Floyd used a counterfeit $20US bill (Hill et al., 2022). Despite being accused of a lesser crime than Roof, Floyd was perceived as a greater threat and ultimately lost his life at the hands of law enforcement. Relatedly, recent research finds most mass shootings are associated with domestic violence, with many of those perpetrators being white men (Geller et al., 2021). Yet, despite this known tendency for violent and aggressive behaviors, many states do not block abusers from purchasing firearms (Zeoli et al., 2019) and most mass shootings are committed with legally acquired weapons (National Institute of Justice, 2022). Thus, too often, white men with guns are provided a presumption of good faith while Black men are assumed a danger just for being.

Furthermore, lethal force is also commonly used by law enforcement against Black people with mental health issues (Nelson, 2016). Of those Black men killed by police in 2015, around 25% of them were dealing will some underlying mental health problem (Hirschfield, 2016). Furthermore, the Treatment Advocacy Center reports individuals with untreated mental health concerns are 16 times more likely to die in an encounter with law enforcement (Fuller et al., 2015). Despite the rhetoric in the media, research finds very few perpetrators of mass violence have diagnosable psychiatric conditions that played a role in the attack (National Threat Assessment Center, 2023; Peterson et al., 2022), but those that do in the US are more likely to utilize semi- or fully-automatic firearms (Brucato et al., 2021). Thus, it is vital for scholars to continue exploring these disparities in mass violence incidents and police encounters to understand how to prevent needless death—at the hands of police and others—in the future.

To aid in such understanding, we explore the roles of race and mental health in the on-scene outcomes of mass public shootings since 1966. Specifically, we investigate if disparities exist in the likelihood of being apprehended, taking one’s own life, or being killed by police or others during a public mass shooting incident based on one’s race and known mental health status employing data from The Violence Project (Peterson & Densley, 2022). We begin, however, by providing an overview of the current mass shooting literature on race and mental health.

Mass Shootings

Within the mass shooting literature, different studies have different qualifications for a shooting to be included. For instance, the Gun Violence Archive (2023) defines mass shootings as an incident where four or more people are shot or killed, not including the shooter. Others take a more limited definition referring only to incidents in which four or more victims are killed with a firearm (Duwe, 2020); while still others require such violence and fatalities to occur in a public space (Follman et al., 2023). The average mass shooter is a white man around 30 years old (Silva, 2022). Scholars categorize mass shooting in three ways: public, family, and felony. Public mass shootings refer to an incident that occurred in public or populated locations. Family mass shootings refer to an incident that involves the perpetrator’s immediate or extended family. Felony mass shootings refer to an incident that involves underlying criminal activity (e.g., robbery/burglary or gang violence) (Silva, 2022).

When comparing the three types of mass shootings, there are some distinct factors that separate them. Public mass shootings are defined by incident characteristics, family by victim characteristics, and felony by offender characteristics (Fridel, 2021). Silva (2022) found that public mass shootings have higher rates of victim deaths and injuries, and almost exclusively involve firearms. Family mass shootings are known to target intimate partners or family members and occur in private homes. Felony mass shootings are more likely to be committed by Black people, have multiple perpetrators, and have victims who are acquaintances.

While it is important to understand mass shootings more broadly, the focus of this study is public mass shootings. In addition to incident characteristics, public mass shootings do have some prevalent characteristics that make them stand out among the other two (Fridel, 2021; Silva, 2022). Public mass shooters are more likely to have military experience and involve victims who were strangers. Public mass shooters are less likely to have a violent criminal history and involve the perpetrator surviving the shooting (Silva, 2022). Additionally, public mass shooters are more likely to have mental health issues, though it is not a significant difference (Fridel, 2021). And, research suggests public mass shooters are more likely to perceive being victimized, either directly, such as being bullied, disrespected, or harmed by others, or suffered more general perceived grievances toward a group they are affiliated with and perceive to be marginalized or oppressed (Lankford, 2013, 2018; Levin & Madfis, 2009; Newman & Fox, 2009; Vossekuil et al., 2002). Given these associations, it is unsurprising that most of the perpetrators of mass public shootings are white (Lankford, 2016). Thus, race is a vital factor to understanding variations in the factors associated with mass shootings, particularly across types. The current study seeks to address this by utilizing a database, from The Violence Project, that focuses exclusively on public mass shootings.

On-Scene Outcome

While research suggests public mass shooters are less likely to survive than other mass shooters (Silva, 2022), there are mixed results on how that death is brought about. For instance, the National Threat Assessment Center (2018) found that apprehension by law enforcement was the most common outcome for public mass shooters in 2017. On the other hand, Silva (2022) found that the most common outcome was suicide. Within the literature, multiple studies have discussed the increased suicidal tendencies of mass shooters (Duwe, 2004, 2020; Palermo, 1997; Silva, 2022). However, these studies often combine on-scene outcomes instead of looking at them separately. For example, Duwe (2020) found that 57% percent of shooters were suicidal because 42% committed suicide and 15% were killed by police or a citizen. Someone being killed by law enforcement after committing a crime should not be confused with having suicidal tendencies as that is a mere consequence of committing a crime. Additionally, without talking with the perpetrator, we cannot assume that they are suicidal simply because they “let” themselves be killed after committing a violent crime.

Other studies that have examined on-scene outcomes have not thoroughly accounted for the varied possibilities of how an incident may culminate. For example, some studies have utilized a binary variable and only examined whether the shooter survived the incident or not (Lankford, 2015, 2016), which does not account for how a perpetrator died. Capellan and Gomez (2018) examined on-scene outcomes using a categorical variable and found that around 44% of shooters were arrested, 14% were killed, and 41% committed suicide, but no statistical differences were found between the outcomes. We build upon previous literature by examining the variations in factors associated with three distinct public mass shooting outcomes—being apprehended, taking one’s own life, or being killed by someone else.

Although, the majority of public mass shooters are white, there are still people of Color who commit such actions. For instance, from 1966 to 2019, 52% of public mass shooters were white, 20.5% were Black, 8.2% were Latine,Footnote 4 and 5.8% were Asian (Peterson & Densley, 2019). In other words, only a slim majority of public mass shooters are white. It is important to analyze mass shootings from a racial perspective to understand if there are any racial differences in the characteristics and outcomes of mass shootings. To date, Lankford (2016) is the only known study to examine racial differences in on-scene outcomes among other characteristics of mass shootings, revealing 38% or more of white, Latine, and Asian offenders died, but only 15% of Black offenders died during the incident. Lankford (2016), however, assessed all types—public, family, and felony—of mass shooting, thus not allowing for understanding of the factors associated with outcomes between type. Such distinction is needed as it is known that varied types of mass shootings are commonly associated with different races (Fridel, 2021). Thus, as mentioned above, the three types of mass shootings are distinct and warrant individual analysis. The current study will address this by examining the relationship between race and on-scene outcomes for public mass shootings only.

Mental Health

After each mass shooting, mental health is discussed as an explanation for why someone would commit such a crime (Skeem & Mulvey, 2020), especially if the perpetrator is white. However, there is mixed research on the role that mental health plays in mass shootings. Studies examining the mental health of mass shooters have found that around 20–60% of all mass shooters have some indicator of a mental health issue (Addison, 2020; Capellan & Gomez, 2018; Duwe, 2020; Taylor, 2018). Of the three types of mass shootings, public mass shootings have the highest number of shooters dealing with mental health concerns (Fridel, 2021). Furthermore, Peterson et al. (2022) found mass shooters motivated by mental illness were similar to other mass shooters and often had the same risk factors for violence. In fact, they found serious mental health issues only contributed to about 31% of the mass shootings they analyzed. While it is unclear how many mass shooters experience mental health issues or the true link between such and their behaviors, there is enough evidence to warrant increased attention on mental health as a whole.

However, it is hard to get a full picture of mental health’s impact on mass shootings as mental health is not always operationalized consistently or comprehensively. For example, Fridel (2021) examined mental health by looking at whether the perpetrator was receiving mental health treatment prior to the shooting and whether they committed suicide or suicide-by-cop. Looking solely at treatment, however, is only a conservative indicator of mental health as it misses those who may struggle with mental health but lacked access to treatment. Moreover, looking at whether the offender committed suicide or suicide-by-cop is not indicative of mental health issues by itself. Mass shootings are violent crimes where a defendant can be given the death penalty if convicted. Therefore, committing suicide can be an indicator of the perpetrator wanting to have control over their last moments of life or being overcome with the reality of their decisions. Thus, suicide is not necessarily an indicator of mental health concerns.

Lankford and Cowan (2020) addressed the limitations mentioned above by examining mental health problems’ relationship with on-scene outcomes. They found that public mass shooters with a diagnosis or signs of mental health issues were not more likely to die because of the shooting compared to those without diagnoses or signs of mental health issues. However, they discuss that the database (The Violence Project database) they used had a significant number of missing values that could have influenced their results (Lankford & Cowan, 2020). The current study will address this limitation using an updated version of the same database that has more complete information about the mental health status of each shooter.

The Current Study

Building on previous research, the current study examines the associations between race and mental health with the on-scene outcomes of public mass shooters. Specifically, the current study examines how the perpetrator’s race and mental health status influence their likelihood of killing themselves, being killed by someone else, or apprehended, controlling for known factors associated with public mass shootings—victimization, criminal history, military experience, extremist affiliations, and age. With the increasing attention brought to law enforcements’ responses to crimes committed by Black people, it is important to continue to examine whether any racial disparities exist between on-scene outcomes, particularly for the most violent incidents. To explore these issues, we hypothesize the following:

Hypothesis 1

White people are at a greater risk of being apprehended than they are of being killed on-scene.

Hypothesis 2

White people are at a greater risk of killing themselves than they are of being killed on-scene.

Hypothesis 3

The more indicators of mental health problems a person has, the more they are at risk of being killed on-scene than being apprehended.

Hypothesis 4

The more indicators of mental health problems a person has, the more they are at risk of being killed on-scene than killing themselves.

Method

Data

Data for the current study came from The Violence Project Mass Shooter database which is a “comprehensive source of information on the psychosocial histories of public mass shooters” (Peterson & Densley, 2022). The researchers developed the database utilizing public records and open-source data for 190 public mass shootings from 1966 to 2022. For mass shootings to be included in the database, they had to follow a specific definition of mass shooting:

“…a multiple homicide incident in which four or more victims are murdered with firearms—not including the offender(s)—within one event, and at least some of the murders occurred in a public location or locations in close geographical proximity (e.g., a workplace, school, restaurant, or other public settings), and the murders are not attributable to any other underlying criminal activity or commonplace circumstance (armed robbery, criminal competition, insurance fraud, argument, or romantic triangle)” (Krouse & Richardson, 2015).

To be included in the analytic sample, the mass shooter had to be identified as a man or boy. All other gender identities were excluded due to only accounting for around three percent of the sample. Additionally, any mass shooting incident with missing values was excluded from the analytic sample. In the end, the analytic sample included 173 mass shootings from 1966 to 2022. Demographically, the sample consists of 173 mass shootings largely committed by white (56%) men and boys ranging from ages 11 to 70. On average, 60% of the mass shooters had a criminal record and 27% had at least some military experience. The majority also experienced at least some form of victimization in the past. Thirty-six percent of the perpetrators killed themselves, 21% were killed on-scene, and 43% were apprehended.

Measures

Dependent Variable

The dependent variable in the current study is on-scene outcome. On-scene outcome was measured by whether the mass shooter killed themselves, were killed on scene by someone else (police officer, school resource officer, or a civilian), or if they were apprehended (1 = killed self, 2 = killed on-scene, and 3 = apprehended). Two of the mass shooters committed suicide after being apprehended. Because the on-scene outcome was still apprehension with the mass shooter waiting until being in custody to commit suicide, these cases were coded as apprehended.

Independent Variables

The first independent variable of interest is race which was measured by whether the mass shooter was white or not (white = 1). Race was coded as a dummy variable due to the small size of the analytic sample. Fifty-six percent of the sample was white, 22% were Black, 9% were Latine, 6% were Asian, 4% Middle Eastern, 1% were Native American, and 1% identified as Moroccan or Bosnian.

The second independent variable of interest is history of mental health. This variable includes whether the mass shooter had been hospitalized for psychiatric reasons, received counseling, been prescribed psychiatric medication, been diagnosed with a disorder, and if they were receiving psychiatric treatment within the six months prior to the shooting. Values were added together for each mental health item and higher values indicate higher levels of mental health issues.

Control Variables

The current study controlled for age, criminal record (criminal record = 1), military experience (military experience = 1), extremist views, and history of victimization. For military experience, the mass shooters were coded as having military experience even if they only joined but did not make it through training. For extremist views, a variable was created that included whether the mass shooter was affiliated with a hate group and/or a terror group. Higher values indicate a higher level of extremist views. The history of victimization variable includes whether the mass shooter was bullied, physically abused, sexually abused, emotionally abused, had childhood trauma, and if their mother was abused. Higher values indicate higher levels of victimization.

Analytic Strategy

To assess the hypotheses, we first evaluated bivariate differences between on-scene outcomes and the independent variables in the study. To do so, one-way ANOVA tests were conducted. Then we employed multinominal logistic regressions to examine the associations between race and mental health with on-scene outcomes while controlling for age, extremist views, military experience, criminal record, and victimization. The multinominal logistic regressions were conducted in a multi-stage approach. First being apprehended served as the base outcome while accounting for the control variables. Then the regressions were repeated with killed self as the base outcome to determine the relationship between killed self and killed on-scene. Killed self was utilized as the base outcome because there were more observations for that variable than killed on-scene. A relative risk ratio (RRR) greater than one indicates that with an increase in the independent variable there is an increased risk of a case being in the comparison category and a decreased risk of a case falling into the baseline category. A RRR less than one indicates that with an increase in the independent variable there is a decreased risk of a case being in the comparison category and an increased risk of a case being in the baseline category. The following equation depicts how the estimations of the logistic regressions were conducted:

$${\text{Pr}}\left({y}_{i }=1|{x}_{i}\right)={\beta }_{0}+{\beta }_{1}{x}_{1}+{\beta }_{2}{x}_{2}\dots +{\beta }_{7}{x}_{7}$$

Preliminary assessments reveal the data do not suffer from multicollinearity. Analyses were completed in STATA 15.1.

Results

Table 1 provides descriptive statistics by on-scene outcomes. On average, white public mass shooters are more likely to kill themselves than be killed on-scene or apprehended. The most notable difference is that 39% of public mass shooters killed on-scene were white, meaning public mass shooters of Color account for the majority of perpetrators killed during an incident, despite not being the majority of the shooters. Table 1 further reveals that on average, for all outcomes, mass shooters are likely to have at least one indicator of a history of mental health concerns.

Table 1 Descriptive statistics by on-scene outcomes (N = 173)
Full size table

Table 2 provides the results of the one-way ANOVA tests that were conducted between the on-scene outcomes and the variables of interest. There are no significant differences between on-scene outcomes between races, mental health, criminal record, military experience, victimization, or age. There is, however, a significant difference between on-scene outcomes and extremist views (F = 0.95, P ≤ 0.05) such that those with higher levels of extremist views are more likely to be killed on-scene than kill themselves or be apprehended.

Table 2 One-way ANOVA tests
Full size table

Table 3 provides the results of the multinominal logistic regressions and the assessments of our hypotheses. Again, in terms of race, we hypothesize white men are more likely to be apprehended than killed on scene, and that white men are more likely to take their own life than be killed by law enforcement or others during an incident. Furthermore, in terms of mental health, we hypothesize those with a history of mental health problems are more likely to be killed than apprehended, and that they are more likely to be killed by others than themselves during a public mass shooting. The first model depicts the relationship between killed on-scene and killed self with killed self as the base outcome. There is a significant relationship between on-scene outcome and race (RRR = 0.33, P ≤ 0.01) such that white men are at a lower risk of being killed on scene and a higher risk of killing themselves at the mass shooting event compared to shooters of Color, supporting our race hypotheses. There was not a significant association between on-scene outcomes and mental health in the present models. Thus, we did not find support for our mental health predictions. Also, there was not a significant association between on-scene outcome and any of the control variables. The second model of Table 3 depicts the relationship between killed self and apprehended with apprehended as the base outcome. There are no significant relationships between on-scene outcome and any of the independent or control variables.

Table 3 Multinominal logistic regressions
Full size table

The third model of Table 3 depicts the relationship between killed on-scene and apprehended with apprehended as the base outcome. There is a marginally significant relationship between on-scene outcome and race (RRR = 0.49, P ≤ 0.10) such that white men are at a lower risk of being killed on-scene and a higher risk of being apprehended compared to men of Color, further lending support to our race hypotheses. There is also a marginally significant association between on-scene outcomes and military experience (RRR = 2.35, P ≤ 0.10) such that those with military experience are at a higher risk of being killed on scene and a lower risk of being apprehended compared to those without military experience. There was not a significant association between on-scene outcomes and mental health or the additional control variables.

Discussion

While mass shootings are rare events, it often seems that they are quite routine in the United States (Peterson & Densley, 2019). It does not help that they bring a lot of media attention and are constantly circulating on the news channels. Because of this, it is important that scholars continue to examine mass shootings to help us understand them and find ways to prevent them. The current study adds to the literature and our understanding of mass shootings by examining the roles that race and mental health play in the on-scene outcomes of public mass shootings.

Our results lend support to our race hypotheses. Specifically, the multinominal logistic regressions demonstrated that white men are at a lower risk of being killed on-scene and a higher risk of killing themselves at a public mass shooting event compared to men of Color. This finding confirms that there are racial differences in on-scene outcomes for public mass shooters. If white people are not at the same risk of being killed on-scene than people of Color, then they have an advantage when committing violent crimes. Thus, white men have to worry less about someone else ending their life despite ending the lives of other people.

Additionally, this finding raises questions about the treatment of people of Color by law enforcement in general. One explanation that is often touted for racial disparities in treatment by police officers is that people of Color are more likely to commit certain crimes. However, that is not the case for public mass shootings, as from 1966 to 2022, 56% of the shooters were white. It is possible that the implicit biases about people of Color held by law enforcement may influence their judgement and behaviors even with crimes that are mainly committed by whites. Thus, even though society has perpetrated the idea that only white men commit mass shooting, these violent crimes are still relatively rare. Therefore, until law enforcement deals with racialized implicit biases, specifically ideas of who is a threat, disparities in on-scene outcomes are likely to persist. However, explaining the true causes of such disparities is beyond the present scope and additional research is warranted on the causes and consequence of varied on-scene outcomes.

Somewhat surprisingly, we did not find any significant associations between mental health and on-scene outcomes. In particular, no significant associations were found between mental health and the shooter killing themselves, demonstrating that mental health and suicide after a mass shooting are separate concepts. Thus, while we did not find support for our mental health hypotheses, our findings still add to the literature. First, future research should refrain from utilizing suicide after a mass shooting as an indicator of a mental health problem. Suicide is a distinct action that warrants further investigation and should not be assumed an indicator of mental health. Furthermore, our lack of a direct association adds to the complex and mixed results between mental health and mass violence. On average, the public mass shooters in the current study showed at least one indicator of a mental health issue. It is still unclear what exact role mental health plays in mass shootings, but it seems some association is there, and we need to examine it further. It is important that society continues to talk about mental health to diminish the stigma surrounding it and increase access to mental healthcare.

Our findings also provide evidence that debunks the racialized moral panic around mass shootings and mental health touted by the media. White perpetrators of mass violence are often framed by the media as sympathetic characters that engaged in such violent actions due to their mental health problems, whereas Black perpetrators are often framed as a perpetual threat (Bridges et al., 2022; Duxbury et al., 2018). Such (supposed) associations and rhetoric are often peddled by (conservative) politicians in the wake of a mass shooting turning focus on the individual’s mental health over the national plague of gun violence (e.g., Mitchell, 2022; Vargas et al., 2020). Our findings lend to the complexity of the associations between mental health and mass shootings and further point out the racial disparities in how perpetrators are treated in the moment. Understanding these disparities adds further nuance to how they are later framed by the media and the links between these lines of research warrant investigation.

As with all research, there are some limitations in the current study. For starters, because mass shootings are rare, particularly when limited to events with multiple deaths and not just injuries, our sample size was small, thus limiting the statistical power of the analyses. Therefore, while statistical modeling allows the assessments of some associations, deeper investigations with qualitative methods would add greatly to the literature. Additionally, as noted above, the definition used for mass shootings is very narrow and only applies to public mass shootings. For that reason, the findings are not generalizable to other forms of mass shootings. Another limitation of the study is in the availability of mass shooting information. The Violence Project Mass Shooter database (Peterson & Densley, 2022) utilizes public records and open-source information which may not be complete or accurate for each perpetrator. The information gathered about each incident is secondhand and not gathered from the shooters or victims directly. Thus, data about broader mental health and contextual factors are missing. Despite these limitations, Peterson and Densley (2022) provide the most comprehensive data on public mass shootings to date. Future research is warranted to build upon their endeavors to further our understanding of mass violence more broadly.

Conclusion

It feels like a day does not go by in America without another report of mass violence, and this year—2023—that is the reality. We are met with calls for action (and often backlash)—to support mental healthcare, to restrict access to guns, to make schools safer, and ban assault rifles (e.g., Acevedo et al., 2022; Baker et al., 2023; Cathey, 2023). During the same time that we have experienced a rise in mass shootings, more people of Color have been killed by police (Mapping Police Violence, 2023). What is often lost in the discussion are the disparities in the outcomes of these events. While not a direct comparison, our results show there are racial disparities in on-scene outcomes of public mass shootings; and furthermore, mental health may not be as directly tied to such outcomes as often proposed. This study serves as a steppingstone to continue to unpack the causes and consequence of mass violence. Furthermore, we hope scholars will continue to look at the larger factors that link the rise in gun violence across America—racial divisions, biases in police culture, access to guns, the mental health crisis, and so forth. Such events do not occur in a vacuum. Understanding there are disparities is just the first step. We must find the causes if we are to prevent them. We hope this study aids understanding and inspires future investigations into the remaining gaps to inform policy and social change which will lead to a more just, safe, equitable, and less violent world, for all.