ÌÇÐÄvlog

Key Points

QuestionÌý What types of firearms are used in school shootings, and how are the firearms obtained?

FindingsÌý In this cross-sectional study of data from 262 adolescents who discharged firearms in 253 school shootings spanning 26 years from The American School Shooting Study, school shooting incidents in the US were typically executed using low- and moderate-powered firearms. These weapons were most frequently stolen from family members or relatives of the perpetrators.

MeaningÌý These findings inform the debate on how school shooters obtain their firearms and indicate that secure storage of firearms could benefit public health.

ImportanceÌý Within gun violence research, the types of firearms involved in US school shootings and the origins of these weapons have not been well explored.

ObjectiveÌý To examine the type, make, and power of firearms involved in school-related gun violence as well as the sources and methods through which adolescents obtained these weapons.

Design, Setting, and ParticipantsÌý This cross-sectional study used descriptive statistics and logistic regression to analyze 253 school shootings executed by 262 adolescents (aged ≤19 years) in the US from January 1, 1990, to December 31, 2016, and to calculate adjusted estimates for missing cases. The data were sourced from The American School Shooting Study (TASSS), which compiles information about these shooting events from open-source materials. Data analysis was performed from June 23, 2023, to July 20, 2023.

ExposuresÌý Firearm type, make, and power.

Main Outcomes and MeasuresÌý The firearm type refers to whether a weapon was a handgun, rifle, or shotgun. The firearm’s make differentiates semiautomatic or fully automatic guns from other action mechanisms. The firearm’s power refers to the caliber and velocity, distinguished as small, moderate, or higher power. The acquisition source captures from whom the firearm was received. The obtainment method refers to how the shooter obtained the firearm. Estimates were adjusted by total number of documents and TASSS reliability score as well as neighborhood demographic, socioeconomic, and land use variables measured at the census tract level, and were weighted to account for item nonresponse.

ResultsÌý A total of 262 adolescents (mean [SD] age, 16.2 [1.9] years; 256 [97.8%] male) were studied. In the adjusted analyses, handguns were the most used weapon in school shootings (85.5%; 95% CI, 80.6%-89.4%). Firearms were predominantly lower (37.0%; 95% CI, 29.9%-44.7%) or moderate (39.7%; 95% CI, 32.0%-47.8%) in power. Adolescents mainly obtained their guns from relatives (41.8%; 95% CI, 31.7%-52.6%), friends or acquaintances (22.0%; 95% CI, 13.2%-34.5%), the illegal market (29.6%; 95% CI, 19.3%-42.5%), strangers or persons who were shot (4.7%; 95% CI, 1.8%-11.6%), or licensed dealers (1.9%; 95% CI, 0.7%-5.2%). Most firearms were procured via theft from relatives (82.1%; 95% CI, 69.4%-90.3%).

Conclusions and RelevanceÌý Despite the more frequent use of lower to moderately powered firearms in school shootings, this study’s results indicated stability in the use of high-powered weapons in adolescent school shootings throughout the research timeframe. A majority of the firearms were procured from the perpetrator’s family or relatives or from friends or acquaintances. These findings may significantly influence discussions around gun control policy, particularly in advocating for secure firearm storage to reduce adolescents’ access to weapons.

Fatal and nonfatal gun violence at K-12 schools in the US is a significant social issue that has substantial policy and public health implications.^1-3 The US Department of Education’s Institute of Education Sciences has tracked school shootings since the turn of the millennium. Its latest findings, documenting 93 incidents between 2020 and 2021, are particularly alarming, as numbers are almost 4 times higher than the yearly averages reported in prior research.^4,5 One pivotal aspect of this mounting problem revolves around the age of the individuals responsible for these acts. Empirical research has consistently shown that more than 70% of school shooters in the US are adolescents aged 19 years or younger.⁵ In light of escalating gun violence among US youth and the daunting reality that gun-related injuries are now a leading cause of adolescent deaths, carefully examining the firearms deployed in these school shootings remains crucial for improving knowledge of this urgent public health concern.⁶

Contrary to the widespread perceptions fueled by media narratives, most adolescent school shootings do not fit the stereotype of mass casualty rampages involving assault-style weapons.⁷ Rather, a detailed analysis⁸ of school violence from 1990 to 2016 revealed that adolescents were responsible for only 7 mass casualty shootings, defined as causing 4 or more gunshot fatalities. Most incidents, approximately 98%, resulted in 1 or 2 fatalities. These shootings often involved handguns rather than assault rifles and were typically rooted in interpersonal disputes, largely reflecting broader patterns of gun violence within our society.^5,8,9

As with studies of gun violence in schools, the existing literature suggests that small- or medium-caliber semiautomatic handguns, with their concealability and simple design, feature prominently in firearm assaults and murders across US communities, where youth frequently acquire them illegally through trusted channels, such as family members, friends, or underground markets.^10-12 Unfortunately, despite growing research on the characteristics and patterns of school-related gun violence, it remains unclear whether these findings are specifically true for school shooters. Few empirical studies have systematically investigated the firearms used in adolescent school shootings, including their characteristics, acquisition sources, and obtainment methods, underscoring the critical need for further research to address this complex issue.

The primary goal of this study was to bridge this gap by leveraging innovative data from The American School Shooting Study (TASSS). In particular, we analyzed the type, make, and potency of firearms involved in school-related gun violence, as well as the sources (eg, family, friends, or illegal market) and methods (eg, theft and illegal purchase) through which adolescents obtained firearms. By studying these critical aspects, we provide essential insights that can inform evidence-based policies and preventive measures to reduce school shootings and safeguard the well-being of students and educators alike. At a time marked by increasing concerns over the safety of educational environments, our research seeks to advance our understanding and inform strategies to address this pressing issue comprehensively.

This cross-sectional study used data from the TASSS database. TASSS relies on public records (eTables 1 and 2 in Supplement 1) to document every known instance of firearm discharges on K-12 school property resulting in at least 1 gunshot injury or death in the US from January 1, 1990, to December 31, 2016, the most current period the data are available. Being one of the longest-running projects on US school shootings, TASSS has gathered data from various open sources, including news media (online and print), court records, police files, and academic reports (eFigure 2 in Supplement 1). With its broad scope, TASSS encompasses more than 30 000 documents and 90 000 pages of information chronicling incidents of gun violence at US schools. Researchers have used these resources to extract variables into a quantitative database.^5,13 The University of South Carolina, Michigan State University, John Jay College of Criminal Justice, and The University of Texas at Dallas institutional review boards approved this study as non–human participant research; thus, there was no need for informed consent from participants.¹⁴ This study followed the Strengthening the Reporting of Observational Studies in Epidemiology () reporting guideline.

Due to the public and media attention that school gun violence attracts, TASSS likely captures the known population of school shooting incidents and offenders.¹⁵ The database has information on 354 school shootings involving 373 publicly known individuals (adults and adolescents) who committed firearm-related assaults, as some incidents involved multiple shooters (eFigure 1 in Supplement 1). Given the importance of studying youth gun-carrying and acquisition behaviors, we extracted all 262 named adolescent perpetrators (aged ≤19 years) who discharged firearms in 253 school shootings for our analysis. We then examined all open-source documents housed in TASSS related to these school shootings to code variables about firearm attributes and adolescents’ methods to obtain them. We included item nonresponse and results from intercoder reliability using the Cohen κ coefficient for all variables analyzed. Missing data ranged from 4% to more than 50%. The Cohen κ results indicated substantial reliability for every variable, with scores between 0.789 and 0.988 (eTable 3 in Supplement 1).^16,17 All demographic information used in this study was reported in the TASSS database.

Three indicators captured the make and configuration of the guns discharged during school shootings. Firearm type included handguns, shotguns, and rifles. The firearm’s action mechanism coded the operational apparatus that loads, fires, and ejects ammunition as semiautomatic or fully automatic and otherwise (eg, revolvers, lever/bolt action, derringers, and pump action). The firearm’s power encompassed its caliber size and velocity as lower, moderate, and higher power. Lower power included the 0.22, 0.25, and 0.32 calibers and textual indicators from the open sources that a small-caliber gun was used. Moderate power included 0.38 caliber, 0.380 caliber, 9 mm, 0.410 gauge, and 20 gauge. Higher power included the 0.357 magnum, 10 mm, 0.40 caliber, 0.45 caliber, and 0.44 magnum or high-velocity rifles and shotguns, such as the 7.62 × 39-mm NATO, 5.56 × 45-mm NATO, 0.223 caliber, 0.270 caliber, 0.30 to 0.06 caliber, 0.30 to 30 caliber, 16 gauge, and 12 gauge (eMethods in Supplement 1).¹⁸

Two additional variables detailed the firearm’s obtainment sources and methods. Acquisition source captured the immediate individual (source) who supplied the firearm to the school shooter (knowingly or unknowingly), coded as a licensed dealer, family or relative (eg, parents or guardians, siblings, or grandparents), friends or acquaintances, street or illegal market, and strangers or from the person who was shot during the incident. The obtainment method examined how the school shooter procured the firearm from the acquisition source as a legal purchase, borrowed, taken or stolen, illegal purchase, or found randomly (eMethods in Supplement 1).^10,11 Importantly, because the obtainment method captures how the gun was procured from the acquisition source, there necessarily is an overlap between these 2 variables, limiting potential combinations. For example, the variable licensed dealer indicates that the firearm was procured legally or taken or stolen, whereas other gun sources offer diverse obtainment methods. The findings should thus be interpreted with this limitation in mind.

Our analysis uses descriptive statistics to report the variable’s distributions within the school shooting population and across time. We present unadjusted and adjusted estimates to address item nonresponse (ie, missingness). To adjust the data, we fit a logistic regression model, linking indicators of open-source quantity (eg, total number of documents) and reliability (TASSS reliability score) and neighborhood factors with the likelihood of missing values on each of the gun variables (with 1 indicating missing and 0 not missing). Using data from the US Census Bureau’s decennial surveys (1990, 2000, and 2010) that correlate with violence and neighborhood characteristics, we incorporated several contextual indicators associated with the school at the census tract level.^19,20 These represented 3 general categories.

First, variables capturing sociodemographic characteristics included the percentages of residents who were Black, Latino/a/x, White, and of other racial or ethnic groups (including Asian or Pacific Islander and Indigenous peoples of the Americas); non–US-born residents; and residents younger than 18 years. Second, economic indicators included the percentage of residents living below the poverty line, unemployment, public assistance dependency, households headed by women, and residents without a high school diploma. Third, we included variables that display land use characteristics of neighborhoods as the proportion of vacant total land, percentage of owner-occupied homes, residents who lived in the same house 5 years earlier, and urbanicity. We selected these sociodemographic and economic variables because they captured the most comprehensive set of measures available across all 3 US Census surveys and provided a picture of the groups of people living in the neighborhoods where school shootings occurred. These characteristics are inextricably connected to the underlying population of those who attend certain schools because most locations are composed of students who live nearby.

As noted, we assessed the probability of missingness for each variable to weigh the cases and compute the proportions shown. Hence, the adjusted data incorporated sampling weights (ie, the reciprocal of the adjusted probability). Because missing values were unavoidable yet vital to consider, this statistical adjustment is a significant contribution, allowing the data to accurately represent the recognized US school shooting population. Data analysis was performed from June 23, 2023, to July 20, 2023, with Stata BE, version 17.0 (StataCorp LLC).

A total of 119 of 253 adolescent school shooting incidents (47.0%) involved at least 1 fatality, and 7 (2.8%) were mass killings that caused 4 or more deaths. Among the 262 adolescent shooters, the overwhelming majority (256 [97.8%]) were male and 6 (2.3%) were female, with a mean (SD) age of 16.2 (1.9) years. As reported in TASSS, 141 (57.8%) of these individuals were Black, 21 (8.6%) were of Latino/a/x ethnicity, 68 (27.9%) were White, and 14 (5.7%) represented other racial or ethnic groups, including Asian or Pacific Islander and Indigenous peoples of the Americas. At the time of the incident, 144 (55.0%) of these adolescent shooters were enrolled as students at the school they targeted (eTable 4 in Supplement 1).

Regarding the census tracts associated with the school, a mean (SD) of 36.9% (36.8%) of residents were Black, 10.4% (15.7%) were Latino/a/x, and 46.3% (35.2%) were of races other than White. In these census tracts, a mean (SD) of 7.9% (10.7%) of residents were born outside the US, and 27.3% (6.2%) were younger than 18 years. Regarding economic characteristics, 12.3% (7.6%) of households were characterized as headed by women. Moreover, 26.1% (14.1%) of residents did not have a high school diploma, and 10.1% (6.5%) were unemployed. Additionally, 8.1% (7.5%) of the population depended on public assistance, whereas 19.7% (13.6%) lived below the poverty line. A mean (SD) of 58.9% (20.8%) of the residents were homeowners, and 54.7% (13.6%) had resided in the same house for 5 years or more. The census tracts were predominantly urban (81.6% [37.4%]), and 10.5% (8.3%) of the total land in these areas was vacant (eTable 5 in Supplement 1).

Table 1 reports the unadjusted and adjusted proportions describing the characteristics of the firearms used in US school shootings involving adolescent perpetrators. Focusing on the adjusted proportions, the firearms discharged were primarily handguns (85.5%; 95% CI, 80.6%-89.4%), followed by rifles (8.6%; 95% CI, 5.7%-12.7%) and shotguns (5.9%; 95% CI, 3.6%-9.6%). Most of these guns were semiautomatic or fully automatic (58.5%; 95% CI, 49.8%-66.8%). Regarding the firearm’s inherent size and power, 37.0% (95% CI, 29.9%-44.7%) were lower powered, and 39.7% (95% CI, 32.0%-47.8%) were moderately powered. The 0.22-caliber (50 of 186 [26.9%]) and 9-mm (29 of 186 [5.6%]) were the most common lower to moderately powered weapons. It was rare for adolescents to deploy higher-powered firearms in school shootings between 1990 and 2016 (23.3%; 95% CI, 16.6%-31.7%). When they did, most used the 12-gauge shotgun (9 of 186 [4.8%]), or a 0.40-caliber (8 of 186 [4.3%]), 0.357-caliber (6 of 183 [3.2%]), or 0.45-caliber (5 of 186 [2.7%]) handgun. Among rifles, typical recreational firearm calibers, such as the 0.223, 0.30 to 0.30, 0.30 to 0.06, 0.30, and 0.270, were used most often; 4 of the 186 shootings (2.1%) involved 7.62 × 39-mm NATO rifle rounds.

Extending this discussion, the Figure reports changes in firearm size and power across time. The use of lower to moderately powered firearms in school shootings involving adolescents in the US has decreased. The early 1990s marked the height of such firearm use, with the lowest point in the 2010s. The frequency of adolescents using higher-powered firearms has been stable over time, with a modest increase since the mid-1990s. Furthermore, the number of incidents in which the firearm’s power was undisclosed in open sources has notably increased between 1990 and 2016.

Table 2 reports the sources of firearm acquisition and methods of obtaining guns for both the unadjusted and adjusted data. Considering the adjusted data, 41.8% (95% CI, 31.7%-52.6%) of the adolescents involved in US school shootings procured firearms from family members or relatives, primarily parents. A total of 22.0% (95% CI, 13.2%-34.5%) obtained weapons from friends or acquaintances, whereas 29.6% (95% CI, 19.3% 42.5%) secured them from the illegal market. Fewer adolescents sourced guns directly from strangers or the persons who were shot (4.7%; 95% CI, 1.8%-11.6%) or used personally owned weapons bought legally (1.9%; 95% CI, 0.7%-5.2%).

Most firearms were procured via theft (78 of 128), predominantly from family or relatives (82.1%; 95% CI, 69.4%-90.3%), followed by strangers or the persons who were shot (9.5%; 95% CI, 3.7%-22.0%) and friends or acquaintances (8.4%; 95% CI, 3.5%-18.7%). When firearms were loaned or gifted to the shooter, the majority came from friends or acquaintances (93.5%; 95% CI, 79.2%-98.2%). All illegally purchased or randomly discovered firearms were traced to the street or illegal market.

This study addresses existing gaps in knowledge by presenting key trends relating to the firearms used in US school shootings perpetrated by adolescents from 1990 to 2016 and the methods of obtaining these weapons. Although today’s media coverage primarily focuses on high-profile school shootings in which assault-style rifles are used, our findings suggest these cases do not historically represent the broader issue, despite the emergence of this new trend in recent years, which falls beyond the scope of TASSS (ie, 2017 to 2023). We found that handguns, likely due to their higher accessibility and concealability, accounted for most of the firearms used, and many of these weapons were semiautomatic or fully automatic and lower to moderately powered. Our findings thus align with prior research on adolescent gun-carrying behaviors, showing that small-caliber (eg, 0.22, 0.25, and 0.32) and medium-caliber (eg, 0.38 and 0.380) handguns constitute most firearms acquired by criminally involved youths in US communities.¹⁰ Overall, these results indicate that the firearms involved in community violence are similarly deployed in school shootings, suggesting a potential overlap in the factors driving these violent incidents collectively.

Throughout our study period, we also observed a notable decrease in the use of lower- to moderate-power firearms, such as 0.22-caliber and 9-mm firearms, whereas the prevalence of higher-powered weapons remained consistent. This trend mirrors earlier research detailing the evolution of handgun designs, especially the change from revolvers to semiautomatics and the increased prominence of calibers like the 9 mm.²¹ Several factors could account for these findings: alterations in firearm manufacturing and availability, increasing adolescent interest in and familiarity with more potent weapons, or changes in documentation methods. Moreover, we discovered a sharp increase in unreported firearm size and calibers in the open sources in recent years. Although school shootings typically receive extensive media coverage,²² this finding indicates that the content and nature of reporting practices may have changed over time.

Therefore, despite more extensive media coverage, the increase of online publications and social media in the 21st century has not necessarily deepened our understanding of firearms used in shootings. This situation underscores the need for a standardized national reporting system for these incidents instead of solely depending on media sources. The 2014 death of Michael Brown in Ferguson, Missouri, for instance, highlighted this gap: without a national data registry at the time, it was challenging to discern nationwide patterns in police use of deadly force. Such barriers can impede our ability to study vital societal issues. Consequently, future efforts must prioritize consistent reporting procedures, especially for infrequent but significant events, such as school shootings and other notable violent acts.

Another crucial discovery from this study is the predominant sources of firearms deployed in adolescent school shootings: many were procured from family members (primarily parents), friends, or street markets and usually through illegal means, such as theft. Previous studies corroborate that criminally engaged youths younger than 18 years typically acquire guns from these same sources.¹¹ This finding again suggests a parallel between school gun violence and broader community violence. School shootings may not be isolated occurrences but could share underlying factors with off-campus violence. Hence, policy solutions should be comprehensive, intertwining with strategies to address gun violence more broadly, especially those targeting weapon accessibility.

Overall, these findings stress the critical public health message concerning the secure storage of firearms, especially in households with adolescents. Our study suggests that initiatives limiting adolescent access to firearms, such as child access prevention laws or efforts to decrease illegal gun trafficking, might effectively prevent school shooting incidents.^23,24 Furthermore, hospital-based initiatives centered on screening for firearm accessibility and exposure for inpatients could be fruitful in preventing gun violence, both inside and outside schools.²⁵

This study has some limitations. We used publicly sourced data, presenting benefits and challenges when evaluating firearm characteristics and acquisition methods. Despite TASSS’s rigor and this study’s strong intercoder reliability, open sources can include vague, imprecise, and conflicting information, leading to human and other measurement errors. Therefore, we advise readers to approach the results with caution. Moreover, although we adjusted for missing values, which had minimal impact on the estimates, significant data were missing for some variables, which may distort this study’s findings if these missing values could ever be captured. In addition, TASSS data are current only up to 2016. Analyzing more recent data in future research could reveal new and evolving trends.

The results of this study provide valuable insights that can inform both public health interventions and policy measures. The study highlights the need for comprehensive strategies, including better firearm safety practices at home, legal measures, and community education, to address the multifaceted issue of school shootings. Ultimately, this study supports policies encouraging secure firearm storage in households and limiting adolescents’ access to firearms through legislative measures, educational campaigns, or public health initiatives. The study further underlines the role played by the illegal market and informal networks (such as friends and acquaintances) in firearm acquisition, implying the need to bolster efforts to curtail illegal firearm trafficking.

Accepted for Publication: September 20, 2023.

Published Online: November 27, 2023. doi:10.1001/jamapediatrics.2023.5093

Correction: This article was corrected on January 8, 2024, to fix 2 percentages that were reported erroneously in the Results section: In the third paragraph of the Results, the adjusted proportion of rifles should be 8.6%, not 9.6%; and in the fifth paragraph, the adjusted proportion of shooters who procured firearms from family members should be 41.8%, not 51.8%.

Corresponding Author: Brent R. Klein, PhD, Department of Criminology and Criminal Justice, University of South Carolina, 1305 Greene St, Columbia, SC 29208 (bk17@mailbox.sc.edu).

Author Contributions: Dr Klein had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Klein, Schnell.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Klein, Schnell, Lewis.

Critical review of the manuscript for important intellectual content: Klein, Trowbridge, Schnell.

Statistical analysis: Klein, Trowbridge, Schnell.

Obtained funding: Klein.

Administrative, technical, or material support: Klein.

Supervision: Klein, Schnell.

Conflict of Interest Disclosures: Dr Klein reported grants receiving from the National Institute of Justice during the conduct of the study. No other disclosures were reported.

Funding/Support: The data used for this manuscript were collected as part of 3 National Institute of Justice awards: 2016-CK-BX-0013; 2018-R2-CX-0002; 2020-CK-BX-0003.

Role of the Funder/Sponsor: The National Institute of Justice had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The views and statements contained herein are those of the authors and do not necessarily reflect the National Institute of Justice.

Data Sharing Statement: See Supplement 2.