Firearm availability and homicide: A review of the literature

https://doi.org/10.1016/S1359-1789(03)00044-2Get rights and content

Abstract

This article reviews the most commonly cited, representative, empirical studies in the peer-reviewed literature that directly investigate the association of gun availability and homicide victimization. Individual-level studies (n=4) are reviewed that investigate the risks and benefits of owning a personal or household firearm. The research suggests that households with firearms are at higher risk for homicide, and there is no net beneficial effect of firearm ownership. No longitudinal cohort study seems to have investigated the association between a gun in the home and homicide. Two groups of ecological studies are reviewed, those comparing multiple countries and those focused solely on the United States. Results from the cross-sectional international studies (n=7) typically show that in high-income countries with more firearms, both men and women are at higher risk for homicide, particularly firearm homicide. Time series (n=10) and cross-sectional studies (n=9) of U.S. cities, states, and regions and for the United States as a whole, generally find a statistically significant gun prevalence–homicide association. None of the studies prove causation, but the available evidence is consistent with the hypothesis that increased gun prevalence increases the homicide rate.

Introduction

In the past 35 years, homicide rates in the United States have moved in cycles. The rate of homicide victimization doubled from the mid-1960s to the late 1970s. After peaking in 1980 at a rate of 10.2/100,000, it fell to 7.9 in 1985. The rate rose again and peaked at 9.8 in 1991 and then declined to a low of 5.7 in 1999 (Federal Bureau of Investigation, 2002). Between 1985 and 1991, the nonfirearm homicide rate remained relatively stable, while the firearm homicide rate increased significantly, thus accounting for the changes in the overall homicide rate. The recent reductions in the overall homicide rate have been attributed to reductions in firearm homicides in large cities—the same areas responsible for a large proportion of the increase in the overall firearm homicide rate during the preceding rise. Stratified analysis of homicide victimization rates for youth and young adults reveals that these groups were primarily responsible for the dramatic increase in firearm homicide rates during the mid-1980s to the early 1990s. Researchers attribute the recent decline in homicide to different causes, including reduced unemployment, increased policing, and the decline and stabilization of the illegal drug markets (Wintemute, 2000).

The gender and race of victims and offenders have not changed significantly over time. Males commit approximately 90% of all homicides and represent 75% of the victims. African Americans are disproportionately represented among homicide victims and offenders. Prior to 1985, 25- to 34-year olds experienced the highest homicide victimization rates; however, after 1985, 18- to 24-year olds have had the highest rates of homicide victimization (Federal Bureau of Investigation, 2002).

The homicide rate is higher in urban than in rural areas (Federal Bureau of Investigation, 2002). Homicide rates have been consistently higher in the Southern and Western regions of the United States. This is especially true for firearm homicides (Fox & Zawitz, 2000). Most researchers believe these regional differences are a result of cultural differences and higher gun ownership levels (Messner & Rosenfeld, 1999).

Over 60% of all homicides in the United States in 1999 involved a firearm (Federal Bureau of Investigation, 2002). Differences in the lethality of the weapons used in criminal acts have been described as the “weapon instrumentality effects” Cook, 1991, Zimring, 1968. The theory is that, for a given level of intent, firearms are more lethal when used during assaults and robberies compared to other weapons such as knives. Zimring (1991) suggests that firearms allow the assailant to kill without sustained effort and allow a weaker person to attack a stronger victim in a way that may not be physically possible with another weapon.

Firearm ownership in the United States, and particularly handgun ownership, is much more common than in other developed nations. From their national survey, Cook and Ludwig (1997) estimated that, in 1996, 44 million Americans owned 192 million firearms. Of these firearms, 65 million were handguns. Approximately 75% of gun owners reported owning more than one gun. Many more men than women owned guns, and ownership was highest in rural areas.

In 2000, it was estimated that 32% of American households owned at least one firearm. This number has declined from 51% in 1977. Although the number of firearms owned by individuals increased during this same time period and the percentage of individuals owning firearms remained relatively constant, changing household demographics has led to the decline in the household ownership figure. In the past 30 years, the number of households with an adult male has fallen significantly (Smith, 2000).

Neither most states nor the federal government has reporting requirements for firearm sales or ownership. Gun ownership rates have generally been estimated through surveys, such as the General Social Survey (GSS) or some state-level Behavioral Risk Factor Surveys (BRFSS). The results of the GSS are representative of the nation and the nine census regions but not of individual states (Davis & Smith, 1998).

Researchers, therefore, have generally been forced to develop proxies to measure ownership levels at the state or city levels. Usually, these proxies are validated by comparing them to a survey data.

Cook (1979) developed a well-known proxy for city gun prevalence. To investigate the relationship between gun prevalence and robbery rates in cities, he used the average of the percent of homicides involving a gun and the percent of suicides involving a gun to estimate gun ownership. Often referred to as “Cook's index,” this proxy has been used in several other studies Hemenway & Miller, 2000, Miller et al., 2001, Sloan et al., 1988. Other proxies for gun ownership levels have included the number of hunting licenses per capita (Krug, 1967), subscriptions to Gun and Ammo magazine Duggan, 2001, Lester, 1988, the percent of robberies and other crimes in which guns are used Kleck, 1984, McDowall, 1991, and the number of handgun purchase licenses Bordura, 1986, Fisher, 1976, Newton & Zimring, 1969. Some researchers have attempted to create factor scores of multiple measures of prevalence Kleck & Patterson, 1993, Seitz, 1972. Additional measures have included the percent of suicides by firearm Azrael et al., 2001, Hemenway & Miller, 2000, Hemenway et al., 2002, Miller et al., 2001, Miller et al., 2002a, Miller et al., 2002b, Miller et al., 2002c, the accumulated number of new firearms manufactured in the United StatesKleck, 1979, Kleck, 1984, and the rate of firearm accidents (Lester, 1988).

Recent research has attempted to determine which proxy measures most accurately measure gun ownership. Azrael et al. (2001) evaluated several proxies by comparing them to survey data. The proxies included the percent of suicides committed by firearm (FS/S), the percent of homicides committed by firearm (FH/H), Cook's index, the unintentional firearm death rate, subscription rates to Guns and Ammo magazine, and the number of National Rifle Association members per capita. They found that FS/S produced the most consistent results when compared with multiple survey-based estimates of gun ownership at the county, state, and regional levels.

Even when measured accurately, household gun ownership levels are not an ideal measure of firearm availability. The percentage of households with a firearm tells us nothing about the average number of firearms in each of those households or how lethal and concealable the guns are. Nor does it tell anything directly about how easy it is for high-risk users, such as teenagers or criminals, to obtain guns. For example, in the United States, guns move easily across state lines. Teens and criminals in states with few guns and strict gun laws can readily obtain firearms that were originally purchased in states with many firearms and lax gun regulations (Bureau of Alcohol Tobacco and Firearms, 1999). Nonetheless, most research uses household gun or handgun ownership levels as the “gold standard” for estimates of gun availability.

There are various theories that predict a negative relationship between household firearm ownership and homicide. A prime reason for gun ownership is protection. If a gun in the home often deters or thwarts assaults and violent robberies, we might expect to find that households with firearms had lower homicide victimization rates.

Greater gun availability in the community could also reduce homicide. Citizens could use their guns both to protect themselves and to protect their neighbors against the criminal element. We might thus expect to find lower rates of homicide in communities where guns were more available.

Areas with high levels of crime and homicide, particularly gun homicide, might create a demand for public policies that try to reduce the number of firearms available to criminals and others. Strict gun control laws might reduce the levels of gun ownership and homicides could either fall somewhat (or stay unchanged or rise), but homicide rates might still be high relative to low crime or rural areas. Comparisons across geographic areas might find that areas with high homicide rates had lower levels of gun ownership.

On the other hand, there are reasons to expect a positive relationship between firearm ownership levels and homicide rates. Higher availability of firearms could lead to higher homicide rates if there is a greater use of guns in robberies and criminal assaults, making these encounters more lethal. The ready availability of firearms could also contribute to the escalation of the level of violence in arguments or disputes. Having a gun could embolden individuals to take risks, to engage in activities, or go places that they might otherwise avoid because they were too dangerous. We might also expect a positive relationship between firearm ownership levels and homicide rates if high homicide rates in an area lead households to arm themselves for protection.

The relationship between firearm availability and homicide can clearly be bidirectional. For example, high levels of firearm availability might lead to higher homicide rates, and higher homicide rates may lead to more people acquiring firearms. Some of the studies reviewed in this article directly address this issue Duggan, 2001, Kleck & Patterson, 1993, Magaddino & Medoff, 1984. Also, firearm regulations can directly affect both gun availability and homicide rates, and both firearm ownership levels and homicide rates may influence whether or not firearm regulations are enacted. For example, firearm regulations may cause fewer individuals to acquire firearms, and as a result of low levels of gun ownership, passing further restrictive laws may meet less political resistance.

This article does not address research that has investigated the relationship between firearm regulations and homicide. Nor do we claim to review every study examining the association of firearm availability and homicide. In addition, we do not discuss related studies that provide indirect evidence on the relationship between firearm availability and homicide (e.g., Do criminals who assault with knives have different levels of lethal intent than those who use guns). Rather, this article provides a review of the most commonly cited, representative, and empirical studies in the peer-reviewed literature that directly investigate the association between gun availability and homicide.

The first section describes individual case control and cohort studies. The second section describes international ecological studies that have compared the United States to other countries. The third section describes ecological studies of the United States that have contrasted the levels of gun availability and homicide across regions, states, and rural and urban areas.

None of the studies can prove causation. They merely examine the statistical association between gun availability and homicide. They provide evidence on whether or not the real world is consistent with various theories. The actual hypothesis that is usually tested is that there is no statistical association between gun availability and homicide (the “null” hypotheses). The issue is whether this hypothesis can be rejected with a high degree of confidence, and if so whether the direction of the association is positive or negative.

Section snippets

Individual-level studies

The case control and cohort studies discussed below provide information about the benefits and risks associated with owning a personal or household gun. They do not provide information about whether community levels of gun ownership affect the household risk of homicide.

Ecological studies

A key element in case control, cohort, and ecological studies is to account for possible important confounders that affect the relationship between the variable of interest and the outcome; in this case, firearm availability and homicide. Generally, an important first step is to compare likes to likes. For case control studies, that means matching on, at least, age, gender, and neighborhood. For international studies, that means comparing high-income countries to high-income countries, thus

Conclusion

The available evidence is quite consistent. The few case control studies suggest that households with firearms are at higher risk for homicide, particularly firearm homicide. International cross-sectional studies of high-income countries find that in countries with more firearms, both men and women are at higher risk for homicide, particularly firearm homicide. This result is primarily due to the United States, which has the highest levels of household ownership of private firearms, the weakest

Acknowledgements

This research was supported in part by grants from the Joyce Foundation, the Robert Wood Johnson Foundation, and the Open Society Institute.

References (55)

  • B.S Centerwall

    Homicide and the prevalence of handguns: Canada and the United States, 1976–1980

    American Journal of Epidemiology

    (1991)
  • P.J Cook

    The effect of gun availability on robbery and robbery murder

  • P.J Cook

    The technology of personal violence

  • P.J Cook et al.

    Guns in America: National survey on private ownership and use of firearms (NCJ 1654476)

    (1997)
  • J Corzine et al.

    Cultural and subcultural theories of homicide

  • P Cummings et al.

    The association between the purchase of a handgun and homicide or suicide

    American Journal of Public Health

    (1997)
  • P Cunningham et al.

    Patterns and correlates of gun ownership among nonmetropolitan and rural middle school students

    Journal of Clinical Child Psychology

    (2000)
  • J Davis et al.

    General social surveys, 1972–1998

    (1998)
  • M Duggan

    More guns, more crime

    Journal of Political Economy

    (2001)
  • R DuRant et al.

    The association of weapon carrying and fighting on school property and other health risk and problem behaviors among high school students

    Archives of Pediatrics & Adolescent Medicine

    (1997)
  • Federal Bureau of Investigation

    Uniformed crime reports

    (2002)
  • J.C Fisher

    Homicide in Detroit

    Criminology

    (1976)
  • J.A Fox et al.

    Homicide trends in the United States: 1998 update (179767)

    (2000)
  • D Hemenway et al.

    Firearm availability and homicide rates across 26 high-income countries

    Journal of Trauma, Injury, Infection, and Critical Care

    (2000)
  • D Hemenway et al.

    Firearm availability and female homicide victimization rates among 25 populous high-income countries

    JAMWA: Journal Of the American Medical Women's Association

    (2002)
  • A.L Kellerman et al.

    Gun ownership as a risk factor for homicide in the home

    New England Journal of Medicine

    (1993)
  • M Killias

    International correlations between gun ownership and rates of homicide and suicide

    Canadian Medical Association Journal

    (1993)
  • Cited by (146)

    • Extending the Firearm Suicide Proxy for Household Gun Ownership

      2023, American Journal of Preventive Medicine
    • Are gun ownership rates and regulations associated with firearm incidents in American schools? A forty-year analysis (1980–2019)

      2021, Journal of Criminal Justice
      Citation Excerpt :

      This measure offers a robust estimate of gun ownership rates across states that improves on prior analyses. Drawing on the State Firearm Law database, dummy variables were created indicating whether the state required safety training to own a handgun, had a minimum age requirement of 21 years of age to purchase a gun, and had a child access prevention law holding firearm owners criminally responsible in instances of negligence that allow minors to gain access to their guns (e.g. failure to store firearm in securely locked container) (Fla. Stat. Ann. § 790.174, 2011; Hawaii Rev. Stat. § 134-10.5, 2011). Between 1980 and 2019, twenty-nine states implemented a child access prevention law holding firearm owners responsible for allowing minors to gain access to their guns, twenty-one states enacted a minimum age law of 21 years of age for handgun purchases, and six states created laws mandating safety training for all handgun purchases.

    View all citing articles on Scopus
    View full text